BUSINESS RESEARCH METHODS IV - SEMESTER BBA Bangalore University
Prepare !y" Dr#S#HARIHARAPUTHRIAN Pro$#Manage%ent" Ne& Hori'on College" (ast)ri Nagar" Bangalore -*+#
1
BUSINESS RESEARCH METHODS OB,ECTIVE The objective is to create an awareness of the Process of Research, the tools and techniques of research and generation of reports Unit . INTRODUCTION TO RESEARCH *Hrs Meaning – Objectives – Types of Research – Scope of Research – Research pproaches – Research Process – Research !esign – Research Methods "s Research Methodology # Steps in Research – Prob Proble le$ $ %or$ %or$ul ulati ation on – Stat State$ e$en entt of Resea Researc rch h Obje Object ctiv ivee – &' &'pl plor orat atory ory – !esc !escri ript ptiv ivee – &'peri$ental Research( Unit /. METHODS O0 DATA DATA CO11ECTION CO11ECTI ON 23 Hrs Observational and Survey Methods – %ield )or* Plan # d$inistration of surveys # Training field investigators # Sa$pling $ethods # Sa$ple si+e( Unit +. TOO1S 0OR CO11ECTION CO11ECT ION O0 DATA DATA 23 Hrs uestionnai uestionnaire re !esign!esign- ttitude ttitude $easure$ent $easure$ent techniques techniques – Motivation Motivational al Research Research Technique Techniquess – Selection of ppropriate ppropriate Statistical Techniques Techniques Unit *. STATISTICA1 STATISTICA1 METHODS 3 Hrs Tabulation of data # nalysis of data –Testing of .ypothesis, dvanced techniques – /O", 0hi# Square # !iscri$inant nalysis # %actor analysis, 0onjoint analysis # Multidi$ensional Scaling # 0luster nalysis 10oncepts Only2( Unit 4. REPORT 5RITIN6 23 Hrs Types of Reports, 3usiness, Technical and cade$ic Report writing – Methodology Procedure – 0ontents – 3ibliography S(I11 DEVE1OPMENT 4 5llustrate different types of sa$ples with e'a$ples e'a $ples 4 0onstruct a questionnaire for collection of pri$ary data *eeping in $ind the topic chosen for research 4 /arrate your e'perience using observation technique 4 !iagra$$atically present the infor$ation collected through the questionnaire BOO(S 0OR RE0ERENCE 6( O(R(7rishnaswa$y- Research $ethodology in Social Sciences, .P., 899:( 8( R( !ivivedi; Research Methods in 3ehavior 3ehav ior Science, Mac$illan 5ndia
(7( Sachdeva; 3usiness Research Methodology .P. ?( S(/( Murthy, "( "( 3hojanna; 3usiness Research Methods &'cel 3oo*s @(
2
Unit I INTRODUCTION TO RESEARCH
Meaning – Objectives – Types of Research – Scope of Research – Research pproaches – Research Process – Research !esign – Research Methods "s Research Methodology # Steps in Research – Prob Proble le$ $ %or$ %or$ul ulati ation on – Stat State$ e$en entt of Resea Researc rch h Obje Object ctiv ivee – &' &'pl plor orat atory ory – !esc !escri ript ptiv ivee – &'peri$ental Research( GGGGG
Meaning an De$inition o$ Resear78. Systematic investigative process employed to increase or revise current cur rent knowledge by discovering new facts.
5t is divided into two general categories; 162 3asic research is inquiry ai$ed at increasing scientific *nowledge, and 182 pplied research is effort effort ai$ed at using basic research for solving proble$s p roble$s or developing new processes, products, or techniques( Research is wor* that involves studying so$ething and trying trying to to discover facts facts about it(investigation it( investigation,, study,, inquiry study inquiry,, analysis analysis,, e'a$ination e'a$ination,, probe probe,, e'ploration e'ploration,, scrutiny scrutiny,, e'peri e'peri$en $entat tation ion,, delvin delving, g, groundwor* , fact#finding careful careful,, syste$at syste$atic, ic, patient study study and invest investiga igatio tion n in so$e field of *nowledge,, underta*en *nowledge underta*en to to discover or establish facts or principles De$inition o$ resear78
6( The collecting of infor$ation about a particular p articular subject 8( To study 1so$ething2 carefully =( To collect infor$ation about or for 1so$ething2 ?( 0areful or diligent search :( studious inquiry or e'a$ination- especially . investigation or e'peri$entation ai$ed at the discovery and interpretation of facts, revision of accepted theories or laws in the light of new facts, or practical application of such new or revised theories or laws 58at is Resear789
Research is a syste$atic inquiry that investigates hypothesis, suggests new interpretations of data or te'ts, and poses new questions for future research to e'plore( e 'plore( Research consists of: • • •
s*ing a question that nobody has as*ed before!oing the necessary wor* to find the answer- and 0o$$unicating the *nowledge you have acquired to a larger audience(
3
O!:e7tives o$ Resear78.
The purpose of research is to discover answers to questions through the application of scientific procedures( The $ain ai$ of research is to find out the truth which is hidden and which has not been discovered as yet( Though each research study has its own specific purpose, we $ay thin* of research objectives as falling into a nu$ber of following broad groupings; •
•
•
•
To gain fa$iliarity with a pheno$enon or to achieve new insights into it 1studies with this object in view are ter$ed as e'ploratory e' ploratory or for$ulative research studies2To portr portray ay accur accurat ately ely the the chara charact cteri erist stic icss of a part partic icul ular ar indi indivi vidu dual al,, situ situat atio ion n or a group1studies with this object in view are *nown as descriptive d escriptive research studies2To deter$ine the frequency with which so$ething occurs or with which it is associated with so$ething else 1studies with this object in view are *nown as diagnostic research studies2To test a hypothesis of a causal relationship between variables 1such studies are *nown as hypothesis#testing research studies2(
Types of Research : 6( !escriptive Research 8( &'ploratory Research =( pplied Research ?( %unda$ental Research @( uantitative Research B( ualitative Research D( 0onceptual Research :( &$pirical Research E( &'peri$ental Research 69( .istorical Research Spe7i$i7 to $iel o$ st)y
Craduate Craduate students learn about research $ethods used in their particular particular field of study( study( )hatever )hatever $ethods are used, there are $any resources to support research, and any nu$ber of variations to the basic $ethods( 0hoose a $ethod or variation that is $anageable in your you r first years as a new faculty $e$ber to be certain to get your research agenda underway( underway( Often the biggest challenge is often to to get started, so establish research priorities early on, and create a plan to i$pli$ent the$( ;)antitative Resear78
uantitative research is inquiry into an identified proble$, based on testing a theory, $easured with nu$bers, and analy+ed using statistical statistical techniques( The goal of quantitative $ethods $ethods is to deter$ine whether the predictive generali+ations of a theory hold true( )e will e'plore so$e of the issues and
4
challenges associated with quantitative research in this section( See* the advice of faculty $e$bers who have conducted quantitative studies for advice, support and encourage$ent( ;)alitative Resear78
study based upon a qualitative process of inquiry has the goal of understanding a social or hu$an proble$ fro$ $ultiple perspectives( ualitative research is conducted in a natural setting and involves a process of building a co$ple' and holistic holistic picture of the pheno$enon of interest( interest( )e will e'plore so$e of the issues and challenges associated with qualitative research in this section(
There are $any ways to collaborate and thereby enrich your wor* as a faculty $e$ber( Multiple perspectives offer a $ore co$plete view of an issue under study( )hether presenting or publishing, having several iterations of a study in different settings offers a $eans of validating findings( 0ollaboration $ay ta*e $any for$s, including crossing disciplines, types of institutions, or engaging the co$$unity outside outside the ca$pus environ$ent environ$ent(( &'plore &'plore the ways that collaboration collaboration $ay enrich your research( So$e of the issues and challenges associated with collaborative collaborative research are e'plored in this section( Pra7titioner Resear78
Si$ple definitions of practitioner research address the investigator, the setting and the purpose( The invest investiga igator tor is the practi practiti tioner oner,, in wor*pl wor*place ace settin settings gs rangin ranging g fro$ fro$ hos hospit pitals als,, to school schoolss and co$$unities co$$unities(( The general purpose is to better align the practitionerHs practitionerHs purpose purpose with their actions( There are those who argue that practitioner research ste$s fro$ a larger social justice $ove$ent within within qualita qualitativ tivee resear research( ch( &ven &ven when when social social justic justicee is not the sole sole $otiva $otivatin ting g princi principle ple,, an underlying underlying co$$onality co$$onality of purpose purpose is the desire desire to i$prove i$prove upon and develop deeper insights into oneHs oneHs practi practice( ce( Practi Practiti tioner oner resear research ch by its its nature nature offers offers practi practiti tioner onerss a voice voice in the resear research ch conversation( So$e consider it a bridge bridge of sorts between between theory and practice, although practitioners clai$ a rightful place in the research continuu$( So$e of the issues and challenges associated with practitioner research are e'plored in this section(
58at are t8e vario)s types o$ resear789 # P)re resear78 a( This is called as the funda$ental or the the theoretical research( b( 5s basic and original( c( 0an lead to the discovery of a new theory( d( 0an result in the develop$ent or refine$ent of a theory that already e'ists( e( .elps in getting *nowledge without thin*ing for$ally of i$ple$enting it in practice based on the honesty, love and integrity of the researcher for discovering the truth( 5
/# Applie resear78 a( 3ased on the concept of the pure research( b( 5s proble$ oriented( c( .elps in finding results or solutions for real life proble$s( d( Provides evidence of usefulness to society( e( .elps in testing e$pirical content of a theory( f( Ftili+es and helps in developing the techniques that can be used for basic research( g( .elps in testing the validity of a theory but under so$e conditions( h( Provides data that can lead to the acceleration of the process of generali+ation( +# E# A7tion resear78 a( Type of evaluation study( stud y( b( 5s a concurrent evaluation study( SCOPE O0 RESEARCH 6
6( To *now the 3usiness 0o$petition – &nviron$ent &nv iron$ent a( To *now the 0usto$ers;# 7now about custo$ers cu sto$ers needs, b( To *now the Product – !esign, Price, &'pectations 8( To *now the 5ndustry 0o$petition; =( Maturing of Manage$ent as a group of disciplines ? Sta*eholders de$anding greater influence @( To *now the Clobal 0o$petition B( Covern$ent 5ntervention D( &cono$ical !ata 0ollection
RESEARCH APPROACHES. The four main approaches (Types of
research) Approa78es o$ Resear78
6( uantitative pproach 8( ualitative pproach =( Prag$atic pproach ?( dvocacy I Participatory pproach Quantitative research uantitative research is generally associated with the positivistIpostpositivist paradig$( 5t usually involves collecting and converting data into nu$erical for$ so that statistical calculations can be $ade and conclusions drawn( The process Researchers will have one or $ore 8ypot8eses# These are the questions that they want to address which include predictions predictions about possible relationships relationships between the things they want to investigate investigate ?varia!les@ ( 5n order to find answers to these questions, the researchers will also have various instru$ents and $aterials 1e(g( paper or co$puter tests, observation chec* lists etc(2 and a clearly defined plan of action(
!ata is collected by various $eans following a strict procedure and prepared for statisti7al analysis ( /owadays, this is carried out with the aid of sophisticated statistical co$puter pac*ages( The analysis enables the researchers to deter$ine to what e'tent there is a relationship between two or $ore variables( This could be a si$ple association 1e(g( people who e'ercise on a daily basis have lower blood pressure2 or a causal relationship 1e(g( daily e'ercise actually leads to lower blood pressure2( Statistical analysis per$its researchers to discover co$ple' causal relationships and to deter$ine to what e'tent one variable influences another( 7
The results of statistical analyses are presented in journals in a standard way, the end result being a P val)e ( %or people who are not fa$iliar with scientific research jargon, the discussion sections at the end of articles in peer reviewed journals usually describe the results of the study and e'plain the i$plications of the findings in straightforward ter$s Principles Objectivity Objectivity is very i$portant in quantitativ quantitativee research( research( 0onsequently 0onsequently,, researchers researchers ta*e great care to avoid their own presence, presence, behaviour or attitude attitude affecting affecting the results 1e(g( by changing changing the situation situation being studied or causing participants to behave differently2( They also critically e'a$ine their $ethods and conclusions for any possible bias(
Researchers go to great lengths to ensure that they are really $easuring what they clai$ to be $easuring( %or e'a$ple, if the study is about whether bac*ground $usic has a positive i$pact on restlessness in residents in a nursing ho$e, the researchers $ust be clear about what *ind * ind of $usic to include, the volu$e of the $usic, what they $ean by restlessness, how to $easure restlessness and what is considered a positive i$pact( This $ust all be considered, prepared and controlled in advance( &'ternal factors, which $ight affect the results, $ust also be controlled for( 5n the above e'a$ple, it would be i$portant to $a*e sure that the introduction of the $usic was not acco$panied by other changes 1e(g( the person who brings the 0! player chatting with the residents after the $usic session2 as it $ight be the other factor which produces the results 1i(e( the social contact and not the $usic2( So$e possible contributing factors cannot always be ruled out but should be ac*nowledged by the researchers( The $ain e$phasis of quantitative research is on deductive reasoning which tends to $ove fro$ the general to the specific( This is so$eti$es referred to as a top down approach( The validity of conclusions is shown to be dependent on one or $ore pre$ises 1prior state$ents, findings or conditions2 being valid( ristotleHs fa$ous e'a$ple of deductive reasoning was; ll $en are $ortal JSocrates is a $an J Socrates is $ortal( 5f the pre$ises of an argu$ent are inaccurate, then the argu$ent argu$ent is inaccurate( inaccurate( This type of reasoning is often also associated with the fictitious fictitious character character Sherloc* .ol$es( .owever, $ost studies also include an ele$ent of inductive reasoning at so$e stage of the research 1see section on qualitative research for $ore details2( Researchers rarely have access to all the $e$bers of a particular group 1e(g( all people with de$entia, carers or healthcare professionals2( .owever, they are usually interested in being able to $a*e inferences fro$ their study about these larger groups( %or this reason, it is i$portant that the people involved in the study are a representative sa%ple of the wider populationIgroup( .owever, the e'tent to which generali+ations are possible depends to a certain e'tent on the nu$ber of people involved in the study, how they were selected and whether they are representative of the wider group( group( %or e'a$ple e'a$ple,, general generali+at i+ation ionss about about psy psychi chiatr atrist istss sho should uld be based based on a study study involv involving ing psychiatrists and not one based on psychology students( 5n $ost cases, rando$ sa$ples are preferred 1so that each potential participant has an equal chance of participating2 but so$eti$es researchers 8
$ight want to ensure that they include a certain nu$ber of people with specific characteristics and this would not be possible using rando$ sa$pling $ethods( Cenerali+ability of the results is not li$ited to groups of people but also to situations( 5t is presu$ed that the results of a laboratory e'peri$ent reflect the real life situation which the study see*s to clarify( )hen loo*ing at results, the P val)e is i$portant( P stands for probability( 5t $easures the li*elihood that a particular finding or observed difference is due to chance( The P value is between 9 and 6( The closer the result is to 9, the less li*ely it is that the observed difference is due to chance( The closer the result is to 6, the greater the li*elihood that the finding is due to chance 1rando$ variation2 and that there is no difference between the groupsIvariables(
Qualitative research ualitative research is the approach usually associated with the social constructivist paradig$ which e$phasises the socially constructed nature of reality( 5t is about recording, analysing and atte$pting to uncover the deeper $eaning and significance of hu$an behaviour and e'perience, including contradictory beliefs, behaviours and e$otions( Researchers are interested in gaining a rich and co$ple' understanding of peopleHs e'perience and not in obtaining infor$ation which can be generali+ed to other larger groups( The process The approach approach adopted adopted by qualit qualitati ative ve resear researche chers rs tends tends to be induct inductive ive which which $eans $eans that that they they develop a theory or loo* for a pattern of $eaning on the basis of the data that they have collected( This involves a $ove fro$ the specific to the general and is so$eti$es called a botto$#up approach( .owever, $ost research projects also involve a certain degree of deductive reasoning 1see section on quantitative research for $ore details2(
ualitative researchers do not base their research on pre#deter$ined hypotheses( /evertheless, they clearly identify a proble$ or topic that they want to e'plore and $ay be guided by a theoretical lens # a *ind of overarching theory theor y which provides a fra$ewor* for their investigation( The approach to data collection and analysis is $ethodical but allows for greater fle'ibility than in quantitative research( !ata is collected in te'tual for$ on the basis of observation and interaction with the participants e(g( through participant observation, in#depth interviews and focus groups( 5t is not converted into nu$erical for$ and is not statistically analysed( !ata collection $ay be carried out in several stages rather than once and for all( The researchers $ay even adapt the process $id#way, deciding to address additional issues or dropping questions which are not appropriate on the basis of what they learn during the process( 5n so$e cases, the researchers will interview or observe a set nu$ber of people( 5n other cases, the process of data collection and analysis $ay continue until the researchers find that no new issues are e$erging(
9
Principles Researchers will tend to use $ethods which give participants a certain degree of freedo$ and per$it spontaneity rather than forcing the$ to select fro$ a set of pre#deter$ined responses 1of which none $ight $ight be appropr appropriat iatee or accura accuratel tely y descri describe be the partic participa ipantH ntHss thought thoughts, s, feelin feelings, gs, attitu attitudes des or behaviour2 and to try to create the right at$osphere to enable people to e'press the$selves( This $ay $ean adopting a less for$al and less rigid approach than that used in quantitative research(
5t is believed that people are constantly trying to attribute $eaning to their e'perience( Therefore, it would $a*e no sense to li$it the study to the researcherHs view or understanding of the situation and e'pect to learn so$ething new about the e'perience of the participants( 0onsequently, the $ethods used $ay be $ore open#ended, less narrow and $ore e'ploratory 1particularly when very little is *nown about a particular subject2( The researchers are free to go beyond the initial response that the participant gives and to as* why, how, in what way etc( 5n this way, subsequent questions can be tailored to the responses just given( ualitative research often involves a s$aller nu$ber of participants( This $ay be because the $ethods used such as in#depth interviews are ti$e and labour intensive but also because a large nu$ber of people are not needed for the purposes of statistical analysis or to $a*e generali+ations fro$ the results( The s$aller nu$ber of people typically involved in qualitative research studies and the greater degree of fle'ibility does not $a*e the study in any way Kless scientificL than a typical quantitative study involving $ore subjects and carried out in a $uch $ore rigid $anner( The objectives of the two types of research and their underlying philosophical assu$ptions are si$ply different( .owever, as discussed in the section on Kphilosophies guiding researchL, this does not $ean that the two approaches cannot be used in the sa$e study(
Pragmatic approach to research (mixed methods The prag$atic approach to science involves using the $ethod which appears best suited to the research proble$ and not getting caught up in philosophical debates about which is the best approach( Prag$atic researchers therefore grant the$selves the freedo$ to use any of the $ethods, techni technique quess and procedu procedures res typica typically lly associ associate ated d with with quantit quantitati ative ve or qualit qualitati ative ve resear research( ch( They They recog recogni nise se that that ever every y $eth $ethod od has has its its li$i li$ita tati tion onss and that that the the diff differ eren entt appr approa oache chess can can be co$ple$entary( They $ay also use different techniques at the sa$e ti$e or one after the other( %or e'a$ple, they $ight start with face#to#face interviews with several people or have a focus group and then use the findings to construct a questionnaire to $easure attitudes in a large scale sa$ple with the ai$ of carrying out statistical analysis( !epending on which $easures have been used, the data collected is analysed in the appropriate $anner( .owever, it is so$eti$es possible to transfor$ qualitative data into quantitative data and vice versa although transfor$ing quantitative data into qualitative data is not very co$$on( 1!
