Sampling Methods I

29
1 Sampling Methods Probability , Non Probability and Mixed Probability Sampling Procedures

description

how to sample things

Transcript of Sampling Methods I

Page 1: Sampling Methods I

1

Sampling Methods

Probability , Non Probability and Mixed Probability Sampling Procedures

Page 2: Sampling Methods I

2

What is Sampling

Sampling is simply the process of learning about the population on the basis of sample drawn from it.Sampling may be defined as selection of some part of an aggregate or totality on the basis of which a judgement or inference about the aggregate or totality is made.

Page 3: Sampling Methods I

3

Key Terms

Mean,S.D,Correlation

Statistics

Parameter

Mean,S.D,Correlation

Population

Sample

Page 4: Sampling Methods I

4

Key Terms

1. Sample

2. Population

3. Sampling frame

4. Sample statistics

5. Population parameter.

6. Estimator.

7. Estimate.

Page 5: Sampling Methods I

5

Advantages of Sampling over Advantages of Sampling over Complete CensusComplete Census

1.1. Less Time.Less Time.

2.2. Reduced Cost of Survey.Reduced Cost of Survey.

3.3. Greater Accuracy of Results.Greater Accuracy of Results.

4.4. Greater Scope (statistical analysis).Greater Scope (statistical analysis).

5.5. Larger Population.Larger Population.

6.6. Hypothetical Population.Hypothetical Population.

7.7. Testing is Destructive.Testing is Destructive.

Page 6: Sampling Methods I

6

Example I

A manufacturer of electric tubes is interested in knowing the average life of its product.Then it is desirable to take a sample of some tubes and perform experiments on tubes to know the average life of tubes .It is not wise to display all tubes .Infact this is an example where only sampling can be used.

Page 7: Sampling Methods I

7

Advantages of Sampling over Advantages of Sampling over Complete CensusComplete Census

Prof.R.A Fisher (1950) in a report on “The Prof.R.A Fisher (1950) in a report on “The Subcommission on Statistical Sampling of Subcommission on Statistical Sampling of United Nations”sums the advantages of United Nations”sums the advantages of sampling techniques over complete census in sampling techniques over complete census in the following four words.the following four words.Adaptability.Adaptability.Speed.Speed.Economy.Economy.Scientific Approach.Scientific Approach.

Page 8: Sampling Methods I

8

Sampling Errors and Non-Sampling Errors

Sampling Errors have their origin in Sampling and arise due to the fact that only a part of Population has been used to estimate the Population Parameters and draw inferences about the Population.As such Sampling Errors are absent in a complete Enumeration Survey.

Page 9: Sampling Methods I

9

Sampling Errors and Non-Sampling Errors

Data obtained in a Complete Census,although free from sampling Errors would still be subjected to Non-Sampling Errors whereas data obtained in a sample survey would be subjected to both Sampling and Non-Sampling Errors.

Page 10: Sampling Methods I

10

Sources of Non-Sampling Errors

1. Faulty Planning or Definitions.

2. Response Errors.

3. Non-Response Errors or Bias.

4. Errors in Coverage.

5. Compiling Errors.

6. Publication Errors.

Page 11: Sampling Methods I

11

The Sampling Process

1. Define population .

2. Determine the Sampling Frame.

3. Sample Design.

4. Determine the Sample Size.

5. Execute the Sampling Process.

Page 12: Sampling Methods I

12

Sample Designs

Designs

Probability Samples

Non-Probability Samples

Mixed ProbabilitySamples

Page 13: Sampling Methods I

13

Lower class Upper class

People who have own houses/Cars/posh colonies

BiasedSample

Population

Lower Class Upper Class

Unbiased Sample

Population

Unbiased, representative sample drawn at random from the entire population.

Biased, unrepresentative sample drawn from people who have cars /own houses/posh colonies.

Sample survey –Standard of living in Delhi

Page 14: Sampling Methods I

14

Probability versus Non-Probability Samples

Probability Samples (Statistical Samples)1. Representative Samples.2. Each element of the population has the

known probability of being included in the sample.

