Determining How to Select a Sample Edited & Complied By
Sanjeev S. Malage Associate Professor FMS Department , NIFT, Bangalore
Learning Objectives 1. To underst understand and the conce concept pt of sampli sampling. ng. 2. To learn the steps in developing developing a sampling sampling plan. plan. 3. To understand the concepts of sampling error and nonsampling error. 4. To distinguish between probability samples, and nonprobability samples. 5. To understand understand sampling sampling implicat implications ions of surveying surveying over over the Internet. Sanjeev Sadashiv Malage NIFT, Bangalore
2
Definition of sampling
Procedure by which some members of a given population are selected as representatives of the entire population
Sanjeev Sadashiv Malage NIFT, Bangalore
3
The Concept of Sampling
To understand the concept of sampling.
Ø
Sampling Defined:The Defined:The process of obtaining information from a subset of a larger group.
Ø
A market researcher takes the results from the sample to make estimates of the larger group.
Ø
Sampling a small percentage of a population can result in very accurate estimates.
Sanjeev Sadashiv Malage NIFT, Bangalore
4
Why do we use samples ?
Get information from large populations – At minima minimall cos costt – At max maximu imum m speed speed – At increa increased sed accur accuracy acy – Using Using enhan enhanced ced too tools ls
Sanjeev Sadashiv Malage NIFT, Bangalore
5
What we need to know
• Concepts – Repr Repres esent entat ativ ivene eness ss – Samp Sampliling ng met metho hods ds – Choice Choice of the the right right desi design gn
Sanjeev Sadashiv Malage NIFT, Bangalore
6
Sampling and representativ representativeness eness
Sampling Population
Sample
Target Population
Target Population
Sampling Population
Sanjeev Sadashiv Malage NIFT, Bangalore
Sample 7
Steps in Developing a Sample Plan
Step 7. Execute Operational Plan
Step 2. Choose Data Collection Method Step1. Define the Population of Interest
Step 6. Develop Operational Plan
Step 5. Determine Sample Size
Step 3. Choose Sampling Frame
(4) Select a Sampling Method
Sanjeev Sadashiv Malage NIFT, Bangalore
8
Steps In Developing A Sampling Plan
To learn the steps in developing a sample plan.
Step One: Defining the Population of Interest Specifying the characteristics from whom information is needed. Define the characteristics of those that should be excluded. Step Two: Choose Data Collection Method Impacts for the sampling process. Step Three: Choosing Sampling Frame A list of elements or members from which we select units to be sampled. Sanjeev Sadashiv Malage NIFT, Bangalore
9
Basic Concepts in Sampling • Population: the entire group under study as defined by research objectives – Researchers Researchers define define populations populations in specific specific terms terms such such as “heads “heads of of households located in areas served by the company who are responsible for making the decision.”
10 Sanjeev Sadashiv Malage, NIFT, Bangalore
Basic Concepts in Sampling • Sample: a subset of the population that should represent the entire group • Sample unit: the basic level of investigation • Census: an accounting of the complete population
11 Sanjeev Sadashiv Malage, NIFT, Bangalore
Steps In Developing A Sampling Plan
To learn the steps in developing a sample plan.
Step Four: Select a Sampling Method The selection will depend on: • The objecti objectiv ves of of the the study study • The finan financial cial resourc resources es availab available le • Tim Time limi limita tati tion ons s • The The natur nature e of the pro proble blem m
12 Sanjeev Sadashiv Malage, NIFT, Bangalore
Steps In Developing A Sampling Plan Step Five: Determine Sample Size • Availa Available ble budget budget • Rules of thumb thumb Step Six: Develop of Operational Procedures for Selecting Sample Elements Specify whether a probability or nonprobability nonprobabil ity sample is being used Step Seven: Execution the Sampling Plan The final step of the operational sampling plan Include adequate checking of specified procedures. 13 Sanjeev Sadashiv Malage, NIFT, Bangalore
Classification of Sampling Methods
Sampling methods
Probability samples
Nonprobability samples
Simple random
Systematic
Cluster
Stratified
Convenience
Judgement
Snowball
Quota 14
Sanjeev Sadashiv Malage, NIFT, Bangalore
Two Basic Sampling Methods • Prob Probab abil ilit ity y sa sam mpl ples es:: ones in which members of the population have a known chance (probability) of being selected into the sample • NonNon-pr prob obab abililit ity y sa samp mple les: s: instances in which the chances (probability) of selecting members from the population into the sample are unknown 15 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling: Simple Random Sampling • Si Simp mple le ra rand ndom om sa samp mpliling ng:: • the the prob probabi abilit lity y of of being being selec selecte ted d into into the sample is “known” and equal for all members of the population – E.g., Blind Draw Method Method – Random Numbers Method 16 Sanjeev Sadashiv Malage, NIFT, Bangalore
Simple random sampling
• P r i n ci p l e –Equal chance of drawing each unit
• Procedure –Numbe – Numberr all units units –Rando – Randomly mly draw units 17 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling: 1.Simple Random Sampling – Advan Advanta tage ge:: • Known Known and equal chance chance of selection selection
– Disadvan Disadvantage tages: s: • Complete Complete accounting accounting of population population needed • Cumber Cumbersom some e to provide provide uniqu unique e designations to every population member 18 Sanjeev Sadashiv Malage, NIFT, Bangalore