3eing able to $i' different approaches has the advantages of enabling triangulation( Triangulation is a co$$on feature of $i'ed $ethods studies( 5t involves, for e'a$ple; •
the use o" a variet# o" data sources (data triangulation
•
the use o" several di$erent researchers (investigator triangulation
•
the use o" multiple perspectives p erspectives to interpret the results (theor# triangulation
•
the use o" multip multiple le methods methods to stud# stud# a resea researc rch h pro%le pro%lem m (m (metho ethodol dologi ogical cal triangulation
5n so$e studies, qualitative and quantitative $ethods are used si$ultaneously( 5n others, first one approach is used and then the ne't, with the second part of the study perhaps e'panding on the results of the first( %or e'a$ple, a qualitative study involving in#depth interviews or focus group discussions $ight serve to obtain infor$ation which will then be used to contribute towards the develop$ent of an e'peri$ental $easure or attitude scale, the results of which will be analysed statistically(
&dvocac#'participator# approach to research (emancipator# To so$e degree, researchers adopting an advocacyIparticipatory approach feel that the approaches to research described so far do not respond to the needs or situation of people fro$ $arginalised or vulnerable groups( s they ai$ to bring about positive change in the lives of the research subjects, their approach is so$eti$es described as e$ancipatory( 5t is not a neutral stance( The researchers are li*ely to have a political agenda and to try to give the groups they are studying a voice( s they want their research to directly or indirectly result in so$e *ind of refor$, it is i$portant that they involve the group being studied in the research, preferably at all stages, so as to avoid further $arginali+ing the$( The researchers $ay adopt a less neutral position than that which is usually required in scientific research( This $ight involve interacting infor$ally or even living a$ongst the research participants 1who are so$eti$es referred to as co#researchers in recognition that the study is not si$ply about the$ but also by the$2( The findings of the research $ight be reported in $ore personal ter$s, often using the precise words of the research participants( )hilst this type of research could by criticised for not being objective, it should be noted that for so$e groups of people or for certain situations, it is necessary as otherwise the thoughts, feelings or behaviour of the various $e$bers of the group could not be accessed or fully understood( "ulnerable groups are rarely in a position of power within society( %or this reason, researchers are so$eti$es $e$bers of the group they are studying or have so$ething in co$$on with the $e$bers of the group(
11
Resear78 Pro7ess - 2 Steps Research process contains a series of closely related activities which has to carry out by a researcher( Research process requires patients( There is no $easure that shows shows your research is the the best( 5t is an art rather than a science( %ollowing are the $ain steps in social or business research process(
6( Sele Select ctiion of Resear78 Pro!le% 8( &' &'te tens nsiv ivee
The selection of topic for research is a difficult job( )hen we select a title or research state$ent, then other activities would be easy to perfor$( So, for the understanding thoroughly the proble$ it $ust have to discuss with colleagues, friend, e'perts and teachers( The research topic or proble$ should be practical, relatively i$portant, feasible, ethically and politically acceptable( /# 1iterat) 1iterat)re re Revie& Revie& or E
fter the selection of research research proble$, the second step is that of literature literature $ostly connected with the topics( The availability of the literature $ay bring ease in the research( %or this purpose acade$ic journals, conference and govt( reports and library $ust be studied( +# Mai Maing ng Hypo Hypot8 t8es esis is
The develop$ent of hypothesis is a technical wor* depends on the researcher e'perience( The hypothesis is to draw the positive A negative cause and effect aspects of a proble$( .ypothesis narrows down the area of a research and *eep a researcher on the right path( *# Prep Prepari aring ng t8e t8e Resea Resear7 r78 8 Design Design
fter the for$ulation of the proble$ and creating hypothesis for it, research !esign is to prepare by the researcher( 5t $ay draw the conceptual structure of the proble$( ny type of research design $ay 12
be $ade, depend on the nature and purpose of the study( !aring R( !esign the infor$ation about sources, s*ill, ti$e and finance is ta*en into consideration( con sideration( 4# Sa%pling
The researcher $ust design a sa$ple( 5t is a plan for ta*ing its respondents fro$ a specific areas or universe( The sa$ple $ay be of two types; 6( Prob Probabi abili lity ty Sa$pl Sa$plin ing g 8( /on# /on#pr prob obab abil ilit ity y Sa$pl Sa$plin ing g =# Data Data 7oll 7olle7 e7ti tion on
!ata collection is the $ost i$portant wor*, is researcher( The collection of infor$ation $ust be containing on facts which is fro$ the following two types t ypes of researcher( Pri%ary Data Colle7tion. Pri$ary data $ay be fro$ the following(
6( &'peri$ ri$ent 8( ues uesttionn ionnai aire re =( Observat vation ?( 5nterview Se7onary ata 7olle7tion. it has the following categories;
6( Revi Review ew of lite litera ratu ture re 8( Offici Official al and and non#of non#offi ficia ciall report reportss =( # Data Data Anal nalysis ysis
)hen data is collected, it is forwarded for analysis which is the $ost technical job( !ata analysis $ay be divided into two $ain categories( Data Pro7essing. it is sub#divided into the following(
!ata editing, !ata coding, !ata classification, !ata tabulation, !ata presentation, !ata $easure$ent Data E
!escription, &'planation, /arration, 0onclusionI%indings, Reco $$endationsISuggestions 3# Hypo Hypot8 t8es esis is Te Testin sting g 13
Research data is then forwarded to test the hypothesis( !o the hypothesis are related to the facts or not To find the answer the process of testing hypothesis is underta*en which $ay result in accepting or rejecting the hypothesis( # 6en 6eneral erali'at i'ation ion an Interpr Interpreta etation tion
The acceptable hypothesis is possible for researcher to arrival at the process of generali+ation or to $a*e A theory( So$e types of research has no hypothesis for which researcher depends upon on theory which is *nown as interpretation( 2# Preparation Preparation o$ Report Report
researcher should prepare a report for which he has done is his h is wor*( .e $ust *eep in his $ind the following points; Report Design in Pri%ary Stages
The report should carry a title, brief introduction of the proble$ and bac*ground followed by ac*nowledge$ent( There should be a table of contents, grapes and charts( Main Te
5t should contain contain objectives, objectives, hypothesis, hypothesis, e'planations e'planations and $ethodology $ethodology of the research( 5t $ust be divided into chapters and every chapter e'plains separate title in which su$$ary of the findings should should be enlisted( enlisted( The last section would be clearly clearly of conclusions conclusions to show the $ain the$e of the R#study( Closing t8e Report
fter the preparation of report, the last step in business research process contains of bibliography, references, appendices, inde' and $aps or charts for illustration( %or this purpose the infor$ation should $ore clearer(
RESEARCH DESI6N Meaning.
The resear78 refers to the overall overall strate strategy gy that that you choose choose to integr integrate ate the differe different nt resear78 esign refers co$ponents of the study in a coherent and logical way, thereby, thereby, ensuring you will effectively address the resear78 proble$- it constitutes the blueprint for the collection, $easure$ent, and analysis of data( detailed outline of how an investigation will ta*e place( research design will typically include how data is to be collected, what instru$ents will be e$ployed, how the instru$ents will be used and the intended $eans for analy+ing data collected(
The key features of a research design are:
14
1i2 5t is a plan which describes the sources and *inds of infor$ation strongly related to the research proble$( 1ii2 5t is a strategy indicating which $ethod will be e$ployed for collecting and e'a$ining the data( 1iii2 5t also consists of the ti$e and cost budgets because $ost studies are done under these two li$itations( 5n a nutshell, research design $ust, at least, contain 1a2 a clear state$ent of the research proble$1b2 processes and $ethods to be utili+ed for collecting data 1c2 the population to be researched- and 1d2 techniques to be e$ployed in processing and e'a$ining data( Research design is what $a*es the entire research project p roject wor* ( More e'plicitly, the the design decisions happen to be in respect of; •
)hat is the study about
•
)hy is the study being $ade
•
)here will the study be carried out
•
)hat type of data is required
•
)here can the required data be found
•
)hat periods of ti$e will the study include
•
)hat will be the sa$ple design
•
)hat techniques of data collection will be used
•
.ow will the data be analy+ed
•
5n what style will the report be prepared
Avantages o$ resear78 esign
6( 0ons 0onsu$ u$es es les lesss ti$e ti$e(( 8( &ns &nsure uress proj project ect ti$e ti$e sche schedul dule( e( 15
=( .elps researc researcher her to prepare prepare hi$self hi$self to carry out researc research h in a proper and a syste$atic syste$atic way( way( ?( 3etter 3etter docu$entatio docu$entation n of the various various activiti activities es while while the project project wor* is is going on( @( .elps in in proper planning planning of the the resources resources and their their procure$ procure$ent ent in right right ti$e( ti$e( B( Provides Provides satisfact satisfaction ion and confidence, confidence, acco$pan acco$panied ied with with a sense of success success fro$ fro$ the beginning of the wor* of the research project(
Nee $or Resear78 Design Research design is needed because it facilitates the s$ooth sailing of the various research operations, thereby $a*ing research as efficient as possible yielding $a'i$al infor$ation with $ini$al e'penditure of effort, ti$e and $oney( Research design has a significant i$pact on the reliability of the results obtained( 5t thus acts as a fir$ foundation for the entire research( %or e'a$ple, econo$ical and attractive construction of house we need a blueprint 1or what is co$$only called the $ap of the house2 well thought out and prepared by an e'pert architect, si$ilarly we need a research design or a plan in advance of o f data collection and analysis for our research project( Research design stands for advance planning of the $ethods to be adopted for collecting the relevant data and the techniques to be used in their analysis(
T8e nee $or resear78 esign is as $ollo&s. •
5t reduces inaccuracy-
•
.elps to get $a'i$u$ efficiency and reliability-
•
&li$inates bias and $arginal errors-
•
Mini$i+es wastage of ti$e-
•
.elpful for collecting research $aterials-
•
.elpful for testing of hypothesis-
•
Cives an idea regarding the type of resources required in ter$s of $oney, $anpower, ti$e, and efforts-
•
Provides an overview to other e'pertse'p erts-
•
Cuides the research in the right direction(
Resear78 Met8os Vs Resear78 Met8oology
16
De$inition o$ Resear78 Met8o Research $ethod pertains to all those $ethods, which a researcher e$ploys to underta*e research process, to solve the given proble$( The techniques and procedure, that are applied during the course of studying research proble$ are *nown as the research $ethod( 5t enco$passes both qualitative and quantitative $ethod of perfor$ing research operations, such as survey, case study, interview, questionnaire, observation, etc(
These are the approaches, which help in collecting data and conducting research, in order to achieve specific objectives such as theory testing or develop$ent( ll the instru$ents and behaviour, used at various levels of the research activity such as $a*ing observations, data collection, data processing, drawing inferences, decision $a*ing, etc( are included in it( Research $ethods are put into three categories; •
•
•
co vered( Such $ethods are used 0irst Category ; The $ethods relating to data collection are covered( when the e'isting data is not sufficient, to reach the solution( Se7on Category. 5ncorporates the processes of analysing data, i(e( to identify patterns and establish a relationship between data and un*nowns( un *nowns( T8ir Category ; 0o$prise of the $ethods which are used to chec* the accuracy of the results obtained(
hat are )esearch *ethods+ s $entioned $entioned above, resear78 %et8os are t8e %et8os )se $or ata 7olle7tion in resear78 ( Research methods involve surveys, interviews , case , case studies , observation , experiments, etc( 5t can be said that research $ethods are $ainly used to gather infor$ation so that the researcher can find answers to his research proble$( )hen spea*ing spea*ing of research research $ethods whether whether it is the natural sciences sciences or else the social sciences there is a vast range of $ethods that can be used( 5n the natural sciences, the researcher is $ostly interested in gaining quantitative data that will allow hi$ to provide specific conclusions( 3ut in the social sciences the research $ethods $ostly provide the researcher with quantitative data( .owever, this does not $ean that in social sciences qualitative data is ignored( On the contrary, a co$bination of data can be used for social research( De$inition o$ Resear78 Met8oology Research Methodology, as its na$e suggest is the study of $ethods, so as to solve the research proble$( 5t is the science of learning the way wa y research should be perfor$ed syste$atically( 5t refers to the rigorous analysis of the $ethods applied in the strea$ of research, to ensure that the conclusions drawn are valid, reliable and credible too(
The researcher ta*es an overview of various steps that are chosen by hi$ in understanding the proble$ at hand, along with the logic behind the $ethods e$ployed by the researcher during study( 17
5t also clarifies the reason for using a particular $ethod or technique, and not others, so that the results obtained can be assessed either by the researcher hi$self or any other party(
hat is a )esearch *ethodolog#+ Resear78 Resear78 %et8oology %et8oology e
hat are the ,i$erence -et.een )esearch *ethods and )esearch *ethodolog#+ Denitions of Research Methods and Research Methodology: Resear78 %et8os. Research $ethods are the $ethods used for data collection in a research( Resear78 %et8oology. Research $ethodology e'plains the overarching theoretical and philosophical fra$ewor*s which guide the research( Characteristics of Research Methods and Research Methodology: Content. Resear78 Met8os. Research $ethods involve surveys, interviews, case studies, observation, e'peri$ents, etc( Resear78 Met8oology. Research $ethodology involves the theoretical fra$ewor*s and learning of the various techniques that can be used in the conduct of research and the conduct of tests, e'peri$ents, surveys and critical studies( Ai%. Resear78 Met8os. Research $ethods ai$ at finding solutions to research proble$s( 18
Resear78 Met8oology. Research $ethodology ai$s at the e$ploy$ent of the correct procedures to find out solutions( Relations8ip. Resear78 Met8os. Research $ethods are the end of any research( Resear78 Met8oology. Research $ethodology is the beginning(
/e# ,i$erences -et.een )esearch *ethod and )esearch *ethodolog# The differences between research $ethod and research $ethodology can be drawn clearly on the following grounds; 6( The resear research ch $ethod $ethod is defined defined as the procedure procedure or techniq technique ue applied applied by the resear researcher cher to underta*e research( On the other hand, research $ethodology is a syste$ of $ethods, used scientifically for solving the research proble$( 8( The research research $ethod $ethod is nothing nothing but the the behaviour behaviour or tool, tool, e$ployed e$ployed in selecti selecting ng and building building research technique( 0onversely, research $ethodology i$plies the science of analysing, the $anner in which research is conducted appropriately( =( The research research $ethod $ethod is concerned concerned with carrying carrying out e'peri$ e'peri$ent ent,, test, test, surveys, surveys, intervi interviews ews,, etc( s against this, research $ethodology is concerned with learning various techniques which can be e$ployed in the perfor$ance of e'peri$ent, test or survey( ?( Resear Research ch $ethod $ethod covers covers variou variouss invest investiga igatio tion n techni techniques ques(( Fnli*e Fnli*e,, resear research ch $ethod $ethodolo ology gy,, which consists of co$plete approach aligned towards the attain$ent of purpose( @( Research Research $ethod intends intends to discover discover the solution solution to the the proble$ at hand( hand( 5n contrast, contrast, research research $ethodology aspires to apply appropriate procedures, with a view to ascertaining solutions( Con7l)sion The scope of research $ethodology is wider than that of research $ethod, as the latter is the part of the for$er( %or understanding the research proble$ thoroughly, the researcher should *now the research $ethodology along with the $ethods(
5n a nutshell, research $ethod refers to the technique which can be adopted to e'plore the nature of the world that surrounds us( On the contrary, research $ethodology is the foundation, which helps us to understand the deter$inants influencing the effectiveness of the $ethods applied(
0ormulating the research pro%lem Once the general topic or proble$ has been identified, this should then be stated as a clear resear78 pro!le%, that is, ta*en fro$ just a state$ent about a proble$atic situation to a clearly defined researchable proble$ that identifies the issues you are trying to address( 19
5t is not always easy to for$ulate the research proble$ si$ply and clearly( 5n so$e areas of scientific research the investigator $ight spend years e'ploring, thin*ing, and researching before they are clear about what research questions they are see*ing to answer( Many topics $ay prove too wide#ranging to provide a researchable proble$( 0hoosing to study, for instance a social issue such as child poverty, does not in itself provide a researchable proble$( The proble$ is too wide#ranging for one researcher to address( Ti$e and resources would $a*e this unfeasible and the results fro$ such a study would consequently lac* depth and focus( Statement of research problem
n adequate state$ent of the research proble$ is one of the $ost i$portant parts of the research( !ifferent researchers are li*ely to generate a variety of researchable proble$s fro$ the sa$e situation since there are $any research issues that can arise out of a general proble$ situation( Nour research will be able to pursue only one in depth( %or a proble$ state$ent to be effective in the planning of applied research it should have the following characteristics 1ndrew and .ildebrand 6E:82( 6( The proble proble$ $ refl reflect ectss felt felt needs needs 8( The proble$ proble$ is non#hypothet non#hypothetical, ical, ie ie it $ust $ust be based based on factual factual evidence evidence =( 5t should should suggest $eaningf $eaningful ul and testable testable hypotheses hypotheses # to avoid avoid answers answers that are are of little little or no use to the alleviation of the proble$ ?( The probl proble$s e$s shoul should d be relevan relevantt and $anage $anageable able %or$ulating the research proble$ allows you to $a*e clear, both to yourself and the reader, what the purpose of your research is( Subsequent elaboration of $ethod should be oriented to providing infor$ation to address that proble$( The proble$ state$ent is therefore a very i$portant device for *eeping you on trac* with your research( 5t is also one $eans by which your research will be evaluated # does the research address add ress the proble$ as stated(
5 Ways to ormulate the Research Pro!lem 1 peci"# the )esearch %ectives clear state$ent of objectives will help you develop effective research. research. 5t will help the decision $a*ers evaluate your project( p roject( 5tHs critical that you have $anageable objectives( 1Two 1Two or three clear goals will help to *eep your research project focused and an d relevant(2
2 )evie. the nvironment or ontext o" the )esearch Pro%lem s a $ar*eting researcher, you $ust wor* closely with your tea$( This will help you deter$ine whether the findings of your project will produce enough infor$ation to be worth the cost( 2!
5n order to do this, you have h ave to identify the environ$ental variables that will affect the research project(
3 xplore the ature o" the Pro%lem Resear78 pro!le%s range fro$ si$ple to co$ple', depending on the nu$ber of variables and the nature of their relationship(
5f you understand the nature of the problem as a researcher researcher , you will be able to better develop a solution for the proble$( To help you understand all di$ensions, you $ight want to consider focus groups of consu$ers, sales people, $anagers, or professionals to provide what is so$eti$es $uch $uch needed insight(
4 ,ene the aria%le )elationships Mar*eting plans often focus on creating a sequence of behaviors that occur over ti$e, as in the adoption of a new pac*age design, or the introduction of a new product( Such progra$s create a co$$it$ent to follow so$e behavioral pattern in the future( Studying such a process involves; •
!eter$ining which variables affect the solution to the proble$(
•
!eter$ining the degree to which each variable can be controlled(
•
!eter$ining the functional relationships between the variables and which variables are critical to the solution of the proble$(
!uring the pro!le% $or%)lation stage, you will want to generate ge nerate and consider as $any courses of action and variable relationships as possible(
5 :he onse;uences o" <ernative ourses o" &ction There are always consequences to any course of action( nticipating and co$$unicating the possible outco$es of various courses of action is a pri$ary responsibility responsibility in the research process( process( Resear78 O!:e7tives an Hypot8eses Resear78 O!:e7tives are state$ents of what the researcher intends to do( The objectives flow directly fro$ the proble$( They co$$unicate what the researcher plans to do( Structurally, the objectives are seen as s$all particles which constitute the proble$(
21
The proble$ $ay be stated broadly but the objective should be should be stated in $ore specific and $easurable ter$; 0or e
The i%%eiate or general o!:e7tive specifies the activity that will ta*e place and the variables that will be e'a$ined( The Spe7i$i7 o!:e7tives $ay be viewed as sub#objectives or s$all particles of the general objective( The following should be e'a$ined in stating the specific objectives; 6( Specific variables 8( "ariables "ariables e'pressed in $easurable ter$s =( Suggestion on the type of analysis to be done
Resear78 Hypot8esis 8ypot8esis is defined as an educated guess or a tentative answer to the research question ( it is a state$ent about an e'pected relationship betwwn two variables( Types o$ 8ypot8eses 1$ain categories2 n)ll 8ypot8esis is a negative state$ent which indicate s no relationship nor correlation between two variables( Example Kdeter$ine whether there is a significant significant relationship relationship between the e'tent of $ass (iven )b*ective; Kdeter$ine $edia e'posure and attitude towards land refor$ a$ong lowland far$ersL The +ull ,ypothesis; K There is no significant relationship between the e'tent of $ass $edia e'posure and attitude towards land refor$ a$ong lowland far$ers(L
n alternative 8ypot8esis is also called the resear78 8ypot8esis ( 5t is the positive for$ of the null hypothesis( Example The +ull ,ypothesis- “ There is a sinificant relationship bet!een mass media exposure and attitude to!ards land reform amon lo!land farmers.”
22
Ot8er 7lassi$i7ations o$ 8ypot8eses
!irectional vs( /on#directional hypotheses ire7tional hypothesis states whether the relationship between two variables is direct 1positive2 or inverse 1negative2( positive or direct relationship is present when the value of one variable increases with the increase in the value of another( negative relationship is present when the value one variable increases as the value of other variable decreases( Example“ The hiher the level of exposure of farmers to mass media" the more favourable their attitude to!ards land refor” positive “ The more time employees spend in meetins" the less productive they are.” are.” neative “ the hiher the advertisement expenses of food establishments" the hiher their monthly ross sales.” positive
non-ire7tional hypothesis does not specify the direction of relationship between variables( 5t $erely states the presence or absence of a relationship between two variables or that one influences 1 or does not influence2 the other( Example“The !omen’s educational attainment is sinificantly associated !ith their extent of participation in decisionma/in at home.”