3. A chance mechanism is used in the selection process.

4. Eliminates bias in the selection process.

Page 15: Sampling Methods I

15

Probability versus Non-Probability Samples

Non-Probability Samples (Non- Statistical Samples)

1. Use some non-random methods to choose items.

2. Open to selection Bias.

3. Do not make use of appropriate data collection methods.

Page 16: Sampling Methods I

16

Probability Samples

1. Simple Random Sample.

2. Stratified Sample.

3. Systematic Sample.

4. Cluster Sample.

5. Area Sample.

6. Multistage Sample.

7. Sequential Sample.

Page 17: Sampling Methods I

17

Simple Random Sample

I. Number each item unit from 1 to N.

II. Use a random number table or a random number generator to select n distinct numbers b/w 1 and N inclusively.

III. Notation: n = no. of items in Sample.

IV. N = no. of items in Population.

Page 18: Sampling Methods I

18

Stratified Sample

I. A population is divided into subgroups called strata based on some criteria eg.Geographic Location,Gender,Age,Income,Profession.

II. A random sample is selected from each stratum (Items within strata are relatively homegeneous).

III. Reasons for stratifyingA. Ensure representation of subgroups.B. Compare Subgroups.

Page 19: Sampling Methods I

19

Stratified Random Sample

Population of Radio FM listeners stratified by age

1. 20 – 30 Years old. ( Alike) Strata 1

2. 30 – 40 Years old. ( Alike) Strata 2

3. 40 – 50 Years old. ( Alike) Strata 3

Heterogeneity b/w 1and 2, 2 and 3,1 and 3.

Homegeneity within groups.

Page 20: Sampling Methods I

20

Cluster Sampling

Population is divided into non-overlapping clusters or areas often based on geographical location.

I. Each cluster is a miniature ,or microcosm of the population.

II. A subset of clusters is selected randomly for the sample.

III. All items in clusters are enumerated.

Page 21: Sampling Methods I

21

Example of Cluster Sampling

Cluster sampling is frequently used in consumer surveys in rural areas.We select a few villages randomly and include every household in the selected villages in our sample.A village can be regarded as the natural geographic cluster of households.

Page 22: Sampling Methods I

22

Systematic Sample

Determine Sampling Interval (N/n)

Population : 500 MBA Students.

Sample : 50 MBA Students.

N/n = 500 = 10

50

Draw a random number b/w 1 – 10

(Say number is 2 )

Include students numbered 2,12,22,--------.

Page 23: Sampling Methods I

23

Non-Probability Samples

The selection of units within a sample involves human judgement rather than pure chance.The maximum information available “per rupee” which can be determined from a probability sample is not possible in this case and moreover, the degree of accuracy is not known.

Page 24: Sampling Methods I

24

Non-Probability Samples

I. Convenience Sampling: Sample elements are selected for the convenience of the researcher.

II. Quota Sampling:Sample elements are selected until the quota controls are satisfied.

III. Judgement Sampling:Sample elements are selected by the judgement of the researcher.

IV. Volunteer Sampling.

Page 25: Sampling Methods I

25

Example of Convenience Sampling

Suppose a marketing research study aims at estimating the proportion of shops which store Daburs Real Juices.Suppose a sample of 100 is needed .Then the researcher may go to the shops according to his convenience.

Note: This is definitely not a representative sample.

Page 26: Sampling Methods I

26

Examples

I. “People- on- the- street interviews.

II. The first 100 customers to enter a departmental store.

III. The first three callers in a radio contest

Page 27: Sampling Methods I

27

Example of Quota Sampling

We have a population of three categories,i.e.,Rich,Medium,Poor.These are given in the ratio 20%,30%,50% respectively.Using quota sampling determine the number of representatives for each category of a sample size of 200.Solution: Category

1. Rich = 20% of 200 = 40 2. Medium = 30% of 200 = 60 3. Poor = 50% of 200 = 100

Page 28: Sampling Methods I

28

Deliberate or Purposive or Judgemental Sampling

I. Statisticians often use this method for in exploratory studies like pretesting of questionnaires and focus groups.

II. Medical research where choice of experimental subjects(animals,human beings, vegetable) reflects the investigators pre-existing beliefs about the population.

III. Expert interviews, panel discussions.

Page 29: Sampling Methods I

29

Volunteer Sampling

A. Pre Poll survey (news channel).B. Post Poll surveys.C. Television and radio often use call-in

polls.D. In psychological experiments or

pharmaceuticals trials(drug testing), it would be difficult and unethical to enlist random participants from general public.