Simple random sampling Example: Example: evaluate evaluate the the prevalence prevalence of tooth tooth decay among the 1200 children attending a school • List List of of chil childr dren en att atten endi ding ng the the scho school ol • Chil Childr dren en num numerat erated ed from from 1 to to 120 1200 0 • Samp Sample le size size = 100 100 chil childr dren en • Rand Random om sam sampl plin ing g of 100 100 numb number ers s between 1 and 1200 How to randomly select? 19 Sanjeev Sadashiv Malage, NIFT, Bangalore
Simple random sampling
20 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Systematic Sampling • Syst System emat atic ic sa samp mpli ling ng:: way to select a random sample from a directory or list that is much more efficient than simple random sampling – Skip interval=po interval=population pulation list size/sample size
21 Sanjeev Sadashiv Malage, NIFT, Bangalore
Systematic sampling • N = 1200, and n = 60 sampling fraction = 1200/60 = 20 • List List pers perso ons from from 1 to to 120 1200 0 • Rand Randomly omly selec selectt a number number betw between een 1 and 20 (ex : 8) st person selected = the 8th on Þ 1 the list nd person = 8 + 20 = the 28th Þ 2 etc ..... 22 Sanjeev Sadashiv Malage, NIFT, Bangalore
Systematic sampling
23 Sanjeev Sadashiv Malage, NIFT, Bangalore
1
2
3
4
5
6
16 17 18 19 20
31 32
33 34 35
46 47 48 49
7
8
9
21 22 23 24
10 11 12 13 14 15
25 26 27 28 29
36 37 38 39 40 41 42
50 51 52 53 54 55
30
43 44 45
……..
24 Sanjeev Sadashiv Malage, NIFT, Bangalore
Systematic sampling
25 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Systematic Sampling – Advan Advanta tages ges:: • Approx Approxima imate te known known and and equal equal chance of selection…it is a probability sample plan • Efficiency Efficiency…do …do not not need to to designate designate every population member • Less Less expensi expensive… ve…fas faster ter than than SRS
– Disadvan Disadvantag tage: e: • Small Small loss loss in sampli sampling ng precisi precision on 26 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Cluster Sampling • Clu lus ste terr sam sampl plin ing g: method in which the population is divided into groups, any of which can be considered a representative sample – Area samplin sampling g
27 Sanjeev Sadashiv Malage, NIFT, Bangalore
Cluster sampling • Principle – Random sample sample of groups groups (“clusters (“clusters”) ”) of units – In selected selected clusters, clusters, all all units or proportion (sample) of units included
28 Sanjeev Sadashiv Malage, NIFT, Bangalore
Cluster Sampling • In clust cluster er sam sampl pling ing the the popu populat lation ion is divided into subgroups, called “clusters.” • Each Each clust cluster er should should repres represen entt the the population. • Area Area sam sampli pling ng is a form form of clust cluster er samplin sampling g – the geogra geographic phic area is divided into clusters. 29 Sanjeev Sadashiv Malage, NIFT, Bangalore
Cluster Sampling • One One clu clust ster er may may be sel selec ecte ted d to to represent the entire area with the one-step area sample. • Seve Severa rall clus cluste ters rs may may be sel selec ecte ted d using the two-step area sample.
30 Sanjeev Sadashiv Malage, NIFT, Bangalore
A Two-Step Cluster Sample • A two two-st -step ep clust cluster er sample sample (samp (samplin ling g several clusters) is preferable to a one-step (selecting only one cluster) sample unless the clusters are homogeneous.
31 Sanjeev Sadashiv Malage, NIFT, Bangalore
Example: Cluster sampling Section 1
Section 2
Section 3
Section 5 Section 4 32 Sanjeev Sadashiv Malage, NIFT, Bangalore
cluster sampling To evaluate vaccination coverage: • Witho ithout ut list list of pers person ons s • Tota Totall popu popula lati tion on of of vil villa lage ges s • Rand Random omly ly choo choose se 30 clu cluster sters s • 30 clus cluste terr of of 7 chi child ldre ren n eac each= h= 210 210 chil childr dren en
33 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Cluster Sampling – Advan Advanta tage ge:: • Economic Economic efficie efficiency… ncy…fa faste sterr and less expensive than SRS – Disadvan Disadvantag tage: e: • Cluster Cluster specifica specification tion error… error…the the more homogeneous the clusters, the more precise the sample results 34 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling • When When the the res resea earc rche herr know knows s the the answers to the research question are likely to vary by subgroups…
35 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Stratified Sampling • Stra Strati tifi fied ed sa samp mpli ling ng:: method in which the population is separated into different strata and a sample is taken from each stratum – Proportionate Proportionate stratified stratified sampl sample e – Disproportionate Disproportionate stratified stratified samp sample le
36 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified sampling • Principle : –Classify population into internally homogeneous subgroups (strata) – Draw sample sample in each each strata – Combine Combine results results of all strata
37 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling – Research Question Question:: “To what what extent do you value your college degree?” Answers are on a five point scale: 1= “Not valued at all” and 5= “Very highly valued” • We would would expect expect the answer answers s to vary vary depending on classification. Freshers are likely to value less than Alumni. We would expect the mean scores to be higher as classification goes up. 38 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling – Research Question Question:: “To what what extent do you value your college degree?” • We would would also also expect expect there there to be be more more agreement (less variance) as classification goes up. That is, seniors should pretty much agree that there is value. Freshers will have less agreement.