TPES O0 RESEARCH # E
23
E
Once upon a ti$e, so$eone had the idea that the world was flat and that if you went too far, you would surely fall off( )e now *now that is not n ot true( )e *now this because of &R( )hen you conduct &R, you are an e'plorer, li*e Magellan or
/# DESCRIPTIVE RESEARCH De$inition
s the na$e i$plies, descriptive research $ethods are used when the researcher !ants to describe specific behavior as it occurs in the environment environment ( There are a variety of descriptive research $ethods available, and once again, the nature of the question that needs to be answered drives which $ethod is used( Traditionally, descriptive descriptive research involves three $ain categories; observation" case studies" a t each of these individually( and surveys(
:( Cenerali+ations andIor predictions
*# E
Observational and Survey Methods – %ield )or* Plan # d$inistration d$inistration of surveys # Training field investigators # Sa$pling $ethods # Sa$ple si+e(
25
Data Collection De$inition !ata collection is the process of gathering and $easuring data, infor$ation or any variables of intere interest st in a standa standardi rdi+ed +ed and establ establish ished ed $anner $anner that that enables enables the collec collector tor to answer answer or test test hypothesis and evaluate outco$es of the particular collection( This is an integral, usually initial, co$ponent of any research done in any field of study such as the physical and social sciences, business, hu$anities and others( !ata collection is concerned with the accurate acquisition of data- although $ethods $ay differ depending on the field, the e$phasis on ensuring accuracy re$ains the sa$e( The pri$ary goal of any data collection endeavor is to capture quality data or evidence that easily translates to rich data analysis that $ay lead to credible and conclusive answers to questions that have been posed( !ata collection is the syste$atic approach to gathering and $easuring infor$ation fro$ a variety of sources to get a co$plete and accurate picture of an area of interest( !ata !ata collection collection enables a person or organi+ation organi+ation to answer relevant relevant questions, evaluate outco$es and $a*e predictions about predictions about future probabilities and trends(
Types of Data: Primary and "econdary data There are $any ways of classifying data( co$$on classification is !ase )pon who collected the data( Pri%ary ata; !ata collected by the investigator hi$selfI herself for a specific purpose( Examples; !ata collected by a student for hisIher thesis or research project(
15n $ovies2 The hero is directly told by the heroine that he is her Kideal $anL( Se7onary ata ; !ata collected by so$eone else for so$e other purpose 1but being utili+ed by the investigator for another purpose2( Examples; 0ensus data being used to analy+e the i$pact of education on career choice and earning(
So%e Avantages o$ )sing Pri%ary ata ;
6( The investig investigator ator collects collects data specifi specificc to the proble proble$ $ under study study(( 8( There is no no doubt about about the quality quality of the the data collected collected 1for 1for the investi investigator2 gator2(( =( 5f required, required, it it $ay be possible possible to obtain obtain addition additional al data during during the study study period( period(
26
So%e Disavantages o$ )sing Pri%ary ata ?$or rel)7tant )nintereste investigators@; 6( The investig investigator ator has has to contend contend with with all the the hassles hassles of data data collectio collection# n# •
deciding why, what, how, when to collect
•
getting the data collected 1personally or through others2
•
getting funding and dealing with funding agencies
•
ethical considerations 1consent, per$issions, etc(2
8( &nsuring the data data collected collected is of a high standard# standard# •
all desired data is obtained accurately, and in the for$at it is required in
•
there is no fa*eI coo*ed up data
•
unnecessaryI useless data has not been included
=( 0ost of obtaining obtaining the data is often the $ajor e'pense in studies So%e Avantages o$ )sing Se7onary ata. 6( The dataH dataHss already already there# there# no hassles hassles of data data collectio collection n 8( 5t is less less e'pe e'pens nsiv ivee =( The investig investigator ator is not persona personally lly responsib responsible le for the quality quality of data 1K5 didnHt didnHt do itL2 itL2
So%e isavantages o$ )sing Se7onary ata. 6( The investig investigator ator cannot cannot decide decide what is collect collected ed 1if specific specific data about about so$ething so$ething is required required,, for instance2( 8( One can can only only hope that the data is of of good good quali quality ty =( Obtaining Obtaining additio additional nal data 1or 1or even clarificati clarification2 on2 about about so$ething so$ething is not possibl possiblee
Data Colle7tion Te78ni)es 5nfor$ation you gather can co$e fro$ a range of sources(
27
•
5nterviews
•
uestionnaires and Surveys
•
Observations
•
%ocus Croups
•
&thnographies, Oral .istory, and 0ase Studies
•
!ocu$ents and Records
vervie. o" ,i$erent :echni;ues ,ata ollection :echni;ues Te78ni)e
(ey 0a7ts
•
•
Intervie&s •
•
•
;)estionnaires an S)rveys
•
•
E
structured= One#on#one conversation with parent or in"ormall# of at#ris* youth who can help you Questions should %e "ocused= clear= and encourage open> ended responses
understand the issueto see a sa$ple *ey infor$ant interview(
)esponses can %e anal#?ed .ith ;uantitative methods %# assigning numerical values to @iAert>t#pe scales
Results of a satisfaction survey or opinion survey to see an e'a$ple of a survey created using the 0N%&RnetS&R0. )esults are generall# easier (than ;ualitative techni;ues to 5nteractive Survey 3uilder feature(to anal#?e see a sa$ple survey on $iddle school youth ris* behavior( behavio r( Pretest'Posttest Pretest'Posttest can %e compared and anal#?ed
28
•
&llo.s "or the stud# o" the d#namics o" a situation= "re;uenc# counts o" target %ehaviors= or other %ehaviors as indicated %# needs o" the evaluation
Site visits to an after#school progra$
O!servations
•
•
•
0o7)s 6ro)ps
•
•
•
•
Et8nograp8ies " Oral History" an Case St)ies
•
•
•
Bood source "or providing to docu$ent the interaction between additional in"ormation a%out a particular group= can use video youth and staff within the progra$ to provide documentation an produce ;ualitative (eg= narrative data and ;uantitative data (eg= "re;uenc# counts= mean length o" interactions= and instructional time
& "acilitated group intervie. .ith individuals that have something in common Bathers in"ormation a%out com%ined perspectives and opinions
group of parents of teenagers in an after#school progra$ are invited to infor$ally discuss progra$s that $ight benefit and help their children succeed
)esponses are o"ten coded into categories and anal#?ed thematicall#
29
Shadowing a fa$ily while recording e'tensive field notes to study the e'perience and issues associated with youth who have a parent or guardian that has been deployed for an e'a$ple of an oral oral history( history( for an additional e'a$ple of an oral history(
•
Do7)%ents an Re7ors •
onsists o" examining existing data in the "orm o" data%ases= meeting minutes= reports= attendance logs= nancial records= ne.sletters= etc
To understand the pri$ary reasons students $iss school, records on student absences are collected and analy+ed for an e'a$ple of a searchable database of aggregate data :his can %e an inexpensive .a# on youth ris* behavior( to gather in"ormation %ut ma# %e an incomplete data source
0IE1D 5OR( P1AN.
Meaning collectio tion n of infor$ infor$ati ation on outsid outsidee a laborat laboratory ory,, library library or 0iel resear78 resear78 or $iel&or is the collec wor*place setting( The approaches and $ethods used in field research vary across disciplines( %or e'a$ple, biologists who conduct field research $ay si$ply observe ani$als interacting with their environ$ents, whereas social scientists conducting field research $ay interview or observe people in their natural environ$ents to learn their languages, fol*lore, and social structures( %ield research involves a range of well#defined, although variable, $ethods; infor$al interviews, direct observation, participation in the life of the group, collective discussions, analyses of personal docu$ents produced within the group, self#analysis, results fro$ activities underta*en off# or on# line, and life#histories( lthough the $ethod generally is characteri+ed as qualitative research, it $ay 1and often does2 include quantitative di$ensions(
A%inistration o$ S)rveys.
S)rvey Resear78 an A%inistration S)rvey resear78 is a $ethod in which data is collected fro$ a target population, called the sa$ple, by personal interviews, online surveys, the telephone, or paper questionnaires( So$e for$s of survey research research such as online surveys surveys $ay be co$pleted co$pleted in an auto$ated auto$ated fashion( fashion( The professional professionalss at Statistics Solutions provide survey ad$inistration help to $asterHs and doctoral candidates in the survey ad$inistration phase of their research(
The choice of survey instru$ent1s instru$ent1s22 used to gather data for your thesis or dissertati dissertation on is critical( critical( 5f you are planning to create your own survey instru$ent and ad$inister it online 1e(g(, Survey Mon*ey, uestionPro, Psych!ata or oo$erang2, Statistics Solutions can help you create the survey questions and any subscales so they can be easily analy+ed and answer your research questions( Our consultants can then help you validate your instru$ent and e'pedite the 5R3 approval process by helping you avoid the typical t ypical university and co$$ittee pitfalls(
Sa%pling %et8os
3!
Meaning o$ Sa%pling. Sa$pling is the process of selecting units 1e(g(, people, organi+ations2 fro$ a population of interest so that by studying studying the sa$ple sa$ple $ay fairly fairly generali+e our results bac* to the population population fro$ which they were chosen(
•
Pro!a!ility Sa%pling ; Sa$ple has a *nown probability of being selected Non-pro!a!ility Sa%pling ; Sa$ple does not have *nown probability of being selected as in convenience or voluntary response surveys
Pro!a!ility Sa%pling
5n probability sa$pling it is possible to both deter$ine which sa$pling units belong to which sa$ple and the probability that each sa$ple will be selected( The following sa$pling $ethods are e'a$ples of pro!a!ility sa%pling ; 6( Si%ple Rano% Sa%pling ?SRS@ 8( Strati$ie Sa%pling =( Cl)ster Sa%pling ?( Syste%ati7 Sa%pling @( M)ltistage Sa%pling ?in &8i78 so%e o$ t8e %et8os a!ove are 7o%!ine in stages@ Of the five $ethods listed above, students have the $ost trouble distinguishing between strati$ie sa%pling and 7l)ster sa%pling ( popu lation into groups based on Strati$ie Sa%pling is possible when it $a*es sense to partition the population a factor that $ay influence the variable variable that is being $easured( These groups are then called strata( n individual group is called a stratu$( )ith )ith strati$ie sa%pling one should; •
partition the population into groups 1strata2
•
obtain a si$ple rando$ sa$ple fro$ each group 1stratu$2
•
collect data on each sa$pling unit that was rando$ly sa$pled fro$ each group 1stratu$2
Strati$ie sa%pling wor*s best when a heterogeneous population is split into fairly ho$ogeneous groups( Fnder these conditions, stratification stratification generally produces $ore precise precise esti$ates of the
31
population percents than esti$ates that would be found fro$ a si$ple rando$ sa$ple( Ta!le +#/ shows so$e e'a$ples of ways to obtain a stratified sa$ple( )ith 7l)ster sa%pling one should Cl)ster Sa%pling is very different fro$ Stratified Sa$pling( )ith •
divide the population into groups 1clusters2(
•
obtain a si$ple rando$ sa$ple of so $any $an y clusters fro$ all possible clusters(
•
obtain data on every sa$pling unit in each of the rando$ly selected clusters(
5t is i$portant to note that, unli*e with the strata in stratified sa$pling, the clusters should be $icrocos$s, rather rather than subsections, subsections, of the population( &ach cluster should be heterogeneous( heterogeneous( dditionally, the statistical statistical analysis used with cluster sa$pling is not only on ly different, but also $ore co$plicated than that used with stratified sa$pling( Non-pro!a!ility Sa%pling
The following sa$pling $ethods that are listed in your te't are types of non-pro!a!ility sa%pling t8at s8o)l !e avoie ; 1 #olunteer samples 2 hapha$ard (con#enience) samples
Since s)78 non-pro!a!ility sa%pling %et8os are based on hu$an choice rather than rando$ selection, statistical theory cannot e'plain how they $ight be have and potential sources of bias are ra$pant( 5n your te'tboo*, the two two types of non#probability sa$ples listed listed above are called sa$pling disasters(
Sa%ple si'e. T8e De$inition o$ Sa%ple Si'e
Sa$ple si+e $easures the nu$ber of individual sa$ples $easured or observations used in a survey or e'peri$ent( Sample si0e is an i$portant concept in statistics, and refers to the nu$ber of individual pieces of data collected in a survey( survey or statistics sa$ple si+e is i$portant in deter$ining the a77)ra7y an relia!ility of a surveys findings( The sa$ple si+e is an i$portant i$portant feature feature of any e$pirical e$pirical study in which the goal is to $a*e inferences about a population fro$ a sa$ple( 5n practice, the sa$ple si+e used in a study is deter$ined based on the e'pense e'pen se of data collection, and the need to have sufficient statistical power(
hat is a Dample i?eE+
32
sa$ple si+e is a part o$ t8e pop)lation chosen for a survey or e'peri$ent( %or e'a$ple, you $ight ta*e a survey of dog ownerHs brand preferences( Nou wonHt want to survey all the the $illions of dog owners in the country 1either because itHs too e'pensive or ti$e consu$ing2, so you ta*e a sa$ple si+e( That $ay be several thousand owners( The sa$ple si+e is a representation of all dog ownerHs brand preferences( 5f you choose your sa$ple wisely, it will be a good representation(
"ample si$e determination choosing g the nu$ nu$ber ber of obs observ ervati ations ons or replicates to Sa%ple si'e eter%ination is the act of choosin include in a statistical sa$ple( sa$ple( The sa$ple si+e is an i$portant feature of any e$pirical study in which the goal is to $a*e inferences about inferences about a population a population fro$ fro$ a sa$ple( 5n practice, the sa$ple si+e used in a study is deter$ined based on the e'pense of data collection, and the need to have sufficient statistical power ( 5n co$plicated studies there $ay be several different sa$ple si+es involved in the study; for e'a$ple, in a stratified survey there survey there would be different sa$ple si+es for each stratu$( 5n a census, census, data are collected on the entire population, hence the sa$ple si+e is equal to the population si+e( 5n e'peri$ental design, design, where a study $ay be divided into different treat$ent groups, this groups, this $ay be different sa$ple si+es for each group( Sa$ple si+es $ay be chosen in several different ways; •
e'perience – for e'a$ple, include those ite$s readily available or convenient to collect( choice of s$all sa$ple si+es, though so$eti$es so$eti$es necessary necessary,, can result result in wide confidence intervals or ris*s of errors in statistical hypothesis testing( testing (
•
using a target variance for an esti$ate to be derived fro$ the sa$ple eventually obtained
•
using a target for the power of a statistical test to test to be applied once the sa$ple is collected(
•
using a confidence level deter$ines how accurate a result will turn out with lower chances of error( GGGGGGGGGGGGGGGGGGGG Unit III TOO1S 0OR CO11ECTION O0 DATA
uestionnai uestionnaire re !esign!esign- ttitude ttitude $easure$ent $easure$ent techniques techniques – Motivation Motivational al Research Research Technique Techniquess – Selection of ppropriate Statistical Techniques Techniques TOO1S 0OR CO11ECTION O0 DATA Intro)7tion.
•
!ata !ata 0ollec 0ollectio tion n is an i$port i$portant ant aspect aspect of any type type of resear research ch study study(( 5naccu 5naccurat ratee data data collection can i$pact the results of a study and ulti$ately lead to invalid results(
33
•
•
!ata collection $ethods for i$pact evaluation vary along a continuu$( t the one end of this continuu$ are quantitative $ethods and at the other end of the continuu$ are ualitative $ethods for data collection ( !epending on the nature of the infor$ation to be gathered, different instru$ents are used to conduct the assess$ent; for$s for gathering data fro$ official sources such as police or school records- surveysIinterviews to gather infor$ation fro$ youth, co$$unity residents, and others- and focus groups to elicit free#flowing perspectives(
TPES O0 TOO1S
The various $ethods of data gathering involve the use of appropriate recording for$s( These are called tools or instru$ents of data collection( They consist of • Observation schedule • 5nterview guide • 5nterview schedule • Mailed questionnaire • Rating scale • 0hec*list • !ocu$ent scheduleIdata sheet • Schedule for institutions • &ach of the above tools is used for a specific $ethod of data gathering; Observation schedule for for obse observ rvat atio ion n $eth $ethod, od, inte interv rvie iew w sche schedul dulee and and inte interv rvie iew w guide guide for for inter intervi view ewin ing, g, questionnaire for $ail survey, and so on(
%uestionnaire Design Process De$inition. ;)estionnaire is a syste$atic, data collection technique consists of a series of questions required to be answered by the respondents to identify their attitude, e'perience, and behavior towards the subject of research(
One of the $ost critical parts of the survey is the creation of questions that $ust be fra$ed in such a way that it results results in obtaining the desired desired infor$ation fro$ the respondents( There are no scientific principles that assure an ideal questionnaire and in fact, the questionnaire design is the s*ill s*ill which is learned through e'perience( T8e $ollo&ing steps are involve in t8e )estionnaire esign pro7ess.
6( Spe7i$y Thee firs firstt and and the the fore fore$o $ost st step step in desi designi gning ng the the Spe7i$y t8e In$or%a In$or%ation tion Neee. Neee. Th questio questionnai nnaire re is to specify specify the infor$ infor$ati ation on needed needed fro$ fro$ the respon respondent dentss such such that the objective of the survey is fulfilled( The researcher $ust co$pletely review the co$ponents of the proble$, particularly the hypothesis, research questions, and the infor$ation needed( 8( De$ine t8e Target arget Responent. Responent. t the very outset, the researcher $ust identify the target respondent fro$ who$ the infor$ation is to be collected( The questions $ust be designed *eeping *eeping in $ind $ind the type of respond respondent entss under under study study(( Such as, the question questionss that that are appropriate for service$an $ight not be appropriate for a business$an( The less diversified respondent group shall be selected because the $ore diversified the group is, the $ore difficult it will be to design a single questionnaire that is appropriate for the entire g roup( 34
=( Spe7i$y t8e type o$ Intervie&ing Met8o. The ne't step is to identify the &ay in &8i78 t8e personal al interv interview iews, s, the respon respondent dent is presen presented ted with with a responents responents are rea78e rea78e ( 5n person questionnaire and interacts face#to#face with the interviewer( Thus, lengthy, co$ple' and varied questions can be as*ed using the personal interview $ethod( 5n telephone interviews, the respondent is required to give answers to the questions over the telephone( .ere the respondent cannot see the questionnaire and hence this $ethod restricts the use of s$all, si$ple and precise questions( The questionnaire can be sent through $ail or post( 5t should be self#e'planatory and contain all the i$portant infor$ation such that the respondent is able to understand every question and gives a co$plete response( The electronic questionnaires are sent directly to the $ail ids of the respondents and are required to give answers online( ?( Deter%ine Deter%ine t8e Content Content o$ Inivi)al Inivi)al ;)estions. Once the infor$ation needed is specified and the interviewing $ethods are deter$ined, the ne't step is to decide the content of the question( The researcher $ust decide on what should be included in the question such that it contribute to the infor$ation needed or serve so$e specific purpose( 5n so$e situations, the indirect questions which are not directly related to the infor$ation needed $ay be as*ed( 5t is useful to as* neutral questions at the beginning of a questionnaire with intent to establish respondentHs involve$ent and rapport( This is $ainly done when the subject of a questionnaire is sensitive or controversial( The researcher $ust try to avoid the use of o)!le-!arrele )estions ( question that tal*s about two issues si$ultaneously, such as 5s the Real juice tasty and a refreshing health drin* @( Over7o%e Over7o%e ResponentFs ResponentFs Ina!ility an Un&illingness to Ans&er. The researcher should not presu$e that the respondent can provide accurate responses to all the questions( .e $ust atte$pt to overco$e the respondentHs inability to answer( The questions $ust be designed in a si$ple and easy language such that it is easily understood by each respondent( 5n situations, where the respondent is not at all infor$ed about the topic of interest, then the researcher $ay as* the $ilter )estions , an initial question as*ed in the questionnaire to identify the prospective respondents to ensure that they fulfil the require$ents of the sa$ple( sa$ple( !espite being able to answer the question, the respondent is unwilling to devote ti$e in providing infor$ation( The researcher $ust atte$pt to understand the reason behind such unwillingness and design the questionnaire in such a way that it helps in retaining the respondentHs attention( B( De7ie on t8e ;)estion Str)7t)re. The researcher $ust decide on the structure of questions to be included in the questionnaire( The question can be structured or unstructured( The which h are are answ answer ered ed by the the )nstr)7 )nstr)7t)r t)re e )e )estio stions ns are t8e open-en open-ene e )e )estio stions ns whic respondents respondents in their own words( These questions are also called called as a $ree-response or $reeans&er )estion s( )hile, the str)7t)re )estions are 7alle as 7lose-ene )estions that pre#specify the response alternatives( These questions could be a $ultiple choice question, dichoto$ous 1yes or no2 or a scale( 35
D( Deter%i Deter%ine ne t8e ;)e ;)estio stion n 5ori 5oring. ng. The desired question content and structure $ust be transl translate ated d into into &ors &8i78 are easily )nerstoo by the respondents( t this step, the researcher $ust translate the questions in easy words such that the infor$ation received fro$ the respondents is si$ilar to what was intended( 5n case the question is written poorly, then the respondent $ight refuse to answer it or $ight give give a wron wrong g answ answer er(( 5n case case,, the the resp respon onde dent nt is relu reluct ctan antt to give give answ answer ers, s, then then KnonresponseG arises which increases the co$ple'ity of data analysis( On the other hand, if the wrong infor$at infor$ation ion is given, given, then response errorG arises due to which the result is biassed( :( Deter%ine t8e Orer o$ ;)estions. t this step, the researcher $ust decide the se)en7e in &8i78 t8e )estions are to !e ase ( The opening questions are crucial in establishing respondentHs involve$ent and rapport, and therefore, these questions $ust be interesting, non#threate non#threatening ning and easy( easy( Fsually, Fsually, the open-ene )estions which as* respondents for their opinions are considered as good opening questions, because people li*e to e'press their opinions( E( Ienti$y t8e 0or% an 1ayo)t. The $or%at" positioning an spa7ing of questions has a significant effect on the results( The layout of a questionnaire is specifically i$portant for the self#ad$inistered questionnaires( The questionnaires $ust be divided into several parts, and each part shall be nu$bered accurately to clearly define the branches of a question( 69( Repro)7tion o$ ;)estionnaire. .ere, we tal* about the appearan7e o$ t8e )estionnaire" i(e( the quality of paper on which the questionnaire is either written or printed( 5n case, the questionnaire is reproduced on a poor#quality paper- then the respondent $ight feel the research is uni$portant due to which the quality of response gets adversely affected( Thus, it is reco$$ended to reproduce the questionnaire on a good#quality paper having a professional appearance( 5n case, the questionnaire has several pages, then it should be presented in the for$ of a boo*let rather than the sheets clipped or stapled together( 66( Pretesting. Pretesting $eans testing t8e )estionnaires on a $e& sele7te responents or a s$all s$all sa$ple sa$ple of actual actual respond respondent entss with with a purpos purposee of i$pro i$proving ving the questi questionna onnaire ire by identifying and eli$inating the potential proble$s( ll the aspects of the questionnaire $ust be tested such as question content, structure, wording, sequence, for$ and layout, instructions, and question difficulty( The researcher $ust ensure that the respondents in the pretest should be si$ilar to those who are to be finally surveyed( Attit)e Meas)re%ent M eas)re%ent Te78ni)es. Te78ni)es.
ttitude Scales # Rating Scales to $easure data S7aling Te78ni)es $or Meas)ring Data 6at8ere $ro% Responents
The ter$ scaling is applied to the atte$pts to $easure the attitude objectively( ttitude is a resultant of nu$ber of e'ternal and internal factors( !epending upon the attitude to be $easured, appropriate 36
scales are designed( Scaling is a technique used for $easuring qualitative responses of respondents such as those related to their feelings, perception, li*es, disli*es, interests and prefer
Types o$ S7ales 6( /o$inal Scale 8( Ordinal Scale =( 5nterval Scale ?( Ratio Scale "elf Rating "cales
6( Craphic Rating Scale 8( 5te$i+ed Rating Scales a(
)hat is your gender J Male J %e$ale
37
nother e'a$ple is # a survey of retail stores done on two di$ensions # way of $aintaining stoc*s and daily turnover( .ow do you stoc* ite$s at present J 3y product category J t a centrali+ed store J !epart$ent wise J Single warehouse !aily turnover of consu$er is J 3etween 699 – 899 J 3etween 899 – =99 J bove =99 two way classification can be $ade as follows DailySto7 Pro)7t T)rnover Category Met8o
Depart%ent Centrali'e Single &ise Store 5are8o)se
69 9 – 89 9 89 9 – =9 9 b o ve = 9 9 Mode is frequently used for response category( ' rdinal "c "cale
Ordinal scales are the si$plest attitude $easuring scale used in Mar*eting Research( Research( 5t is $ore powerful than a no$inal scale in that the nu$bers possess the property of ran* order( orde r( The ran*ing of certain product attributesIbenefits as dee$ed i$portant by the respondents is obtained through the scale( E
38
The $ost i$portant attribute is ran*ed 6 by b y the respondents and the least i$portant is ran*ed @( 5nstead of nu$bers, letters or sy$bols too can be used to rate in a ordinal scale( Such scale $a*es no atte$pt to $easure the degree de gree of favourability of different ran*ings( E
Talco$ Powder and if a respondent ran*s the$ E
.erein the distance between the various categories unli*e in /o$inal, or nu$bers unli*e in Ordinal, are equal in case of 5nterval Scales( The 5nterval Scales are also ter$ed as Rating Scales( n 5nterval Scale has an arbitrary ero point with further nu$bers placed at equal intervals( very good e'a$ple e' a$ple of 5nterval Scale is a Ther$o$eter( Ill)stration # .ow do you rate your present refrigerator for the following qualities( Company &ame
@ess 1 /no.n
2
3
4
5
ell /no.n
unctions
0e.
1
2
3
4
5
*an#
Price
@o.
1
2
3
4
5
Figh
Design
Poor
1
2
3
4
5
Bood
5
er# atise d
er# #erall ,is> 1 "atisfaction atis ed
2
3
4
Such a scale per$its the researcher to say that position @ on the scale is above position ? and also the distance fro$ @ to ? is sa$e as distance fro$ ? to =( Such a scale however does not per$it conclusion that position ? is twice as strong as position 8 because no +ero position has been established( The data obtained fro$ the 5nterval Scale can be used to calculate the Mean scores scores of each each attrib attribute utess over all respon respondent dents( s( The Standa Standard rd !eviati !eviation on 1a $easur $easuree of dispersion2 can also be calculated( Ratio "cale
39
Ratio Scales are not widely used in Mar*eting Research unless Research unless a base ite$ is $ade available for co$parison( 5n the above e'a$ple of 5nterval scale, a score of ? in one quality does not necessarily $ean that the respondent is twice $ore satisfied than the respondent who $ar*s 8 on the scale( Ratio scale has a natural +ero point and further nu$bers are placed at equally appearing intervals( %or e'a$ple scales for $easuring physical quantities li*e # length, weight, etc( The ratio scales are very co$$on in physical scenarios( uantified responses for$ing a ratio scale analytically are the $ost versatile( Rati scale possess all he characteristics of an internal scale, and the ratios of the nu$bers on these scales have $eaningful interpretations( !ata on certai certain n de$ogr de$ograph aphic ic or descri descripti ptive ve attri attribute butes, s, if they are obtaine obtained d through through open#en open#ended ded questions, will have ratio#scale properties( 0onsider the following questions ; 62 )hat is your annual inco$e before ta'es UUUUUU V 82 .ow far is the Theater fro$ your ho$e UUUUUU $iles nswers to these questions have a natural, una$biguous starting point, na$ely +ero( Since starting point is not chosen arbitrarily, co$puting and interpreting ratio $a*es sense( %or e'a$ple we can say that a respondent with an annual inco$e of V ?9,999 earns twice as $uch as one with an annual inco$e of V 89,999(
el" )ating cales '+ ,raphi ,raphic c Ratin Rating g "cale "cale
The respondents rate the objects by b y placing a $ar* at the appropriate position on a line that runs fro$ one e'tre$e of the criterion variable to another( &'a$ple
! (poor ;ualit#
1 (%ad ;ualit#
5 (neither good nor %ad
7 (good ;ualit#
BRAND
This is also *nown as continuous rating scale( The custo$er can occupy any position( .ere one attribute is ta*en e'#quality of any an y brand of icecrea$(
poor
good
BRAND /
This line can be vertical or hori+ontal h ori+ontal and scale points $ay be provided( /o other indication is there on the continuous scale( range is provided( To quantify the responses to question 4!
that Kindicate your overall opinion about ice#rea$ 3rand 8 by placing a tic* $ar* at appropriate position on the lineL, we $easure the physical distance between the left e'tre$e e' tre$e position and the response position on the line(- the greater the distance, the $ore favourable is the response or attitude towards the brand( 5ts li$itation is that coding and analysis will require substantial a$ount of ti$e, since we first have to $easure the physical distances on the scale for each respondent( + *temi$ *temi$ed ed Ratin Rating g "cales "cales
These scales are different fro$ continuous rating scales( They have a nu$ber of brief descriptions associated with each category( They are widely used in Mar*eting Research( Research( They essentially ta*e the for$ of the $ultiple category categor y questions( The $ost co$$on are #
5t was developed Rensis
4 &gree
3
2 1 ,isagree trongl# disagree
&ach state$ent is assigned a nu$erical score ranging fro$ 6 to @( 5t can also be scaled as #8 to Q8( >2
>1
!
1
2
%or e'a$ple quality of Mother !iary ice#crea$ is poor p oor then /ot Cood is a negative state$ent and Strongly gree with this $eans the quality is not good( &ach degree of agree$ent is given a nu$erical score and the respondents total score is co$puted by su$$ing these scores( This total score of respondent reveals the particular opinion of a person(
Thus, li*ert scale is a five point scale ranging fro$ Hstrongly agree$entHto Hstrongly disagree$entH( /o judging gap is involved in this $ethod( !+ "emantic "emantic Di/erent Di/erential ial "cale "cale
This is a seven point scale and the end points of the scale are a re associated with bipolar labels( 1 Cnpleasan t u%missiv e
2
3
4
5
6
7 Pleasant ,ominant
Suppose we want to *now *no w personality of a particular person( )e have options# 6( Fnpl Fnplea easa sant ntIS ISub ub$i $iss ssiv ivee 8( Plea Pleasa sant ntI! I!o$ o$in inan antt
3i#polar $eans two opposite strea$s( 5ndividual can score between 6 to D or #= to =( On the basis of these responses profiles are $ade( )e can analyse for two or three products and by joining these profiles we get profile analysis( 5t could ta*e any shape depending on the nu$ber of variables( Profile nalysis ###############I############### ##########I#################### ########I###################### and median are used for co$parison( This scale helps to deter$ine overall Mean and si$ilarities and differences a$ong objects( )hen Se$antic !ifferenti !ifferential al Scale is used to develop an i$age profile, profile, it provides provides a good basis for co$paring i$ages of two or $ore ite$s( The big advantage of this scale is its si$plicity, while producing results co$pared with those of the $ore co$ple' scaling $ethods( The $ethod is easy and fast to ad$inister, but it is also sensitive to s$all differences in attitude, highly versatile, reliable and generally valid( c+ "tap "tapel el0s 0s "cal "cale e
5t was developed by >an Stapel( This scale has so$e distinctive features;# i
i ach ach item item has has onl# onl# one one .ord .ord'p 'phra hrase se indi indica cati ting ng the dimension it represents 42
ii ii
ach ach ite item m has has ten ten res respo pons nse e cate catego gori ries es
iii iii
ach ach item item has has an an eve even n num num%er %er o" cate catego gori ries es
iv iv
:he respo response nse cate categor gories ies have have numeric numerical al la%e la%els ls %ut %ut no ver%a ver%all la%el la%els s
%or e'a$ple, in the following ite$s, suppose for quality of ice crea$, we as* respondents to ran* fro$ Q@ to #@( Select a plus nu$ber for words which best b est describe the ice crea$ accurately( Select a $inus nu$ber for words you thin* do not describe the ice crea$ quality accurately( accu rately( Thus, we can select any nu$ber nu $ber fro$ Q@,for words we thin* are very accurate, accu rate, to #@,for words we thin* are very ver y inaccurate( This scale is usually presented vertically( Q@ Q? Q= Q8 Q6 .igh uality #6 #8 #= #? #@ This is a unipolar rating scale( /
M)lt M)ltii Di%e Di%ens nsio iona nall S7a S7aling ling
5t consists of a group of analytical techniques which are used to study consu$er attitudes related to perceptions and preferences( 5t is used to study# 5(
55(
The The $ajor $ajor attr attrib ibut utes es of a give given n class class of prod produc ucts ts perce perceiv ived edby by the the cons consu$ u$er erss in considering the product and by which they co$pare the different ran*s( To study study whic which h brand brand co$pet co$petes es $ost $ost dire directly ctly with with each each othe other( r(
555( 555(
To find find out out wheth whether er the the consu consu$e $ers rs woul would d li*e li*e a new new brand brand with with a co$bi co$bina nati tion on of characteristics not found in the $ar*et(
5"( 5"(
)hat )hat would would be the the consu$ consu$ers ers ideal ideal co$b co$bina inatio tion n of produc productt attrib attribute utes( s(
"(
)hat )hat sale saless and and adv adverti ertisi sing ng $ess $essag ages es are are co$ co$pati patibl blee with with cons consu$ u$er erss bran brand d perceptions(
5t is a co$puter based technique( The respondents are as*ed to place the various brands into different groups li*e si$ilar, very si$ilar, not si$ilar, and so on( 43
goodness of fit is traded off on a large nu$ber of attributes( Then a lac* of fit inde' is calculated by co$puter progra$( The purpose is to find a reasonably s$all nu$ber of di$ensions which will eli$inate $ost of the stress( fter the configuration for the consu consu$e $erH rHss pref prefer erenc encee has has been been deve develo lope ped, d, the the ne't ne't step step is to dete deter$ r$in inee the the preference with regards to the product under study( These techniques atte$pt to identify the product attributes that are i$portant to consu$ers and to $easure their relative i$portance( This This scalin scaling g involv involves es a unreal unrealist istic ic assu$p assu$ptio tion n that that a consu$ consu$er er who co$par co$pares es different brands would perceive the differences on the basis of only one attribute( %or e'a$ple, what are the attributes for joining M(0o$ course( The responses $ay be #to do PC, to go into teaching line,to get *nowledge, appearing in the /&T( There are a nu$ber of attributes, you can not base decision on one attribute only( Therefore, when the consu$ers are choosing between brands, they base their decision on various attributes( 5n practice, the perceptions of the consu$ers involve different attributes and any one consu$er perceives perceives each brand as a co$posite of a nu$ber of different different attributes( This is a shortco$ing of this scale( )heneve )heneverr we choose choose fro$ fro$ a nu$ nu$ber ber of altern alternati atives ves,, go for $ulti# $ulti# di$ensi di$ensional onal scaling( There are $any possible uses of such scaling li*e in $ar*et seg$entation, product life cycle, vendor evaluations and advertising $edia selection( The li$itation of this scale is that it is difficult to clearly define the concept of si$ilarities and preferences( %urther the distances between the ite$s are seen as different Thur Thurs sto ton n "c "cale ales
These are also *nown as equal appearing interval scales( They are used to $easure the attitude towards a given concept or construct( %or this purpose a large nu$ber of state$ents are collected that relate to the concept or construct being $easured( The judges rate these state$ents along an 66 category scale in which each category e'presses a different degree of favourableness towards the concept( The ite$s are then ran*ed according to the $ean or $edian ratings assigned by the judges and are used to construct questionnaire of twenty to thirty ite$s that are chosen $ore or less evenly across the range of ratings( The state$ents are worded in such a way so that a person can agree or disagree with the$( The scale is then ad$inistered to asse$ble of respondents whose scores are deter$ined by co$puting the $ean or $edian value of the ite$s agreed with( person who disagrees with all a ll the ite$s has a score of +ero( So, the advantage of this scale is that it is an interval $easure$ent scale( 3ut it is the ti$e consu$ing $ethod and labour intensive( They are co$$only used in psychology and education research(
44
1 ,uttma ,uttman n "cal "cales2 es2"c "calo alogra gram m 3naly 3nalysis sis
5t is based on the idea that ite$s can be arranged along a continue$ in such a way that a person who agrees with an ite$ or finds an ite$ acceptable will also agree with or find acceptable all other ite$s e'pressing a less e'tre$e position( %or e'a$ple # 0hildren should not be allowed to watch indecent progra$$es or govern$ent should ban these progra$$es or they are not allowed to air on the television( They all are related to one aspect( 5n this scale each score represents a unique u nique set of responses and therefore the total score of every individual is obtained( This scale ta*es a lot of ti$e and effort in develop$ent( They are very co$$only used in political science, anthropology, public opinion, research and psychology( 4 5 The The % "o "ort rt tech techni ni6u 6ue e
5t is used to discri$inate a$ong large nu$ber of objects quic*ly( 5t uses a ran* order procedure and the objects are sorted into piles based on si$ilarity with respect to so$e criteria( The nu$ber of objects to be sorted should be between B9#6?9 appro'i$ately( %or e'a$ple, here we are ta*ing nine brands( On the basis of taste we classify the brands into tasty, tasty, $oderate and non tasty( )e can classify on the basis of price also#
1a2 /on#disguised Structured Techniques( 1b2 /on#disguised, /on#structured Techniques( 1c2 !isguised /on#structured Techniques( Techniques(
45
?a@ Non-isg)ise Str)7t)re Te78ni)es.
This approach e$ploys a standardi+ed questionnaire to collect data on beliefs, feelings, and attitude fro$ the respondent( Single ;)estion Met8o.
15 thin* it is a good product or 5 thin* it is a poor product2( M)ltiple ;)estions Met8o.
1/u$bers of questionnaires as*ed about abo ut the attitude2 and Physiological Tests Tests 1laboratory tests such as galvanic s*in response, eye $ove$ent etc( e tc( $easure attitudes of people towards products2 are carried out under this approval( ?!@ Non-isg)ise" Non-str)7t)re Te78ni)es.
These techniques use a non standardi+ed questionnaire( The techniques are also called depth interview, qualitative qualitative interviews, unstructured interviews, or focussed interviews( ll these techniques are designed to gather infor$ation on various aspects of hu$an behaviour including the KwhyL co$ponent(
?7@ Disg)ise" Non-str)7t)re Te78ni)es.
5n this approach, the purpose of study is not discussed to respondents unli*e above two cases( list of unstructured questions is used to collect data on consu$erHs attitudes( This art of using disguised and unstructured $ethod is referred to as KProjective K Projective TechniquesL( TechniquesL( The projective techniques include several tests given to the respondents( They $ay be as*ed to give their co$$ents on cartoons, pictures, stories etc( The sti$uli used for this purpose are capab le of answering the respondent to a variety of reactions( nu$ber of Projective Techniques, are available to the $ar*et researchers for the purpose of analysing KwhyL part of consu$er behaviour( Qualitative Techniques:
1Projective Techniques and )ord )ord ssociation as follows2( T8e %ain Pro:e7tive Te78ni)es are. # 5or Asso7iation Test ?5#A#I@.
The interviewer calls a series of listed words one by one o ne and the respondents quic*ly replies the first word that enters his $ind( The underlying assu$ption is that by Kfree associatingL with certain
46
sti$uli 1words2 the responses are ti$ed so that those answers which the respondent Kresponse ou tL are identified( /# Senten7e Co%pletion.
Sentence co$pletion test is si$ilar to word association test e'cept that the respondent is required to co$plete an unfinished sentence( %or e'a$ple, K5 do not use sha$poos because((L K0offee that is quic*ly $ade( L +# Story Co%pletion.
5n this technique the respondent is as*ed to co$plete a story, end of which is $issing( This enables a researcher to find out the al$ost e'act version of o f i$ages and feelings of people towards a co$panyHs product( This helps in finalising finalising the advertising and pro$otional the$es for the product in question( *# Resear78 o$ In-!lot Tests ?or Resear78 Tests@.
Motivation Research e$ploys this fa$ous test( These tests are not in $uch use in $ar*eting research( The research test e'presses in a classic way the rationale be hind all projective tests, that is, in filling the $issing parts of a vague and inco$plete sti$ulus, the respondent projects hi$self and his personality into the picture( lot of in* is put on the piece of paper and reference is $ade of co$pany, co $pany, product, and the respondent is as*ed to give his view points after interpreting what he sees in the blot before hi$( The respondent say, Kugly pac*aging of the productL, or Ke'cellent perfor$ance p erfor$ance of the productL( This response will help the seller to finalise his $ar*eting strategies( 4# Psy78ograp8i7 Te78ni)e.
This includes galvanic s*in response, eye $ove$ent and eye blin* test etc( which uses various 5nstru$ents with the physiological responses( =# Espionage Espio nage Te78ni)e. Te78ni)e. T8ere are t&o %et8os in t8is te78ni)e. ?i@ Use o$ Hien Re7orers.
Such as hidden tape recorders, ca$eras used to watch consu$ers as they $a*e $a *e purchases or consu$e ite$s( ?ii@ R)!!is8 Resear78. 47
This is another $ethod of espionage activity( .ere, the researcher shifts through the garbage of individuals or groups and record pattern of consu$ption, waste, and brand preference( 5t gives $ost required esti$ates of consu$ption of cigarettes, $edicines, liquor, and $aga+ines etc( ?@ Disg)ise Str)7t)re Te78ni)es.