39 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling • Why Why is str strat atif ifie ied d samp sampliling ng mor more e accurate when there are skewed populations? – The less varianc variance e in a group, group, the the less less sample size it takes to produce a precise answer. – Why? If If 99% of of the population population (low (low variance) agreed on the choice of Brand A, it would be easy to make a precise pr ecise estimate that the population preferred Brand A even with a small sample size. 40 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling – But, if 33% chose chose Brand Brand A, and 23% chose B, and so on (high variance) it would be difficult to make a precise estimate of the population’s preferred brand…it would take a larger sample size…
41 Sanjeev Sadashiv Malage, NIFT, Bangalore
Stratified Sampling – Stratified Stratified sampling sampling allows allows the researcher to allocate more sample size to strata with less variance and less sample size to strata with less variance. Thus, for the same sample size, more precision is achieved. – This is is normally normally accomplished accomplished by disproportionate sampling. Seniors would be sampled LESS than their proportionate share of the population and freshmen would be sampled more. 42 Sanjeev Sadashiv Malage, NIFT, Bangalore
Probability Sampling Stratified Sampling – Advan Advanta tage ge:: • More accurate accurate overal overalll sample sample of of skewed population…see next slide for WHY
– Disadvan Disadvantag tage: e: • More More comple complex x sampli sampling ng plan plan requiring different sample size for each stratum
43 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • With nonprobability sampling methods selection is not based on fairness, equity, or equal chance. – Convenie Convenience nce sampling sampling – Judgme Judgment nt sampling sampling – Referra Referrall samplin sampling g – Quota Quota samp sampling ling 44 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • May May not not be repre represen senta tati tive ve but but the they y are still used very often. Why? – Decision makers want fast, relatively inexpensive answers… nonprobability samples are faster and less costly than probability samples.
45 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • May May not not be repre represen senta tati tive ve but but the they y are still used very often. Why? – Decision Decision maker makers s can make make a decision based upon what 100 or 200 or 300 people say…they don’t feel they need a probability sample.
46 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • Conv Conven enie ienc nce e sa sam mpl ples es:: samples drawn at the convenience of the interviewer – Error occurs occurs in the form form of members of the population who are infrequent or nonusers of that location
47 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • Judgm gme ent sample les s: samples that require a judgment or an “educated guess” as to who should represent the population – Subjectivity Subjectivity enters enters in here, here, and certain members will have a smaller chance of selection than others 48 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • Referr rra al sam amp ple les s (snowball samples): samples which require respondents to provide the names of additional respondents – Members of the population population who are less known, disliked, or whose opinions conflict with the respondent have a low probability of being selected 49 Sanjeev Sadashiv Malage, NIFT, Bangalore
Nonprobability Sampling • Quota samples: samples that use a specific quota of certain types of individuals to be interviewed – Often used to to ensure ensure that convenience samples will have desired proportion of different respondent classes
50 Sanjeev Sadashiv Malage, NIFT, Bangalore
Online Sampling Techniques • Rando Random m onl onlin ine e int interc ercep eptt sam sampli pling ng:: relies on a random selection of Web site visitors • Invi Invita tati tion on on onliline ne sa samp mpliling ng:: is when potential respondents are alerted that they may fill out a questionnaire that is hosted at a specific Web site
51 Sanjeev Sadashiv Malage, NIFT, Bangalore
Online Sampling Techniques • Onl nlin ine e pan panel el sa samp mpli ling ng:: refers to consumer or other respondent panels that are set up by marketing research companies for the explicit purpose of conducting surveys with representative samples
52 Sanjeev Sadashiv Malage, NIFT, Bangalore
Basic Concepts in Sampling • Sampling er error: any error in a survey that occurs because a sample is used • A sample frame: a mast master er list list of the the entire population • Sam Sampl ple e fr fram ame e er erro ror: r: the degree to which the sample frame fails to account for all of the population…a telephone book listing does not contain unlisted numbers 53 Sanjeev Sadashiv Malage, NIFT, Bangalore