)hen we are to $easure those attitudes which respondents $ight not readily and accurately e'press, we can use disguised structured techniques( The disguised structured questionnaire are easy to ad$inister and code( Respondents are given questions which they are not li*ely to be able to answer accurately( 5n such circu$stances they are co$pelled to Wguess atH the answers( The respondentHs attitude on the subject is assu$ed to be revealed to the e'tent and direction in which these guessing errors are co$$itted( Uses of Motivation Research:
6( Motivation Research leads to useful insights and provides inspiration to creative person in the advertising and pac*ing world( 8( 7nowledge and $easure$ent of the true attitude of custo$ers help in choosing the best selling appeal for the product and the best way to represent the product in the sales tal*, and in deter$ining the appropriateness and weight age of various pro$otional $ethods( =( Motivation Research can help in $easuring changes in attitudes, thus advertising research( ?( 7nowledge and $easure$ent of attitudes provides us with an i$aginative $ar*et seg$entation tool and also enables esti$ating $ar*et potential of each additional seg$ent( @( Strategies to position the offer of the co$pany in a particular $ar*et seg$ent should be based on the findings of $otivation research( Limitations of Motivation Research:
6( 0autions are required to be e'ercised not only in the application of these techniques but also the resultant data should be analysed and interpreted according to the psychological theory( 8( Originally these techniques were developed to collect data fro$ a single individual over a period of ti$e( 5t is not free fro$ draw bac*s while we apply these techniques to gather data fro$ a nu$ber of individuals( =( The designing and ad$inistering of these techniques need qualified and e'peri$ented researchers( Such personnel are not easily available(
Meaning o$ statisti7al te78ni)es. Mathe$atical concepts, for$ulas, $odels, te78ni)es used in statisti7al analysis of rando$ data( 5n co$parison, co$parison, deter$inisti deter$inisticc %et8os are used where the data is easily reproducible or where its behavior is deter$ined entirely by its initial stage and inputs( 48
Di$$erent Di$$erent types o$ statisti7al %et8os. So%e &ell-no&n statisti7al tests an pro7e)res are. •
nalysis of variance 1/O"2
•
0hi#squared test(
•
0orrelation(
•
%actor analysis(
•
Mann–)hitney F(
•
Mean square weighted deviation 1MS)!2
•
Pearson product#$o$ent correlation coefficient(
•
Regression analysis
58at are t8e statisti7al tools9 •
1)hen the distribution is s*ewed statisti7al treat$ent is $ore co$plicated2( The pri$ary para$eters used are the $ean 1or average2 and the standard deviation and the $ain tools the %#test, the t#test, and regression and correlation analysis
A ;)i7 6)ie $or C8oosing t8e Appropriate Statisti7al Test The $ost i$portant step in choosing the appropriate statistical procedure is to *now what the variables of your study are( )hat are the independent independ ent and dependent variables of your study .ow are each of the variables $easured Once you have a better grasp of your variables, you can easily choose the statistical procedure that will best answer your studys questions(
&n xample %ive students are as*ed to design a study that will assess the relationship between using the )ii %it and weight loss in a group of o f 6@9 overweight pre#teens during a $onth#long period( The weight of the participants is ta*en at the beginning and at the end of the study( The The students co$e up with five different proposals; 6( Student Student proposes proposes to rando$ly rando$ly assign half half of the group to the )ii )ii intervention( intervention( This This group will be instructed to do )ii %it erobics training for =9 $inutes five ti$es a wee*( The other half would serve as the control control and will not be told to do anything( anything( Student then then wants to deter$ine whether using the )ii %it would lead to weight loss( ccordingly, ccordingly, she suggests that the participants be classified into one of two groups; no n o weight loss and weight loss(
49
8( Studen Studentt 3 also also proposes proposes to rando$ly rando$ly assign assign half of the group to the )ii )ii intervent intervention ion(( This group will be instructed to do )ii )ii %it erobics training for =9 $inutes five ti$es a wee*( The other half would serve as the control and will not be told to do anything( Student 3 then wants to deter$ine whether using the )ii %it would lead to weight loss( Thus, he suggests that that weig weight ht loss loss shoul should d be defi defined ned in ter$ ter$ss of the the diff differ eren ence ce in weig weight ht prio priorr to and and i$$ediately after the study( =( Student Student 0 proposes to rando$ly rando$ly assign assign =9 pre#teens pre#teens to one of four types types of )ii )ii intervention interventions; s; erobics, Noga, Strength Training, and 3alance Ca$es( These four groups will be instructed to do the specific )ii %it activity for =9 $inutes five ti$es a wee*( The last group of =9 pre# teens would serve as the control and will not be told to do anything( Student 0 then wants to deter$ine whether using the )ii %it would lead to weight loss( Thus, she suggests that weight loss should be defined in ter$s of the difference in weight prior to and i$$ediately after the study( ?( Student Student ! proposes proposes to si$ply si$ply as* the the 6@9 overweight overweight teens teens to record record the nu$ber nu$ber of $inutes $inutes per day they spend using the )ii )ii %it( .e then suggests that the participants be classified into one of two groups; no weight loss and weight loss( @( Studen Studentt & also also proposes proposes to si$ply si$ply as* the 6@9 overwei overweight ght teens teens to record record the nu$ber nu$ber of $inutes per day they spend using the )ii %it( .e then suggests that weight loss should be defined in ter$s of the difference in weight prior to and i$$ediately after the study( 0irst Point. Varia!les o$ t8e St)y )hat are the variables of the study Fsing the )ii %it would be the independent variable of the study while weight loss would be the dependent variable of the study( Se7on Point. De$inition or Meas)re%ent o$ t8e Varia!les %ro$ %ro$ the the e'a$ e'a$pl plee above above it is obvi obviou ouss that that ther theree are are seve severa rall ways ways to defin definee or $eas $easur uree the the independent and dependent dependen t variables of a study( 3ut there are two $ain questions to consider;
6( 5s the independ independent ent variable variable $easure $easured d categorical categorically ly or continu continuously ously 8( 5s the dependen dependentt variable variable $easured $easured categoricall categorically y or continuou continuously sly St)ent A# Student defined )ii %it use in ter$s of using the )ii %it or not using the )ii %it( St)ent ccordingly, Student defined )ii %it use in ter$s of categories( She defined weight loss in ter$s of no weight loss or weight loss( Thus, her definition of weight loss was ca tegorical( St)ent B( Student 3 also defined )ii %it use in ter$s of using the )ii %it or not using the )ii %it( Thus, he defined )ii %it use in ter$s of categories( 3ut Student 3 defined weight loss in ter$s of the difference between weight prior to the study and weight i$$ediately after the study( )eight loss, therefore, was defined continuously( St)ent C( Student 0 defined )ii %it use in ter$s of the type of )ii %it activity( Thus, she defined )ii %it use in ter$s of five categories( She also defined weight loss in ter$s of the difference
5!
between weight prior to the study and weight i$$ediately after the study( )eight loss, therefore, was defined continuously( St)ent D( Student ! defined )ii %it use in ter$s of the nu$ber of $inutes per day spent using the )ii %it( s s such, )ii %it use was defined continuously( .e defined weight loss in ter$s of no weight loss or weight loss( Thus, his definition of weight loss was categorical( St)ent E( Student & defined )ii %it use in ter$s of the nu$ber of $inutes per day spent using the )ii %it( s such, )ii %it use was defined continuously( Student & also defined weight loss in ter$s of the difference between weight prior to the study and weight i$$ediately after the study( )eight loss, therefore, was defined continuously( c ontinuously( T8ir Point. C8oosing t8e Appropriate Statisti7al Pro7e)re Civen that independent and dependent dependent variables variables can be classified classified as categorical categorical or continuous, continuous, the grid below can be used to classify the $ore co$$on statistical procedures(
Inepenent Varia!le Varia!le
St)ent A( Student could thus choose to perfor$ either a cross#tabulation analysis or a logistic regression procedure( These tests are useful when the independent and dependent variables are $easured categorically( St)ent B( Student 3 would need to conduct an independent t#test procedure since his independent variable would be defined in ter$s of categories and his dependent variable would be $easured continuously( n independent t#test procedure is used only when the independent variable has two categories( St)ent C( Student 0 would need to conduct a one#way /O" since her independent variable woul would d be defi define ned d in ter$ ter$ss of categ categor orie iess and her her depen depende dent nt vari variab able le woul would d be $eas $easure ured d contin continuou uously sly(( One#way One#way /O" /O" s are used used when when the indepe independen ndentt variabl variablee has three three or $ore $ore categories( 51
St)ent D( Student ! would use a logistic regression procedure to analy+e his data since his indepen independen dentt variabl variablee would would be $easur $easured ed contin continuous uously ly and his depende dependent nt variabl variablee would would be $easured categorically( 5f Student ! defined his dependent variable in ter$s of three or $ore categories that could be ran*ed 1e(g(, weight gain, no weight loss, weight loss2, then he would use an ordinal regression procedure( St)ent E( Student & could choose to perfor$ either a Pearson correlation procedure or a linear regres regressio sion n procedu procedure re since since both both of her variab variables les would would be define defined d continu continuous ously ly(( Fsuall Fsually y, a correlation test is conducted when there is only one independent variable and one independent variable( 5f Student & wanted to study the relationship between several independent variables 1e(g(, nu$ber of hours spent sleeping, nu$ber of calories consu$ed per day2 and weight loss, then she would use a linear regression procedure(
KKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKK
Unit IV STATISTICA1 METHODS
Tabulation of data # nalysis of data –Testing of .ypothesis, dvanced techniques – /O", 0hi#Square # !iscri$inant nalysis # %actor analysis, 0onjoint analysis # Multidi$ensional Scaling # 0luster nalysis 10oncepts Only2( Statisti7al Met8os ; Methods of collecting, su$$ari+ing, analy+ing, and interpreting variable nu$erical data( Statistical $ethods can be contrasted with deter$inistic deter$inistic $ethods, $ethods, which are appropriate where observations are e'actly reproducable or are assu$ed to be so( )hile statistical $ethods are widely used in the life sciences, in econo$ics, and in agricultural science, they also have an i$portant role in the physical science sciencess in the study study of $easur $easure$e e$ent nt errors errors,, of rando$ rando$ pheno$e pheno$ena na such such as radioa radioacti ctivit vity y or $eteorological events, and in obtaining appro'i$ate results where deter$inistic solutions are hard to apply( 1ata collection involves deciding what to observe in order to obtain infor$ation relevant to the questions whose answers are required, and then $a*ing the observations( Sa$pling Sa$pling involves involves choice of a sufficient nu$ber of observations representing an appropriate population( &'peri$ents with vari variab able le outc outco$ o$es es shou should ld be cond conduc ucte ted d acco accord rdin ing g to prin princi cipl ples es of e'peri e'peri$ental $ental design design((
calculati ation on of appropri appropriate ate statistics statistics 1def( 82 and the display of such 1ata summari0ation is the calcul infor$ation in the for$ of tables, graphs, or charts( !ata $ay also be adjusted to $a*e different sa$ples $ore co$parable, using ratios, co$pensating factors, etc( relatess obs observ erved ed statis statistic tical al data to theoret theoretica icall $od $odels els,, such such as probability Statistic Statistical al analysis analysis relate 52
distributions or $odels used in regression analysis( distributions or analysis( 3y esti$ating para$eters esti$ating para$eters in the proposed $odel and testing hypotheses about rival $odels, one can assess the value of the infor$ation collected and the e'tent to which the infor$ation infor$ation can be applied to si$ilar si$ilar situations( situations( Statistical prediction is the appl applic icat atio ion n of the the $o $odel del thoug thought ht to be $o $ost st appr approp opri riat ate, e, usin using g the the esti esti$a $ate ted d valu values es of the the para$eters(
Ta!)lation Ta !)lation o$ Data an Types o$ Ta!)lation. Ta!)lation. Tabulati Tabulation on is the syste$atic syste$atic arrange$ent arrange$ent of the statistical statistical data in colu$ns colu$ns or rows( 5t involves the orderly and syste$atic presentation of nu$erical data in a for$ designed to e'plain the proble$ under consideration( Tabulation helps in drawing the inference fro$ the statistical figures( Tabulation prepares the ground for analysis and interpretation( Therefore a suitable $ethod $ust be decided carefully ta*ing into account the scope and objects of the investigation, because it is very i$portant part of the statistical $ethods( Types o$ Ta!)lation
5n general, the tabulation is classified in two parts, that is a si$ple tabulation, and a co$ple' tabulation( Si$ple Si$ple tabula tabulatio tion, n, gives gives infor$ infor$ati ation on regard regarding ing one or $ore $ore indepen independent dent questi questions ons(( 0o$ple 0o$ple' ' tabulation gives infor$ation regarding two $utually dependent questions( O/&#)N T3<& !5"5S5O/ POPF<T5O/ 1Millions2 7arachi 69(:D@EB: .yderabad 6?(6:BE@? Su**ur 68(EE??96 This table gives us infor$ation regarding one characteristic infor$ation about the population in different divisions of Sindh( ll questions that can be answered in O/& )N T3<& are independent of each other( 5t is therefore an e'a$ple of a si$ple tabulation, since the infor$ation obtained in it is regarding one independent question, that is the nu$ber of persons in various divisions of Sindh in $illions( T&o-5ay Ta!le
These types of table give infor$ation regarding two $utually dependent questions( %or e'a$ple, question is, how $any $an y $illions of the persons are in the !ivisions- the One#)ay Table Table will give the answer( 3ut if we want to *now that in the population nu$ber, who are in the $ajority, $ale, $ale, or fe$ale( The Two#)a Two#)ay y Tables Tables will answer the question by giving the colu$n for fe$ale and $ale( Thus the table showing the real picture of divisions se' wise is as under; T)O#)N T3<& !5"5S5O/ POPF<T5O/ 1M 1Millions2 Male %e$ale Total 7arachi 53
.yderabad Su**ur T8ree-5ay Ta!le
Three#)ay Three#)ay Table gives infor$ation regarding three $utually dependent and inter#related questions( %or e'a$ple, fro$ one#way table, we get infor$ation about population, and fro$ two#way table, we get infor$ation about the nu$ber of $ale and fe$ale available in various divisions( /ow we can e'tend the sa$e table to a three way table, by putting a question, K.ow $any $ale and fe$ale are literateL Thus the collected statistical data will show the following, three $utually dependent and inter#related questions; 6( Population in various division( 8( Their se'#wise distribution( =( Their position of literacy(
!5"5S5O/
Male 5lliterate
Total
T.R&&# T.R &&#) )N N T3<& POPF<T5O/ 1Millions2 %e$ale
Total 5lliterate
Total
7arachi .yderabad Su**ur This table gives infor$ation concerning the literacy of both $ale and fe$ale in various divisions of Sindh( %ro$ the table we can e'plain the se' which has $ore education in relation to division, and also, we can say whether literacy is low in rural areas than in urban areas( Hig8er Orer Ta!les
.igher order tables are those which provide infor$ation about a large nu$ber of inter related questions( .igher order tables $ay be of four#way, four#way, five#way, si'#way etc( Such *inds of tables are called $anifold tables( Avantages o$ Ta!)lation Ta!)lation
6 The large $ass of confusing data is easily reduced to reasonable for$, that is understandable to *ind( 8(The data once arranged in a suitable for$, gives the condition of the situation at a glance, or gives a bird eye view( = %ro$ the table it is easy to draw so$e reasonable conclusion or inferences( ?( Tables Tables gave grounds for analysis of the data( @( &rrors, and o$ission if if any are always detected in tabulation( Therefore the i$portance of a carefully drawn table is vital for the preparation of data for analysis and interpretation(
&&@G< 0 ,&:&H The process of evaluating data using analytical and logical reasoning to e'a$ine each co$ponent of the data provided( This for$ of analysis is just one of the $any steps that $ust be co$pleted when conduct conducting ing a resear research ch e'peri$ e'peri$ent ent(( !ata !ata fro$ fro$ variou variouss sou source rcess is gathere gathered, d, review reviewed, ed, and then then analy+ed to for$ so$e sort of finding or conclusion( There are a variety of specific data analysis 54
$ethod, $ethod, so$ so$ee of which which includ includee data data $inin $ining, g, te't analyti analytics, cs, bus busine iness ss intell intellige igence, nce, and data data visuali+ations( nalysing data to su$$arise it and loo* for patterns is an i$portant part of every evaluation( The options for doing this have been grouped into two categories # quantitative data 1nu$ber2 and qualitative data 1te't, i$ages2( Options umeric analysis
nalysing nu$eric data such as cost, frequency, physical characteristics( •
•
•
•
•
•
•
•
0orrelation; a statistical $easure ranging fro$ Q6(9 to #6(9 that indicates how strongly two or 0orrelation; $ore variables are related( positive correlation 1Q6(9 to 92 indicates that two variables will either increase or decrease together, while a negative correlation 19 to #6(92 indicates that as one variable increases, the other will decrease( 0rosstabulations; usin 0rosstabulations; using g cont contin ingen gency cy tabl tables es of two two or $o $ore re di$e di$ens nsio ions ns to indi indica cate te the the relati relations onship hip betwee between n no$ no$ina inall 1categor 1categorica ical2 l2 variab variables les(( 5n a si$ple si$ple crosst crosstabul abulati ation, on, one variable occupies the hori+ontal a'is and another the vertical( The frequencies of each are added in the intersecting squares and displayed as percentages of the whole, illustrating relationships in the data( !a ta $in !ata $inin ing g; co$puter co$puter#dri #driven ven auto$ated auto$ated techniques techniques that run through through large large a$ounts a$ounts of te't or data to find new patterns and infor$ation( &'ploratory Te &'ploratory Technique chniques; s; ta*i ta*ing ng a Wfir Wfirst st loo* loo*HH at a data datase sett by su$$ su$$ar aris isin ing g its its $ain $ain characteristics, often by using visual $ethods( %requency tables; tables; a visual way of su$$ari+ing no$inal and ordinal data by displaying the count of observations 1ti$es a value of a variable occurred2 in a table( Measures of central tendency; a tendency; a su$$ary $easure that atte$pts to describe a whole set of data with a single value that represents the $iddle or centre of its distribution( The $ean 1the average average value2 value2,, $edian $edian 1the 1the $iddle $iddle value2 value2 and $od $odee 1the 1the $os $ostt frequen frequentt value2 value2 are all $easures of central tendency( tendenc y( &ach $easure is useful for different conditions( Measures of dispersion; a dispersion; a su$$ary $easure that provides infor$ation about how $uch variation there is in the data, including the range, inter#quartile range and the standard deviation( Multivariate descriptive; descriptive; providing si$ple su$$aries of 1large a$ounts of2 infor$ation 1or data2 with two or $ore related variables( o
Multiple regression
o
%actor analysis
55
•
•
•
•
o
0luster analysis
o
Structural equation $odelling
/on#Para$etric inferential statistics statistics;; $ethods for inferring conclusions about a population fro$ a sa$ples data that are fle'ible and do not follow a nor$al distribution 1ie, the distribution does not parallel a bell curve2, including ran*ing; the chi#square test, bino$ial test and Spear$ans ran* correlation coefficient( Para$etric inferential statistics; statistics ; $ethods for inferring conclusions about a population fro$ a sa$ples data that follows certain para$eters; the data will be nor$al 1ie, the distribution parallels the bell curve2- nu$bers can be added, subtracted, $ultiplied and divided- variances are equal when co$paring two or $ore groups- and the sa$ple should be large and rando$ly selected( Su$$ary stat statistic isticss; provid providing ing a quic* quic* su$ su$$ary $ary of data data which which is partic particular ularly ly useful useful for co$paring one project to another, before and after( Ti$e se Ti$e seri ries es an analy alysi siss; obse observ rvin ing g well# well#def defin ined ed data data ite$ ite$ss obtai obtaine ned d thro throug ugh h repe repeat ated ed $easure$ents over ti$e(
Textual analysis
naly+ing naly+ing words, words, either spo*en or written, written, including questionnai questionnaire re responses, responses, interviews, interviews, and docu$ents( •
•
•
•
0ontent analysis; analysis; reducing large a$ounts of unstructured te'tual content into $anageable data relevant to the 1evaluation2 research questions( qu estions( The$atic coding; coding; recording or identifying passages of te't or i$ages that are lin*ed by a co$$on the$e or idea allowing the inde'ation of te't into categories( %ra$ewor* $atrices; a $atrices; a $ethod $ethod for su$$ari su$$arising sing and analysing analysing qualitati qualitative ve data in a two#by# two#by# two $atri' table( 5t allows for sorting data across case and by the$e( Ti$elines and ti$e# Ti$elines ti$e#order ordered ed $atri $atrices; ces; aids aids analys analysis is by allowi allowing ng for visual visualisa isatio tion n of *ey events, sequences and results(
&&@G< 0 ,&:& A De$inition o$ Data Analysis
The process of evaluating data using analytical and logical reasoning to e'a$ine each co$ponent of the data provided( This for$ of analysis is just one of the $any steps that $ust be co$pleted when conduct conducting ing a resear research ch e'peri$ e'peri$ent ent(( !ata !ata fro$ fro$ variou variouss sou source rcess is gathere gathered, d, review reviewed, ed, and then then analy+ed to for$ so$e sort of finding or conclusion( There are a variety of specific data analysis $ethod, $ethod, so$ so$ee of which which includ includee data data $inin $ining, g, te't analyti analytics, cs, bus busine iness ss intell intellige igence, nce, and data data visuali+ations( 56
!ata analysis is a pri$ary co$ponent of data $ining and 3usiness 5ntelligence 1352 and is *ey to gaining the insight that drives business decisions( Organi+ations and enterprises analy+e data fro$ a $ultitude of sources using 3ig !ata $anage$ent solutions and custo$er e'perience $anage$ent solutions that utili+e data analysis to transfor$ data into actionable insights( !ata analysis for quantitative studies, on the other hand, involves critical analysis and interpretation of figures and nu$bers, and atte$pts to find rationale behind the e$ergence of $ain findings( 0o$parisons of pri$ary research findings to the findings of the literature review are critically i$portant for both types of studies – qualitative and quantitative( qu antitative( !ata analysis $ethods in the absence of pri$ary data collection can involve discussing co$$on patterns, as well as, controversies within within secondary data directly related to the research area( Data Analysis Moel
There are seven steps that re$ain consistent across organi+ations and their data analysis processes; •
•
•
•
•
•
De7ie De7ie on t8e o!:e7tiv o!:e7tives es – !eter$ine objectives for data science tea$s to develop a quantifiable way to deter$ine whether the business is progressing toward its goals- identify $etrics or perfor$ance indicators early Ienti$y !)siness levers – 5dentify goals, $etrics, and levers early in data analysis projects to give scope and focus to data analysis- this $eans the business should be willing to $a*e changes to i$prove its *ey $etrics and reach its goals as well Data 7olle7tion – Cather as $uch data fro$ diverse sources as possible in order to build better $odels and gain $ore actionable insights
5$provee data data quality quality to genera generate te the right results results and avoid $a*ing $a*ing Data 7leaning – 5$prov incorr incorrect ect conclus conclusion ionss- auto$a auto$ate te the proces processs but involv involvee e$ploy e$ployees ees to overse overseee the data data cleaning and ensure accuracy 6ro& a ata s7ien7e tea% – 5nclude on your science tea$ individuals with advanced degr degree eess in stat statis isti tics cs who who will will focu focuss on dat dataa $od $odeli eling ng and and predi predict ctio ions ns,, as well well as infrastructure engineers, software developers, and &T< e'perts- then, give the tea$ the large# scale data analysis platfor$s they need to auto$ate data collection and analysis Opti%i'e an repeat – Perfect your data analysis $odel so you can repeat the process to generate accurate predictions, reach goals, and $onitor and report consistently
Types Types o$ Data Analysis. # Univariate Data Analysis /# Bivariate Data Analysis +# M)ltivariate Data Analysis 57
!ata nalysis is a process of collecting, transfor$ing, cleaning, and $odeling data with the goal of discov discoveri ering ng the requir required ed infor$ infor$ati ation( on( The result resultss so obtain obtained ed are co$$un co$$unica icated ted,, sug sugges gestin ting g conclusions, and supporting decision#$a*ing( !ata visuali+ation is at ti$es used to portray the data for the ease of discovering the useful patterns in the data( The ter$s !ata Modeling and !ata nalysis $ean the sa$e( !ata nalysis Process consists of the following phases that are iterative in nature X •
!ata Require$ents Specification
•
!ata 0ollection
•
!ata Processing
•
!ata 0leaning
•
!ata nalysis
•
0o$$unication
Data Re)ire%ents Spe7i$i7ation
The data required for analysis is based on a question or an e'peri$ent( 3ased on the require$ents of those directing the analysis, the data necessary as inputs to the analysis is identified 1e(g(,
58
Population of people2( Specific variables regarding a population 1e(g(, ge and 5nco$e2 $ay be specified and obtained( !ata $ay be nu$erical or categorical( Data Colle7tion
!ata 0ollection is the process of gathering infor$ation on targeted variables identified as data require$ents( The e$phasis is on ensuring accurate and honest collection of data( !ata 0ollection ensures that data gathered is accurate such that the related decisions are valid( !ata 0ollection provides both a baseline to $easure and a target to i$prove( !ata is collected fro$ various sources ranging fro$ organi+ational databases to the infor$ation in web pages( The data thus obtained, $ay not be structured and $ay contain irrelevant infor$ation( .ence, the collected data is required to be subjected to !ata Processing and !ata 0leaning( Data Pro7essing
The data that is collected $ust be processed or organi+ed for analysis( This includes structuring the data as required for the relevant nalysis Tools( %or e'a$ple, the data $ight have to be placed into rows and colu$ns colu$ns in a table within a Spreadsheet Spreadsheet or Statistical Statistical pplicat pplication( ion( !ata Model $ight have to be created( Data Cleaning
The processed and organi+ed data $ay be inco$plete, contain duplicates, or contain errors( !ata 0leaning is the process of preventing and correcting these errors( There are several types of !ata 0leaning that depend on the type of data( %or e'a$ple, while cleaning the financial data, certain totals $ight be co$pared against reliable published nu$bers or defined thresholds(
!ata that is processed, organi+ed and cleaned would be ready for the analysis( "arious "arious data analysis techniques are available to understand, interpret, and derive conclusions based on the require$ents( !ata "isuali "isuali+ation +ation $ay also be used to e'a$ine the data in graphical graphical for$at, to obtain additional additional insight regarding the $essages within the data( Statistical !ata Models such as 0orrelation, Regression nalysis can be used to identify the relations a$ong the data variables( These $odels that are descriptive of the data are helpful in si$plifying analysis and co$$unicate results( The process $ight require additional !ata 0leaning or additional !ata 0ollection, and hence these activities are iterative in nature( Co%%)ni7ation
The results of the data analysis are to be reported in a for$at as required by the users to support their decisions and further action( The feedbac* fro$ the users $ight result in additional analysis( The data analysts can choose data visuali+ation techniques, such as tables and charts, which help in co$$unicating the $essage clearly and efficiently to the users( The analysis tools provide facility to highlight the required infor$ation with color codes and for$atting in tables and charts(
59
5 Most *mportant Methods or "tatistical Data 3nalysis '+ Mean The arith$etic $ean, $ore co$$only *nown as Kthe average,L is the su$ of a list of nu$bers divided by the nu$ber of ite$s on the list( The $ean is useful in deter$ining the overall trend of a data set or providing providing a rapid snapshot snapshot of your data( nother nother advantage of the $ean is that itHs itHs very easy and quic* to calculate( + "tandard De#iation The standard deviation, often represented with the Cree* letter sig$a, is the $easure of a spread of data around the $ean( high standard deviation signifies that data is spread $ore widely fro$ the $ean, where a low standard deviation signals that $ore data align with the $ean( 5n a portfolio of data analysis $ethods, the standard deviation is useful for quic*ly deter$ining dispersion of data points( 1+ Regression Regression $odels the relationships between dependent and e'planatory variables, which are usually charted on a scatterplot( The regression line also designates whether those relationships are strong or wea*( Regression is co$$only taught in high school or college statistics courses with applications for science or business in deter$ining trends over ti$e( 4+ "ample "i$e Determination )hen $easuring a large data set or population, li*e a wor*force, you donHt always need to collect infor$ation fro$ every $e$ber of that population – a sa$ple does the job just as well( The tric* is to deter$ine the right si+e for a sa$ple to be accurate( Fsing proportion and standard deviation $ethods, you are able to accurately deter$ine the right sa$ple si+e you need to $a*e your data collection statistically significant( 5+ 7ypothesis 7ypothesi s Testing Testing lso co$$only called t testing, testing, hypothesis testing assesses if a certain pre$ise is actually true for your data set or population( 5n data analysis and statistics, you consider the result of a hypothesis test if the results couldnHt have happened by rando$ chance( .ypothesis tests are statistically sinificant if used in everything fro$ science and research to business and econo$ic TESTIN6 O0 HPOTHESIS.
De$inition o$ Hypot8esis. Hypot8esis
hypothesis is an idea idea which which is suggested as a possible a possible e'planation e'planation for a particular particular situation or condition, but which has not yet been proved been proved to to be correct correct((
6!
statisti7al 8ypot8esis , so$eti$e so$eti$ess called 7on$ir%atory ata analysis , is is a hypothesis hypothesis that is testab testable le on the basis basis of observing a proc proces esss that is $odeled via via a set of of rando$ variables( variables(Y6Z inference( 0o$$only, two statistical data sets statisti7al 8ypot8esis test is a $ethod of statistical inference( are co$pared, or a data set obtained by sa$pling is co$pared against a synthetic data set fro$ an ideali+ed $odel( hypothesis is proposed for the statistical relationship between the two data sets, and this is co$pared as an alternative to an ideali+ed null hypothesis that proposes no relationship between two data sets( The co$parison is dee$ed statistically sinificant if the relationship between the the data data sets sets would would be an unli unli*el *ely y real reali+ i+at atio ion n of the the null hypothesis hypothesis according to a threshold probability[the significance level( .ypothesis tests are used in deter$ining what outco$es of a study would lead to a rejection of the null hypothesis for a pre#specified level of significance( The process of distinguishing between the null hypothesis and the alternative hypothesis hypothesis is aided by identifying two conceptual types of errors 1type 6 A type 82, 82, and by specifying para$etric li$its on e(g( how $uch type 6 error will be per$itted(
What is a hypothesis test8 hypothesis test is a statistical statistical test that is used to deter$ine whether there is enough en ough evidence in a sa$ple of data to infer that a certain condition is true for the entire population( hypothesis test e'a$ines two opposing hypotheses about a population; the null h ypothesis and the alternative hypothesis( The null hypothesis is the state$ent being tested( Fsually the null h ypothesis is a state$ent of no effect or no difference( The alternative hypothesis is the state$ent you want to be able to conclude is true( 3ased on the sa$ple data, the test deter$ines whether to reject the null hypothesis( Nou Nou use a p# value, to $a*e the deter$ination( de ter$ination( 5f the p#value is less than or equal to the level of significance, which is a cut#off point that you define, then you can reject the null hypothesis( h ypothesis( co$$on $isconception is that statistical hypothesis tests are designed to select the $ore li*ely of two hypotheses( 5nstead, a test will re$ain with the null hypothesis until there is enough evidence 1data2 to support the alternative hypothesis( &'a$ples of questions you can answer with a hypothesis test include; •
!oes the $ean height of undergraduate u ndergraduate wo$en differ fro$ BB inches
•
5s the standard deviation of their height equal equa l less than @ inches
•
!o $ale and fe$ale undergraduates differ in height
61
Types of 7ypothesis &ull and 3lternati#e hypothesis hypothesis test e'a$ines two opposing hypotheses about a population; the null h ypothesis and the alternative hypothesis( .ow you set up these hypotheses h ypotheses depends on what you are trying to show( N)ll 8ypot8esis ?H 2@ The null hypothesis states that a population para$eter pa ra$eter is equal to a value( The null hypothesis h ypothesis is often an initial clai$ that researchers specify using previous research or *n owledge( Alternative Hypot8esis ?H @ The alternative hypothesis states that the population para$eter is different than the value of the population para$eter in the null nu ll hypothesis( The alternative hypothesis is what you $ight believe to be true or hope to prove true(
What are type * and type ** errors8 )hen you do a hypothesis test, two types of errors are possible; type 5 and type 55( The ris*s of these two errors are inversely related and deter$ined by the level of significance and the power for the test( Therefore, you should deter$ine which error has $ore severe consequences for your situation before you define their ris*s( /o hypothesis test is 699\ certain( 3ecause the test is based on probabilities, probabilities, there is always a chance of drawing an incorrect conclusion( Type 5 error )hen the null hypothesis is true and you reject it, you $a*e a type 5 error( The probability of $a*ing a type 5 error is ], which is the level of significance you set for your hypothesis test( n ] of 9(9@ indicates that you are willing to accept a @\ chance that you are wrong when you reject the null hypothesis( To lower this ris*, you $ust use a lower value for ]( .owever, using a lower value for alpha $eans that you will be less li*ely to detect a true difference if one really e'ists( Type 55 error )hen the null hypothesis is false and you fail to reject it, you $a*e a type 55 error( The probability of $a*ing a type 55 error is ^, which depends on the power of the test( Nou Nou can decrease your ris* of co$$itting a type 55 error by ensuring your test has enough power( Nou can do this by ensuring your sa$ple si+e is large enough to detect a practical difference when one truly e'ists( The probability of rejecting the null hypothesis when it is false is equal to 6–^( This value is the power of the test(
62
&ull 7ypothesis :rue Decisio :rue n
0alse
0ail to reect
orrect ,ecision (pro%a%ilit# I 1 > J
nu ll Type ** 9rror > "ail to reect the null .hen it is "alse (pro%a%ilit# I K
)eect
Type * 9rror > reecting the null .hen it is true (pro%a%ilit# I J
orrect ,ecision (pro%a%ilit# I 1 > K
xample o" t#pe < and t#pe << error To understand the interrelationship between type 5 and type 55 error, and to deter$ine which error has $ore severe consequences for your situation, consider the following e'a$ple( $edica $edicall resear researche cherr wants wants to co$par co$paree the effecti effectivene veness ss of two $edica $edicatio tions( ns( The null null and alternative hypotheses are; /ull hypothesis 1.92; _6` _8 The two $edications are equally effective( lternative hypothesis 1.62; _6 _8 The two $edications are not equally effective( type 5 error occurs if the researcher rejects the null hypothesis and concludes that the two $edications are different when, in fact, they are not( 5f the $edications have the sa$e effectiveness, the researcher $ay not consider this error too severe because the patients still benefit fro$ the sa$e level of effectiveness regardless of which $edicine they ta*e( .owever, if a type 55 error occurs, the researcher fails to reject the null hypothesis when it should be rejected( That is, the researcher concludes that the $edications are the sa$e when, in fact, they are different( This error is potentially life#threatening if the less#effective $edication is sold to the public instead of the $ore effective one( •
•
s you conduct your hypothesis tests, consider the ris*s of $a*ing type 5 and type 55 errors( 5f the consequences of $a*ing one type of error are $ore severe or costly than $a*ing the other type of error, then choose a level of significance and a power for the test that will reflect the relative severity of those consequences( PARAMETRIC PARAMETRIC TESTS USED 0OR TESTIN6 O0 HPOTHESIS. HPOTHESIS. # L-Test /# T-Test +# / - Test # L-Test 63
A z-test test is is a statistical test test used used to determine whether two population means are different when the variances are known and the sample size is large. The test teststatistic statistic is assumed to have a normal distribution, and nuisance parameters such as standard deviation should be known for an accurate z-test test to to be performed.
ANOVA" C8i-s)are
ANOVA ?Analys ?Analysis is o$ o $ Varian7e@ /O" is a statistical technique that assesses potential differences in a scale#level dependent variable variab le by a no$inal#level no$inal#level variable variable having 8 or $ore categories( categories( %or e'a$ple, an /O" /O" can e'a$inee potential differen e'a$in differences ces in 5 score scoress by 0ountr 0ountry y 1FS vs( 0anada vs( 5taly vs( Spain2( Spain2( The /O", developed by Ronald %isher in 6E6:, e'tends the t and and the 0 test test which have the proble$ of only allowing allowing the no$inal level variable variable to have two categories( categories( This test is also called called the %isher analysis of variance( 6eneral P)rpose o$ ANOVA ANOVA
Researchers and students use /O" /O" in $any ways( The use of /O" /O" depends on the research design( 0o$$only, /O" /O"s s are used in three ways; one#way /O", /O", two#way t wo#way /O", /O", and /# way /O"( One-5ay ANOV ANO VA
one#way one#way /O" /O" has has jus justt one ind indepe ependen ndentt vari variabl able( e( %or e'a$ple, e'a$ple, dif differ ferenc encee in 5 can be assessed by 0ountry, and 0ounty can have 8, 89, or $ore different categories to co$pare( T&o-5ay ANOVA
two#way /O" /O" refers to an /O" using two independent independ ent variables( &'panding the e'a$ple above, a 8#way /O" can e'a$ine differences in 5 scores 1the dependent variable2 by 0ountry 1independent 1inde pendent variable variable 62 and Cender 1independent 1independent variable 82( Tw Two#way o#way /O" /O" can can be used to e'a$ine the interaction between the two independent variables( 5nteractions indicate that differences are not unifor$ across all categories categories of the independent variables( variables( %or e'a$ple, e'a$ple, fe$ales $ay have higher 5 scores overall co$pared to $ales, but this difference could be greater 1or less2 in &uropean countries countr ies co$pared to /ort /orth h $er $erican ican countries( Tw Two#way o#way /O" /O"s are also called facto factorial rial /O"s( N-5ay ANOVA
researcher can also use u se $ore than two independent variables, and this is an n#way /O" /O" 1with n being the nu$ber of independent variables you have2( %or e'a$ple, potential differences in 5 scores can be e'a$ined by 0ountry, Cender, ge group, &thnicity, etc, si$ultaneously(
Ass)%ptions o$ ANOVA 64
&ssumptions in the && *odel The one#way /O" $odel with effects coding is written
GitILMNiMOit=iI1=2==vtI1=2==r i Oitiid(!=R 2
The v treat$ents are inde'ed by i and the nu$ber of replicates receiving the i#th treat$ent is ri( )e have so far focused our analysis on the treat$ent $eans( 5n particular, we constructed the /O" table to find a test statistic to test the null hypothesis that all treat$ent $eans are equal( )e also found found ways ways to test test for pairwi pairwise se diffe differenc rences es between between treat$ treat$ent ent $eans, $eans, and tests tests for esti$a esti$able ble contrasts 1linear co$binations2 of treat$ent $eans( ll of these these tests were were constructed constructed under the the assu$ption assu$ption that ϵ itiid/19,82( )e can deco$pose this assu$ption into three parts; !ndependence of Errors
)e assu$e that each trial is independent of all other trials, e'cept for the effect i of the treat$ent on the $ean( Statistical independence of two trials $eans that *nowing the result of one trial doesnt change the distribut distribution ion of the other trial( The $ost co$$on causes of dependence dependence in e'peri$ental e'peri$ental data are confounding factors # so$ething $easured or un$easured that affects the e'peri$ent( Rando$i+ati Rando$i+ation on is a critical critical technique in e'peri$ental e'peri$ental design because it can $ini$i+e the effect effect of any confounder( ormality of Errors
)e assu$e assu$e in the /O" /O" $odel that the error ter$s are nor$ally#distributed with +ero $ean( 5f the data are not nor$ally#distributed, but instead co$e fro$ so$e other distribution1e'ponential or bino$ial, for e'a$ple2, then we $ay not be able to trust our p#values, which were built by assu$ing nor$ality( E"ual Error #ariance #ariance $cross Treatments
The ne't assu$ption is that all error ter$s have the sa$e variance8( 5t is co$$on to see that different treat$ents see$ to result in different $eans /!
C8i-s)are Te Test. st. 65
•
•
•
•
78i-s)a hypoth othesi esiss tes testt wherein 78i-s)are re test, also written as test, is any statistical hyp the sa$pl sa$pling ing distr distributi ibution on of the the test test stat statis isti ticc is a chi#s chi#squared quared dist distribut ribution ion when when the null hypothesis is true( true( )ithou )ithoutt other other qualif qualifica icatio tion, n, chi#s chi#squar quared ed test test often often is used used as sho short rt for 2earson3s chi#squared test( test( The chi#squared test is used to deter$ine whether there is a significant difference between the e'pected frequencies and the observed frequencies in one or $ore categories( 5n the standard applications of the test, the observations are classified into $utually e'clusive classes, and there is so$e theory, or say null hypothesis, which gives the probability that any observation falls into the corresponding class( The purpose of the test is to evaluate how li*ely it is between the observations and the null hypothesis( 0hi#squared tests are often constructed fro$ a su$ of squared errors, errors, or through the sa$ple variance(( Test statistics that follow a chi#squared distribution arise fro$ an assu$ption of variance independent nor$ally distributed data, which is valid in $any cases due to the central li$it theore$(( chi#squared test can be used to atte$pt rejection of the null hypothesis that the theore$ data are independent( lso considered a chi#squared test is a test in which this is asymptotically true, $eaning that the sa$pling distribution 1if the null hypothesis is true2 can be $ade to appro'i$ate a chi# squared distribution as closely as desired by $a*ing the sa$ple si+e large enough(
There are t&o types o$ 78i-s)are tests ( 3oth use the chi#square statistic and distribution for different purposes; •
•
78i-s)are gooness o$ $it test deter$ines if a sa$ple data $atches a population( %or $ore details on this type, see; (oodness of 4it Test ( 78i-s)are test $or inepenen7e co$pares two variables in a contingency table to see if they are related( 5n a $ore general sense, it tests to see whether distributions of categorical categorical variables differ variables differ fro$ each another( anothe r( very $eans that that you yourr obs observ erved ed data data fits fits your very s%all s%all 78i s)are test statist statisti7 i7 $eans e'pected data e'tre$ely well( 5n other words, there is a relationship( very large 78i s)are test statisti7 $eans that the data does not fit very well( 5n other words, there isnHt a relationship( •
•
Ass)%ptions o$ C8i-s)are Test s with para$etric tests, the non#para$etric tests, including the 8 assu$e the data were obtained through rando$ selection( .owever, it is not unco$$on to find inferential statistics used when data are fro$ convenience sa$ples rather than rando$ sa$ples( 1To have confidence in the results when the rando$ sa$pling assu$ption is violated, several replication studies should be perfor$ed with essentially the sa$e result obtained2( &ach non#para$etric test has its own specific assu$ptions as well( T8e ass)%ptions o$ t8e C8i-s)are in7l)e.
66
6( The data in the cells cells should should be frequenci frequencies, es, or counts counts of cases rathe ratherr than percent percentages ages or so$e other transfor$ation of the data( 8( The levels levels 1or categories2 categories2 of the the variables variables are $utually $utually e'clusive( e'clusive( That is, is, a particular particular subject subject fits into one and only one level of each of the variables( =( &ach subject subject $ay $ay contribute contribute data to to one and only only one cell in in the 8( 5f, for e'a$ple, the sa$e subjects are tested over ti$e such that the co$parisons are of the sa$e subjects at Ti$e 6, Ti$e 8, Ti$e =, etc(, then 8 $ay not be used( ?( The study study groups groups $ust be independe independent( nt( This This $eans $eans that a differe different nt test $ust $ust be used if the two groups are related( %or e'a$ple, a different test $ust be used if the researcherHs data consists of paired sa$ples, such as in studies in which a parent is paired with his or her child( @( There There are 8 variab variables les,, and both are $easur $easured ed as categorie categories, s, usually usually at the no$inal no$inal level( level( .owever, data $ay be ordinal data( 5nterval or ratio data that have been collapsed into ordinal categories $ay also be used( )hile 0hi#square has no rule about li$iting the nu$ber of cells 1by li$iting the nu$ber of categories for each variable2, a very large nu$ber of cells 1over 892 can $a*e it difficult to $eet assu$ption B below, and to interpret the $eaning of the results( B( The The valu valuee of the the cel celll expecteds should be @ or $ore in at least :9\ of the cells, and no cell should have an e'pected of less than one 1=2( This assu$ption is $ost li*ely to be $et if the sa$ple si+e equals at least the nu$ber of cells $ultiplied by @( &ssentially, this assu$ption specifies the nu$ber of cases 1sa$ple si+e2 needed to use the 8 for any nu$ber of cells in that 8( This require$ent will be fully e'plained in the e'a$ple of the calculation of the statistic in the case study e'a$ple( Dis7ri%inant Analysis.
!iscri$inant analysis is a technique that is used by the researcher to analy+e the research data when the criterion or the dependent variable is categorical and the predictor or the independent variable is interval in nature( The ter$ categorical variable $eans that the dependent variable is divide divided d into into a nu$ nu$ber ber of categor categories ies(( %or e'a$ple e'a$ple,, three three brands brands of co$put co$puters ers,, 0o$put 0o$puter er , 0o$puter 3 and 0o$puter 0 can be the categorical dependent variable( The objective of discri$inant analysis is to develop discri$inant functions that are nothing but the linear co$bination of independent independen t variables that will discri$inate between the categories of the dependent variable in a perfect $anner( 5t enables the researcher to e'a$ine whether significant differences e'ist a$ong the groups, in ter$s of the predictor variables( 5t also evaluates the accuracy of the classification( !iscri$inant analysis is described by the nu$ber of categories that is possessed by thse dependent variable( s in statistics, everything is assu$ed up until infinity, so in this case, when the dependent variable has two categories, then the type used is two#group discri$inant analysis( 5f the dependent 67
variable has three or $ore than three categories, then the type used is $ultiple discri$inant analysis( The $ajor distinction to the types of discri$inant analysis is that for a two group, it is possible to derive only one discri$inant function( On the other hand, in the case of $ultiple discri$inant analysis, $ore than one discri$inant function can be co$puted( There are $any e'a$ples that can e'plain when discri$inant analysis fits( 5t can be used to *now *now whet whethe herr heavy heavy,, $edi $ediu$ u$ and light light user userss of soft soft drin drin*s *s are are diff differ erent ent in ter$ ter$ss of thei their r consu$ption of fro+en foods( 5n the field of psychology, it can be used to differentiate between the price sensitive and non price sensitive buyers of groceries in ter$s of their psychological ps ychological attributes or characteristics( 5n the field of business, it can be used to understand the characteristics or the attributes of a custo$er possessing store loyalty and a custo$er who does not have store loyalty( %or a researcher, it is i$portant to understand the relationship of discri$inant analysis with Regression and nalysis of "ariance 1/O"2 which 1/O"2 which has $any si$ilarities and differences( Often we can find si$ilarities and differences with the people we co$e across( Si$ilarly, there are so$e si$ila si$ilarit rities ies and differ differenc ences es with with discri discri$i $inant nant analysi analysiss along along with with two other other proced procedure ures( s( The si$ilarity is that the nu$ber of dependent variables is one in discri$inant analysis and in the other two procedures, the nu$ber of independent variables are $ultiple in discri$inant analysis( The difference is categorical or binary in discri$inant analysis, but $etric in the other two procedures( The nature of the independent variables is categorical in nalysis of "ariance 1/O"2, but $etric in regression and discri$inant analysis( anal ysis(
!iscri !iscri$i $inant nant analysis analysis is a stati statisti stical cal $ethod $ethod that that is used used by resear researcher cherss to help help the$ the$ under underst stan and d the the rela relati tion onsh ship ip betwe between en a dep depend enden entt vari variab able le and one one or $o $ore re inde indepe pend ndent ent variables( dependent variable is the variable that a researcher is trying to e'plain or predict fro$ the values of the independent variables( variables( !iscri$inan !iscri$inantt analysis analysis is si$ilar si$ilar to regression regression analysis analysis and analysis of variance 1/O" 1/O"2( 2( The principal difference between discri$inant analysis and the other two $ethods is with regard to the nature of the dependent variable( !iscri$inant analysis requires the researcher to have $easures of the dependent variable and all of the independent variables for a large nu$ber of cases( 5n regression analysis and /O", the dependent variable $ust be a continuous continuous variable( variable( nu$eric variable indicates the degree to which a subject possesses so$e characteristic, so that the higher the value of the variable, the greater the level of the characteristic( good e'a$ple of a continuous variable is a persons inco$e(
T8e steps involve in 7on)7ting is7ri%inant analysis are as $ollo&s.
The proble$ is for$ulated before conducting( The discri$inant function coefficients are esti$ated( The ne't step is the deter$ination of the significance of these discri$inant functions(
68
One $ust interpret the results obtained( The last and the $ost i$portant step is to assess the validity( 0ACTOR ANA1SIS
%actor nalysis is e'tensively used in business research( The purpose of factor analysis in business research is to reduce the nu$ber of variables by using lesser nu$ber of surrogate variables 1factors2 while retaining the variability( The pri$ary objective is to capture so$e psychological states of custo$ersI respondents that cannot be $easured directly( This is usually i$ple$ented by constructing a scale that $easures attitudes, perceptions, $otivations, etc of targeted custo$ers of a specific line of business( 5nitially, the targeted custo$ersI respondents are subjected to answer a detailed questionnaire with different lifestyle state$ents along with nu$eric or verbal scales( Through this, what really are $easured are custo$ers views regarding few di$ensions 1factors2 concerning the success of a business( These di$ensions are psychological states that cannot be $easured directly( So, factor analysis is used to assess these di$ensions 1factors2 indirectly( So, factor analysis is pri$arily used to si$plify a data set before subjecting it to $ultivariate analysis # $ultiple regression, etc( %actor analysis is a way to ta*e a $ass of data and shrin*ing it to a s$aller data set that is $ore $anageable and $ore understandable( 5tHs a way to find hidden patterns, show how those patterns overlap and show what characteristics are seen in $ultiple patterns( 5t 5t is also used to create a set of variables for si$ilar ite$s in the set 1these sets of variables are called di$ensions2( 5t can be a very useful tool for co$ple' sets of data involving psychological studies, socioecono$ic status and other involved concepts( KfactorL is a set of observed variablesthat variablesthat have si$ilar response patternsThey are associ associate ated d with with a hidden hidden variabl variablee 1calle 1called d a confou confounding nding varia variable ble22 that that isnHt isnHt direct directly ly $easured( %actors are listed according to factor loadings, or how $uch variation in the data they can e'plain(
Ass)%ptions o$ 0a7tor analysis#
0ollo&ing are t8e ass)%ptions t8at $a7tor analysis entails.
The variables included can be condensed into one or $ore underlying factors s a dat dataa red reduct uction ion tec techni hnique, que, fac factor tor ana analys lysis is ass assu$e u$ess tha thatt a res resear earch ch inv involv olves es $ul $ulti tiple ple dependent variables
69
The sa$ple is ho$ogeneous in the sense that the sub groups within the sa$ple will not have different patterns of scores on the variables included in the anal ysis The factors are related to the score of each variable in linear $anner
CON,OINT ANA1SIS. ANA1SIS. survey based statis statistic tical al techni technique que used used in $ar*et $ar*et research research that helps helps Con:oint analysis is a survey deter$ deter$ine ine how people people value value differ different ent attrib attribute utess 1featu 1feature, re, functi function, on, benefi benefits2 ts2 that $a*e $a*e up an individual product or service( 0onjoint analysis is also popularly called trade off analysis as buyers have to let go of certain product features that they consider lucrative to $a*e a $ore practical p ractical purchase( %or %o r e'a$ple a large nu$ber of people planning to buy a new s$art phone $ight thin* that however $uch they want an iPhone B, they will have to be content with a less e'pensive phone( Thus we see that consu$ers are put in a situation where they are forced to evaluate the $erit of the phone attributes such as configuration, OS, price, brand, etc( Thus, broadly broad ly conjoint analysis chec*s the co$pro$ises users $a*e while selecting products or services( Operational Develop%ent
The process of conjoint analysis is described in a si$plified $anner in the following steps; 6( Recogni+ing Recogni+ing the product product attribu attributes; tes; configur configuration, ation, brand, brand, price, price, etc in the the above case( case( 8( Selecting Selecting the i$portance i$portance degree of these these attributes( attributes( =( 0reating 0reating virtual virtual products products by fusing fusing several several degrees degrees of these these attribut attributes( es( ?( 0ollec 0ollectin ting g respon responses ses thro through ugh a surv survey ey(( @( naly+ naly+ing ing the the data data stati statisti stical cally ly(( B( Mar*et Mar*et si$ula si$ulatio tion n of the produc product( t( Con:oint Analysis Usage
0onjoint studies aid in avertising ( 3y research the co$pany can selected the $ost desirable attributes to be $ar*eted( %or e'a$ple, after conjoint analysis, the co$pany deter$ines that brand and hardware configuration of the phone is $ost i$portant to its users( 5t would then design advertise$ents that well pro$ote these attributes and that do not focus on price which is a secondary concern in this case(
7!
The $o The $ost st co$$ co$$on on usag usagee is ne& pro)7 0onjoint nt analysi analysiss identi identifie fiess pro)7tt evelop evelop%en %entt ( 0onjoi opportunities by fusing attributes to generate new products and services that are not yet in the $ar*et( The $ethod is also good for test %areting as it provides infor$ation of the degrees of i$portance of each attribute( Prior to releasing the product full#fledged, it is feasible to foresee the success or failure of a product( 0onjoint analysis is also applicable in situations where seg%entation needs to be done( 0ertain clusters of users give preference to one set of attributes, whereas a different set would be $ore i$portant to few others( 0onjoint uncovers this pattern so that the co$pany can target users accordingly(
1i%itations o$ Con:oint Analysis
%or certain *ind of products, products, consu$ers do their evaluation evaluation built on intangible intangible attributes attributes or i$age( i$age( These products $ostly co$prise of lu'ury ite$s where the e$otional factor rather than the rational side do$inates( 5n cases li*e these, the logic of conjoint analysis does not apply( )ith an e'ception to this situation, conjoint is quite ine'pensive as co$pared to other si$ilar $ethods such as concept testing and hence is hugely popular( 5n a nutshell, it is a versatile and powerful tool to predict consu$er choices, foresee their purchase decisions and hence design and launch products accordingly(
MU1TIDIMENSIONA1 SCA1IN6. The $ain function of $ultidi$ensional Scaling 1M!S2 is to re#project the objects 1sites2 in reduced di$ension ordination space( s stated above, the a'es scores cannot be used in subsequent analyses due to a lac* of independence( The real advantage of M!S over other ordination techniques is that it will generally represent the ordering relationships a$ongst objects within a given set of di$ensions better than eigen#based ordinations( )hilst the eigen#based analyses rotates all of the original a'es such that each a'is $a'i$ $a 'i$i+ i+es es th thee va vari rian ance ce pr pres eser erve ved d by th that at a' a'es es,, M! M!S S con const stru ruct ctss a con confi figur gurat atio ion n of ob obje ject ctss specifically for only a no$inated nu$ber of new a'es( Therefore, when a 8 di$ensional ordination is desired, the configuration fro$ an M!S is opti$i+ed for just two di$ensions whereas the eigen# based object configurations are independent of how $any di$ensions you intend to plot( Principle co$ponents analysis 1P02 can be perfor$ed by either spectral 1eigen2 deco$position of an association $atri' or single value deco$position of the original data $atri'( &ither way, it yields a riid rotation of a'es in that the positions of points relative to one another 1euclidean distances2 are $aintained during rotation( The rotated a'es are referred to as principle co$ponents(
71
Obviously, the higher the degree of associations between the variables 1species2, the $ore strongly focused and oriented the data cloud is and therefore the $ore successful the P0 in ter$s of being able to represent $any variables by only a few new variables( The resulting principle co$ponents 1a'es2 and scores 1new variable values2 are independent of one another and therefore the reduced data can be used in subsequent statistical analyses( %or e'a$ple, principle co$ponent 6 1as one $easure of the co$$unity2 could be regressed against an enviro env iron$en n$ental tal var variab iable le 1su 1such ch as alt altitu itude de or rai rainfa nfall ll22 in a reg regula ularr lin linear ear $od $odell elling ing pro proced cedure ure(( Si$ilarly, $ultiple correlated predictor variables 1that otherwise violate $ulti#collinearity2 could be reduced to one or two orthogonal predictors to use in linear $odelling( Principle co$ponents analysis is therefore often an inter$ediate step to used as either responses or predictors in linear $odeling( M!S is $athe$atically and conceptually relatively si$ple, yet co$putationally very intense( Rather than have elegant $athe$atical solutions, M!S wor*s on bruit grunt(
C1USTER ANA1SIS ;
Cl)ster analysis or 7l)stering is the tas* of grouping a set of objects in such a way that objects in the sa$e group 1called a 7l)ster 2 are $ore si$ilar 1in so$e sense or another2 to each other than to those in other groups 1clusters2( 5t is a $ain tas* of e'ploratory data $ini $ining ng,, and a co$$on technique technique for statistical data analy analysis sis,, used used in $any $any fiel fields ds,, incl includi uding ng $achin $achinee learni learning ng,, pattern recognition,, i$age analysis, recognition analysis, infor$ation retrieval, retrieval, bioinfor$atics bioinfor$atics,, data co$pr co$pressio ession n, and co$puter graphics(( graphics
0luster analysis itself is not one specific algorith$ algorith$,, but the general tas* to be solved( 5t can be achieved by various algorith$s that differ significantly in their notion of what constitutes a cluster and and how how to eff efficien cienttly find the$ he$( Popu Popula larr not notions ions of clus cluste ters rs inclu nclude de grou groups ps wit with s$all distances a$on a$ong g the the clus clustter $e$be e$berrs, dens densee area areass of the the dat data space pace,, inter nterva vals ls or particular stati statistica sticall distr distributi ibutions ons(( 0lus 0luste teri ring ng can can ther theref efor oree be for$ for$ul ulat ated ed as a $ulti#objective opti$i+ation proble$( opti$i+ation proble$( The appropriate clustering algorith$ and para$eter settings 1including values such as the distance function to function to use, a density threshold or the nu$ber of e'pected clusters2 depend on the individual data set and intended use of the results( 0luster analysis as such is not an auto$atic tas*, but an iterative process of *nowledge discovery or discovery or interactive $ulti#objective opti$i+ation that involves involves trial and failure( failure( 5t is often necessary to $odify $odify data preprocessing preprocessing and $odel para$eters para$eters until the result achieves the desired properties( 3es 3esides ides the ter$ er$ clusterin , there are a nu$ber of ter$s with si$ilar $eanings, including automatic classification, numeri 1fro$ Cree* Cree* ^ ^ grape grape2 2 numerical cal taxonom taxonomyy, botryoloy1fro$ and typoloical analysis( The subtle differences are often in the usage of the results; while in data 72
$ining, the resulting groups are the $atter of interest, in auto$atic classification the resulting discri$inative power is of interest( Cl)ster analysis is a data e'ploration 1$ining2 tool for dividing a $ultivariate dataset into KnaturalL clusters 1groups2( )e use the $ethods to e'plore whether previously undefined clusters 1groups2 $ay e'ist in the dataset( %or instance, a $ar*eting depart$ent $ay wish to use survey results to sort its custo$ers into categories 1perhaps those li*ely to be $ost receptive to buying a product, those $ost li*ely to be against buying a product, and so forth2(
0luster nalysis is used when we believe that the sa$ple units co$e fro$ an un*nown nu$ber of distinct populations or sub#populations( )e also assu$e that the sa$ple units co$e fro$ a nu$ber of distinct populations, but there is no apriori definition of those populations( Our objective is to describe those populations using the observed data( 0luster 0luster nalysis, nalysis, until relatively relatively recently, recently, has had very little little interest( interest( This has changed because of the interest in the bioinfor$atics and geno$e research( To e'plore 0luster nalysis in our lesson here, we will use an ecological e'a$ple(
&pplications o" luster &nal#sis •
•
•
•
0lustering analysis is broadly used in $any applications such as $ar*et research, pattern recognition, data analysis, and i$age processing( 0lustering can also help $ar*eters discover distinct groups in their custo$er base( nd they can characteri+e their custo$er groups based on the purchasing patterns( 5n the field of biology, it can be used to derive plant and ani$al ta'ono$ies, categori+e genes with si$ilar functionalities and gain insight into structures inherent to populations( 0lustering also helps in identification of areas of si$ilar land use in an earth observation database( database( 5t also helps in the identificat identification ion of groups of houses in a city according according to house type, value, and geographic location(
•
0lustering also helps in classifying docu$ents on the web for infor$ation discovery(
•
0lustering is also used in outlier detection applications such as detection of credit card fraud(
•
s a data data $inin $ining g funct functio ion, n, clus cluste terr analy analysi siss serv serves es as a tool tool to gain gain insi insight ght into into the the distribution of data to observe characteristics of each cluster(
)e;uirements o" lustering in ,ata *ining The following points throw light on why clustering is required in data $ining X •
S7ala!ility X )e need highly scalable clustering algorith$s to deal with large databases( 73
•
•
•
•
•
A!ility to eal &it8 i$$erent ins o$ attri!)tes X lgorith$s should be capable to be applied on any *ind of data such as interval#based 1nu$erical2 data, categorical, and binary data( Dis7overy o$ 7l)sters &it8 attri!)te s8ape X The clustering algorith$ should be capable of detecting clusters of arbitrary shape( They should not be bounded to only distance $easures that tend to find spherical cluster of s$all si+es( Hig8 i%ensionality X The clustering algorith$ should not only be able to handle low# di$ensional data but also the high di$ensional space( A!ility to eal &it8 noisy ata X !atabases contain noisy, $issing or erroneous data( So$e algorith$s are sensitive to such data and $ay $a y lead to poor quality clusters(
clusterin ring g result resultss sho should uld be interp interpret retable able,, co$preh co$prehens ensibl ible, e, and Interpreta!ility X The cluste usable(
lustering *ethods 0lustering $ethods can be classified into the following categories X •
Partitioning Method
•
.ierarchical Method
•
!ensity#based Method
•
Crid#3ased Method
•
Model#3ased Method
•
0onstraint#based Method
Partitioning Method Suppose we are given a database of WnH objects and the partitioning $ethod constructs W*H partition of data( &ach partition will represent a cluster and * k n( 5t $eans that it will classify the data into * groups, which satisfy the following require$ents X •
&ach group contains at least one object(
•
&ach object $ust belong to e'actly one group(
Points to re%e%!er •
%or a given nu$ber of partitions 1say *2, the partitioning $ethod will create an initial partitioning( 74
•
Then it uses the iterative relocation technique to i$prove the partitioning by $oving objects fro$ one group to other(
.ierarchical Methods This $ethod creates a hierarchical deco$position of the given set of data objects( )e can classify hierarchical $ethods on the basis of how the hierarchical deco$position is for$ed( There are two approaches here X •
gglo$erative pproach
•
!ivisive pproach
Agglo%erative Approa78
This approach is also *nown as the botto$#up approach( 5n this, we start with each object for$ing a separate separate group( 5t *eeps on $erging $erging the objects objects or groups that are close to one another( another( 5t *eep on doing so until all of the groups are $erged into one or until the ter$ination condition holds( Divisive Approa78
This approach is also *nown as the top#down approach( 5n this, we start with all of the objects in the sa$e cluster( 5n the continuous iteration, a cluster is split up into s$aller clusters( 5t is down until each object in one cluster or the ter$ination condition holds( This $ethod is rigid, i(e(, once a $erging or splitting is done, it can never be undone( Approa78es to I%prove ;)ality o$ Hierar78i7al Cl)stering
.ere are the two approaches that are used to i$prove the quality of hierarchical clustering X •
•
Perfor$ careful analysis of object lin*ages at each hierarchical partitioning( 5ntegrate hierarchical agglo$eration by first using a hierarchical agglo$erative algorith$ to group group object objectss into into $icro# $icro#clu cluste sters, rs, and then then perfor perfor$in $ing g $acro# $acro#clu cluste sterin ring g on the $icro# $icro# clusters(
Density-!ase Met8o
This $ethod is based on the notion of density( The basic idea is to continue growing the given cluster as long as the density in the neighborhood e'ceeds so$e threshold, i(e(, for each data point within a given cluster, the radius of a given cluster has to contain at least a $ini$u$ nu$ber of points(
75
6ri-!ase Met8o
5n this, the objects together for$ a grid( The object space is quanti+ed into finite nu$ber of cells that for$ a grid structure( Avantage •
The $ajor advantage of this $ethod is fast processing ti$e(
•
5t is dependent only on the nu$ber of cells in each di$ension in the quanti+ed space(
Moel-!ase %et8os
5n this $ethod, a $odel is hypothesi+ed for each cluster to find the best fit of data for a given $odel( $od el( This This $ethod $ethod locate locatess the cluste clusters rs by cluste clusteri ring ng the density density functi function( on( 5t reflec reflects ts spatia spatiall distribution of the data points( This $ethod also provides a way to auto$atically deter$ine the nu$ber of clusters based on standard statistics, ta*ing outlier or noise into account( 5t therefore yields robust clustering $ethods( Constraint-!ase Met8o
5n this $ethod, the clustering is perfor$ed by the incorporation of user or application#oriented constraints( constraint refers to the user e'pectation or the properties of desired clustering results( 0onstraints provide us with an interactive way of co$$unication with the clustering process( 0onstraints can be specified by the user or the application require$ent(
Unit V REPORT 5RITIN6
Types of Reports, 3usiness, Technical and cade$ic Report writing – Methodology Procedure – 0ontents – 3ibliography report is &ritten for a cle clear ar pur purpos posee and to a par partic ticula ularr audi audience ence(( Spe Specif cific ic inf infor$ or$ati ation on and evidence are presented, analysed and applied to a particular proble$ or issue( ((( )hen you are as*ed to &rite a report you will usually be given a report brief which provides you with instructions and guidelines( Report )riting )riting is a well#organi+ed writing process that needs enough eno ugh s*ill, research, and details(
76
This report writing process is usually ti$e#consu$ing and often requires a detailed research for which the report writer $ay have to visit places, $eet people and find a solid conclusion at the end of the day to write a good go od report( Fse of report writing is to e'plain a $atter or issue and infor$ higher authorities to help the$ ta*e a right decision or action in regards to the $atter or issue( There is a little roo$ for creativity in the report writing structure as the $ain focus is to enlighten the readers about a $atter or issue and $a*e $a* e the whole thing self#e'planatory for easy understanding( Report Rep ort )ritin )riting g is the pri pri$ary $ary tool of $ed $edia ia per person sonnel nel thr through ough which they pass on spe specif cific ic infor$ation about an incident or topic( part fro$ the $edia usage, report writing is required in various sectors li*e corporate, govern$ent, politics etc( To offer a better understanding, learning about the essence of this for$ of writing – WThe ReportH is very i$portant( 58at is a report9
report is a $ethodical, well planned docu$ent which outlines and eva luates a subject or proble$, and which $ay include; The record of an order of events &'planation of the i$plication of these events or facts &valuation of the facts or results of research presented !iscussion of the consequences of a conclusion or course of action 0onclusions References
77
)eports must al.a#s %eH 0orrect 0risp 0lear )ell#Structured
)eport riting 0ormat &very organi+ation has a predefined report writing for$at and hen ce a writer canHt re$ain stuc* to a specific for$at for different reports( )hile writing a for$al report, the freelance writer $ust *eep in $ind that the target readers need donHt have to do unnecessary research to ta*e a decision or action after reading the report(
:#pes : #pes o" )eport )eport riting riting
78
&verything $ust be in details( There are $any types t ypes of report writing for organi+ations that are used for various purposes( To for$at for$at your report to serve varied needs, have a loo* at the pri$ary report writing types $entioned below(
0ormal )eport riting %or$al Report )riting is pretty co$ple' and ti$e#consu$ing( Fsually, it de$ands an i$$ense research, e'planation, references, lin*s, lists and $any other things to $a*e the pri$ary point clear enough for the readers( This type of report writing is usually preferred for an i$portant incident, issue or $atter by big organi+ations( %or$al report writing is generally long and e'pensive( %or$al Report writing has an internationally accepted pattern that includes various co$ponents that are $entioned below( The cover of a report is so$ething that drives a readerHs attention first( The reportHs cover Cover The leaves a huge i$pression on the readerHs $ind and heIshe can get an idea regarding the reportHs topic or quality through the cover design( There is a saying – WHdonHt judge a boo* by its coverHH but it is also true – what see$s well, sells well and hence the report writer $ust ta*e care of the cover of the report with has a $ajor role to depict the report brief( This co$ponent includes the reportHs title and the na$e of the writer( part Title O$ T8e Report This fro$ these things, the title can also have a date and the na$e of the organi+ation for which the report has been prepared( The cover also has these things but putting the$ in details in the title section is $andatory( This section includes headings and subheadings of the pri$ary te't written( Ta!le O$ Contents This This is a very i$portant portion of report writing( 5t helps your readers to reach desired sections in your report in a hassle#free $anner( .ere you basically provide the synopsis of the whole reportHs pri$ary te't and you can S)%%ary .ere also call it an infor$ative su$$ary(
79
Many ti$es, it is referred as We'ecutive su$$aryH( Nou can also use descriptive su$$ary which is a si$ple table of contents( The for$at is always decided by the organi+ation( This is the $ost i$portant section of the $ain te't( The $ain te't always includes Intro)7tion Intro )7tion This three co$ponents – introduction, discussion, and conclusion( .ere, you introduce the $ain te't of your report in the $ost intriguing a detailed $anner so that all types of readers can get your point without $uch effort( discus cussio sion n sec sectio tion, n, a rep report ort wri writer ter dis discus cusses ses the $ai $ain n sto story ry of the rep report ort(( Dis7)ssion 5n the dis ccording to your readerHs convenience, you decide the order of the reportHs results( Nou can als also o do a res result ult to the theory ory co$ co$par pariso ison n her heree alo along ng wit with h the analysis analysis,, eva evalua luatio tion n and interpretation of the data included( Con7l)sionRe7o%%enation Nou can present the su$$ary of the discussion section here( .ere,
you $ention your findings and reco$$end the ele$ents to your readers as per your overall evaluation( 5n the appendi' portion, you can attach the graphs, lists, survey and suchli*e stuff that Appeni< Appen i< 5n are related to your report and helps he lps your readers to understand the report co$prehensively( co $prehensively(
-est Process 0or )eport riting To bring up an effective report, the right process has to be followed( .ere are seven steps to underta*e best report writing process( 8!
,ecide the %ectiveH
Cnderstand Gour &udienceH &udience H Right understanding of the audience definitely leads to a quality report( %or e'a$ple, an annual financial report for sta*eholders is co$pletely c o$pletely different fro$ a financial review( nd with an understanding, a writer can alter the use of language, data incorporated and supporting $aterial that can uplift the indulgence for the set audience( .aving a personal touch as per the audiencesH preference can help produce ideas based on their choice( )ith an understanding writer can present the report that suits their preference(
)eport 0ormat and :#pes %or a report to be an effective co$$unication tool, it $ust follow a particular for$at or type( !eciding on para$eters li*e, written report or presentation- what type – for$al, infor$al, financial, annual, technical, fact#finding or proble$#solving report- design te$plates if any available(
81
ollect the 0acts and ,ata dding figures, facts and data adds credibility to the report and strengthens the argu$ent( dding data or facts brings along a crucial responsibility to cite or $ention the sources, li*e interviews, articles, sayings, articles, etc(
tructure the )eportH report typically has four ele$ents, the e'ecutive su$$ary 1this is written after the report is finished2, introduction 1this includes the structure of the report and table of contents2, body 1$ain te't and report is occupied in this portion2, conclusion 1this is a binding portion that brings all the ele$ents of the report into a syste$atical end2(%or a better understanding on the structuring of a report, read The Structure of 5eports by 6ni7earnin (
)eada%ilit# The readability part is definitely a crucial aspect as it beco$es a $ust to $a*e the report enjoyable and accessible to read( great navigation is the best way to $a*e the reader ta*e a unifor$ed path through the infor$ation flow(
82
dding proper for$atting 1h6, h8, h=2, graphicsIvisuals, brea* up of long te't into shorter for better read and giving the te't in report writing structure with bullet points for better understanding(
dit The initial draft of report writing is never perfect 1at least E9\ of ti$es2( This calls for edition and revision of the content( 3est practice can be *eeping aside the report docu$ent for few days and then once $ore start wor*ing over it again or as* a fellow $e$ber to review or proofread it for you( lso while you edit or get it edited, $a*e sure you *eep writing sa$ples handy( These sa$ples can act as a guideline(
xample S ample o" )eports fter attaining all the understanding about writing reports, itHs ti$e to loo* at so$e actual e'a$ples of report writing(
)eport riting xample To grab a better and practical understanding, letHs have a loo* at one of the e'a$ples of report writing for$at here here((
)eport riting amples There are $any $ore report writing sa$ples available over the web that can give a better idea of how to bring up a perfect report( Though, writing a report is often considered as a co$plicated process, but if you lac* the e'pertise in writing your report then outsourcing the writing process can be the best available option( These hired professionals will offer you a report brief before they sub$it you the final version( They will help you with writing reports to perfection by offering a draft and a final copy that is as per your need( sa$ple report before the final copy lets you achieve an effective report( They are well#versed to write reports and essential ele$ents that are i$portant for a good report li*e, how to for$at your report, how to present the content, etc(
83
This the sa$ple report $entioned above, aspiring as well as established report writers can $a*e the best out of report writing( 0inal 5ors
Report writing is the best $ean to end a project and docu$ent the *nowledge or infor$ation for the ne't upco$ing project( These above seven steps offer a syste$atic and logical way to process the te't into a report and develop a report that faultlessly solves its purpose(
Te78ni7al Te 78ni7al 5riting
Technical writing is closely related to the process, or how to, essay( 5ts function is to clearly e'plain the steps to acco$plishing a tas* so anyone can do it( s such, its audience is the average layperson( Many technical writing assign$ents are re$iniscent of instruction $anualsthis is because $anual writing is one type of technical writing( Other types would include busine bus iness ss letters let ters,, $e$os, $e$o s, produc pro ductt descrip des crip tion s, warnin war ning g labels lab els and, and , to so$ so$ee e'tent e't ent,, editor edi torial ial letters( A7ae%i7 5riting.
cade$ic writing is by definition $ore co$ple' because $ost acade$ic writing is tied to a specific discipline or field, which $eans it can beco$e jargon#laden( This occurs even in the busine bus iness ss and co$put co$ puter er sectors sec tors,, two areas are as *no *nown wn for their the ir e$phas e$p hasis is on techni tec hnical cal writing writ ing(( Scholarly articles and te'tboo*s in either of those disciplines will contain as $uch jargon and as $any co$ple' ideas as the $ost challenging essay in literature, art or $usic( This is to be e'pected because all fields have both speciali+ed *nowledge and speciali+ed ter$inology(
ey Di/erence ; 3cademic #s+ Technical Writing &cademic and technical .riting are t.o "orms o" .riting %et.een .hich a Ae# di$erence can %e identied *ost people assume that a technical .riter is= in "act= an academic .riter as .ell :his= ho.ever= is a "alse assumption <hough %oth acad academic emic .rit .riting ing and tech technica nicall .rit .riting ing re re;uir ;uire e ex excell cellent ent .rit .riting ing sAill sAills= s= di/erence rences s %e audience ence the .ey di/e %et.e t.een en th these ese t.o t#p t#pes es o" .ri .ritin ting g ar are e the audi and purpose of
not the case ven a la# person can %e the targ target et audience :hrough :hrough this article let us examine the di$erences %et.een academic and technical .riting hat is &cademic riting+ cade$ic writing is a $or% o$ &riting t8at is )se in a7ae%i7 is7iplines ( This includes both the natural sciences as sciences as well as the social sciences( sciences( Scholars use acade$ic writing for $any reasons( They can use it to present the findings of a new research that they conducted or even to present a new point of view( The target audience of acade$ic writing is usually scholars that belong to a particular discipline( %or acade$ic writing, the writer uses a special jargon( 5f you go through journal through journal articles, articles, research papers,, dissertations papers dissertations,, you will notice that not only the jargon but even the style of writing is quite different to what we see every day because the style is very i$personal( Nou can also notice inter# te'tuality, or else the quoting of previous wor*s to support or oppose certain argu$ents( !eveloping the ability to write acade$ic articles is not an easy tas*, it requires an e'tensive *nowledge of the subject as well as e'cellent writing s*ills(
Preparation o$ reports. 0or preparations or ra$ting o$ reports t8ree steps are involve#
62 %irst !raft; Co%pre8ensiveness or $)llness o$ $a7ts# Pre7ision or A77)ra7y o$ 0a7ts Co8eren7e or logi7 o$ $a7ts" an Move%ent or transition o$ $a7ts an ieas#
8nd( The researcher should give the first draft, at this stage, a shape so that it can be readable, clear and lucid( 0onsiderable tri$$ing or editing will have to be done to $a*e the writing precise, 0oncise and brief( %inally, at the second draft stage, critical evaluation will have to be $ade of all that has been written#facts, findings, conclusions and reco$$endations( =rd( nd final The The final stage in drafting is the preparations preparations of final report( 5t concentrates concentrates $ainly on the finis finish h and final touches, touches, i(e(, on docu$entation docu$entation and polis polish h to $a*e the report weighty, authoritative, convincing and attractive( !ocu$entation indicated the references to the sources, other previous and current wor* and view, additional data and discussion and suggested further reading on the specific proble$ as handled by the researcher( 5n other words, it indicated the thoroughness of the investigation and on the other a guide to further wor* 85
C8ara7teristi7s o$ goo report
62 ttractive 82 0lear Topic =2 3alanced
CONTENTS O0 A RESEARCH REPORT A resear78 report orinarily in7l)es t8e $ollo&ing se7tions.
often required to be no $ore than a given $a'i$u$ nu$ber of words, ABSTRACT( The abstract is often usually between 699 and 6@9( 5t should describe the $ost i$portant aspects of the study, study, including including the proble$ investigated, the type of subjects 1sa$ple2 and data collection $ethod involved, the analytical procedures used, and the $ajor results and conclusions( discussions concerning the practical practical andIor theoretical theoretical INTRODUCTION ( This section includes discussions i$po i$port rtanc ancee of the the topi topicc as well well as a desc descri ript ptio ion n of the the rese resear arch ch prob proble le$( $( 5t ofte often n star starts ts by introducing the reader to the topic and $a*ing a case for the practical significance of the issues being investigated andIor the contribution that the study could $a*e to our understanding of the pheno$enon( The state$ent of research proble$1s2 is intended to indicate what the general purpose of the study is( This is often done through broadly stated questions or state$ents regarding whether and and how how the the rese resear arch ch vari variabl ables es are are 1or 1or are are e'pe e'pect cted ed to be2 rela relate ted d to 1or 1or affe affect cted ed by2 by2 one one another( %inally, the introduction section should include the wor*ing definitions of those those ter$s used in the study that do not have a co$$only *nown $eaning or for which several $eanings $ay be used( authors present present the REVIE5 O0 THE 1ITERAT 1I TERATURE URE AND THE RESEARCH MODE1 ( So$e authors $aterial included here as a separate section under its own heading 1as it is shown here2, while others present it as part of a longer 5/TRO!F0T5O/ section( 5n either case, the review of the related literature describes and analy+es the published studies that are directly related to, andIor have so$e relevance to, the topic and research research questions at hand( Related literature should be integrated with, and weaved into, the $aterial in this section and not be si$ply cataloged( The review could conclude with a brief su$$ary of the literature and its i$plications(
86
The studys theoreticalIconceptual $odel and its hypotheses are developed based on the researchers logical reasoning as well as the i$plications of hisIher literature review( /ote that the studys hypotheses should be stated in a language consistent with its proposed conceptual fra$ewor* and the literature review- they should should not be stated in the null and alternate alternate hypotheses for$s( lso, a well# developed hypothesis is testable- that is, it can be confir$ed or disconfir$ed through the collection and analysis of data( ethodo dollogy ogy sect ection incl ncludes udes a des descri criptio ption n of the res resear earch METHODO1O6( The $etho sa$p sa$ple le 1sub 1subje ject cts2 s2,, data data colle collect ctio ion n $eth $ethod, od, $eas $easure ure$e $ent nt inst instru ru$e $ent nts, s, and and data data analy analysi siss procedures( The description of sa$pleIsubjects includes not only the sa$ple si+e and statistics regarding the subjects but also a definition and description of the population fro$ which the sa$ple was selected( This section also also describes the $ethod $ethod used in selecting the sa$ple or sa$ples( sa$ples( 5n the case of questionnaire surveys, infor$ation on response rates also should be provided( The description of instru$ents should identify and briefly describe all instru$ents used to collect data data pertin pertinent ent to the study study,, be they they tests, tests, questio questionnai nnaires res,, interv interview iew or obs observ ervati ation on for$s for$s,, or unobtrusive data such as absenteeis$ reports or productivity figures( figures( )hen possible, infor$ation infor$ation on validity validity and reliabili reliability ty of the $easures $easures used should be reported( reported( lso, lso, sources should be cited for $easure$ent $easure$ent instru$entsIp instru$entsIprocedur rocedures es 1e(g(, scales2 developed by other researchers( researchers( The $ethod section is usually concluded with a few state$ents about the analysis procedures utili+ed to test the studys hypotheses( section to both present present and discuss the RESU1TS DISCUSSION ( So$e authors use a single section data analysis analysis results( Others deal with the two issues in two separate separate sections( 5n either case, the statis statisti tical cal techniq techniques uestha thatt were were applie applied d to the data data $us $ustt be $entio $entioned ned and the result resultss of each analysis su$$ari+ed, su$$ari+ed, tabulated, and then discussed( %or each research hypothesis, the statistical statistical test of signif significa icance nce select selected ed and applied applied to the data is briefly briefly describe described, d, follow followed ed by a state$ state$ent ent indicating indicating whether the hypothesis hypothesis was supported or not supported( supported( Tables Tables and figures are used to present analyses results in su$$ary andIor graph for$ and to add clarity to the presentation( Cood tables and figures are uncluttered, self#e'planatory, and non#redundant( 5n addition to si$ply presenting the results in a straightforward $anner, the author also has to provide the readers with hisIher interpretation of the results, results, i$plications of the findings, conclusions and reco$$endations( &ach result is discussed in ter$s of the original hypothesis to which it relates and in ter$s ter$s of its agree$en agree$entt or disagr disagree$e ee$ent nt with with result resultss obtain obtained ed by other other resear researche chers rs in si$ila si$ilarI rIrel relate ated d studie studies( s( 5f the result resultss are consis consisten tentt with with the theoret theoretica icall $od $odel, el, resear researcher chers s e'pectations, andIor findings of other researchers, e'planations $ust be provided as to what the results results $ean and what their theoretical theoretical and practical i$plication i$plicationss are( )hen the results do not support the hypotheses andIor contradict previous findings, not only their $eaning, but also possible reasons for the discrepancies $ust be discussed( Often Often during during a study study apparen apparentt andIor andIor intere interesti sting ng relati relations onship hipss will will be noticed noticed that that were were not hypot hypothe hesi si+e +ed d by the the rese resear arch cher er(( Th Thes esee unfo unfore rese seen en resu result ltss shoul should d be ac*no ac*nowl wled edged ged and and discussed( Such results often for$ the basis for future studies specifically designed to e'a$ine the issue $ore carefully( %inally, the researcher should address the studys li$itations and $a*e reco$$endations for future research( 5t is notable that in the discussion portion portion of this section the researcher is often per$itted
87
$ore $ore freedo$ freedo$ to e'pres e'presss opinio opinions ns and reason reasonabl ablee specul speculati ations onsIas Iasser serti tions ons that that $ay be rather rather indirectly and i$plicitly based on data analysis results( si$ilar to the abstract section section e'cept that SUMMAR AND CONC1USIONS ( This section is very si$ilar it appears at the end of the the report 1preceding the R&%&R&/0& section2( section2( 5t su$$ari+es the the studys findin findings gs in an easy to underst understand and $anner $anner(( 5t also also e'plain e'plainss the practi practical cal i$plicat i$plication ionss of those those findings, and points to reco$$ended directions for future research in that area( RE0ERENCES ( The references section, or bibliography, lists all the sources, alphabetically by authors authors last na$es, that were directly used in writing writing the report( report( &very source cited in the paper $ust be included in the references, and every entry listed in the references $ust appear in the paper( Style $anuals, such as the P 1$erican Psychological ssociation2 $anual, will give you the correct procedure for all in#te't and reference reference citations( This for$ is usually different for boo*s, journal articles, and $aga+ine articles( 5t is reco$$ended that you use the P P style( 5t is i$portant that whatever for$ is used be followed consistently( consistently( APPENDIQES ( ppendi'es include infor$ation and data pertinent to the study that either are not i$portant i$portant enough to be included included in the $ain body of the report report or are too lengthy( lengthy( ppendi'es ppendi'es contain such entries as $aterials especially developed for the study 1e(g(, tests, questionnaires, and cover cover letter letters2, s2, coding coding sche$e sche$e,, print print out of raw data, data, and the co$pute co$puterr print# print#out out of statis statistic tical al analyses(
=i!liography =i!liography :
bibliography can be defined as, “8 list of refere reference nce materials materials 9involvin 9involvin any /ind of content : text" music" music" paintins" video etc.; elucidatin the type" nature and other detailed information on the basis of name" date" place and enre of the materials.” )5 “8 complete cateorical compilation of any type of content based on it’s it’s creator9s;" creator9s;" editors and time 9of production" distribution;.”
also *now *nown n as &ors basica call lly y an orde orderl rly y stud study y and and Bi!liograp8y" also &ors 7ite 7ite"" re$er re$eren en7e 7e list list is basi referencing of boo*s and source $aterials used in acade$ic research( 5t $ight or $ight not include any infor$ation on the literary analysis or criticis$ of the $aterials cited( Importance and Use of Bibliographies: $andatory $andatory require$ent of copyright copyright laws and acade$ic conventions is that whenever a research research paper is written, there should be a section at the end of it where you ac*nowledge the sources used( So, bibliography $eans listing all the sources which you have consulted while writing your essay or research article( The sources $ay be in the for$ of printed and online boo*s, websites, web 88
docu$ents, web blogs, newspaper articles, journals, pod casts, wi*is, unpublished $aterial, $aps etc( 0itation ensures that the infor$ation contained in the research paper is based on logic, truth and facts( bsence bsence of references or bibliography indicates that the paper $ay be a piece of plagiaris$(
GGGGG
89
