8/10/2019 Lecture NotesNota 7 Sampling

1/22

Sampling

8/10/2019 Lecture NotesNota 7 Sampling

2/22

Steps in determining sampling

Identify unit of analysisAt what level the data needs to be gathered

May be multi level or single

Specify population and sample Representativeness

Populationgroup of individuals who have the samecharacteristics

Target populationgroup of individuals with somecommon defining characteristics

Samplesubgroup of the target population

8/10/2019 Lecture NotesNota 7 Sampling

3/22

Samplingstrategies

Probability

sampling

Nonprobability

sampling

Simple

Random

sampling

Convenience

sampling

Multistage

Cluster

sampling

Stratified

sampling

Snowball

sampling

8/10/2019 Lecture NotesNota 7 Sampling

4/22

Simple random sampling

Each element in the population has anequal probability of selection AND each

combination of elements has an equalprobability of selection

Names drawn out of a hat

Random numbers to select elements froman ordered list

8/10/2019 Lecture NotesNota 7 Sampling

5/22

Stratified sampling

Divide population into groups that differ inimportant ways

Basis for grouping must be known beforesampling

Select random sample from within each

group

8/10/2019 Lecture NotesNota 7 Sampling

6/22

Random Cluster Sampling

Done correctly, this is a form of randomsampling

Population is divided into groups, usuallygeographic or organizational

Some of the groups are randomly chosen In pure cluster sampling, whole cluster is

sampled. In simple multistage cluster, there is randomsampling within each randomly chosen cluster

8/10/2019 Lecture NotesNota 7 Sampling

7/22

Random Cluster Sampling

Population is divided into groups Some of the groups are randomly selected

For given sample size, a cluster sample hasmore error than a simple random sample

Cost savings of clustering may permit largersample

Error is smaller if the clusters are similar toeach other

8/10/2019 Lecture NotesNota 7 Sampling

8/22

Cluster sampling has very high error if theclusters are different from each other

Cluster sampling is NOT desirable if theclusters are different It IS random sampling: you randomly

choose the clusters

But you will tend to omit some kinds ofsubjects

8/10/2019 Lecture NotesNota 7 Sampling

9/22

Stratified Cluster Sampling

Reduce the error in cluster sampling bycreating strata of clusters

Sample one cluster from each stratum The cost-savings of clustering with the

error reduction of stratification

8/10/2019 Lecture NotesNota 7 Sampling

10/22

Stratification vs. Clustering

Stratification Divide population into

groups different fromeach other: sexes, races,ages

Sample randomly fromeach group

Less error compared tosimple random

More expensive to obtainstratification informationbefore sampling

Clustering Divide population into

comparable groups:schools, cities

Randomly sample someof the groups

More error compared tosimple random

Reduces costs to sampleonly some areas ororganizations

8/10/2019 Lecture NotesNota 7 Sampling

11/22

Stratified Cluster Sampling

Combines elements of stratification andclustering First you define the clusters

Then you group the clusters into strata of clusters,putting similar clusters together in a stratum

Then you randomly pick one (or more) cluster fromeach of the strata of clusters

Then you sample the subjects within the sampledclusters (either all the subjects, or a simple randomsample of them)

8/10/2019 Lecture NotesNota 7 Sampling

12/22

Multi-stage Probability Samples

Large national probability samples involveseveral stages of stratified cluster sampling

The whole country is divided into geographicclusters, metropolitan and rural

Some large metropolitan areas are selected withcertainty (certainty is a non-zero probability!)

Other areas are formed into strata of areas (e.g.middle-sized cities, rural counties); clusters areselected randomly from these strata

8/10/2019 Lecture NotesNota 7 Sampling

13/22

Within each sampled area, the clusters aredefined, and the process is repeated, perhapsseveral times, until blocks or telephoneexchanges are selected

At the last step, households and individualswithin household are randomly selected

Random samples make multiple call-backs topeople not at home.

8/10/2019 Lecture NotesNota 7 Sampling

14/22

The Problem of Non-Response

You can randomly pick elements from samplingframe and use them to randomly select people

But you cannot make people respond Non-response destroys the generalizeability ofthe sample. You are generalizing to people whoare willing to respond to surveys

If response is 90% or so, not so bad. But if it is50%, this is a serious problem

8/10/2019 Lecture NotesNota 7 Sampling

15/22

Multiple call-backs are essential for trying to reduce non-response bias

Samples without call-backs have high bias: cannot really

be considered random samples Response rates have been falling It is very difficult to get above a 60% response rate You do the best you can, and try to estimate the effect

of the error by getting as much information as possibleabout the predictors of non-response.

8/10/2019 Lecture NotesNota 7 Sampling

16/22

Non-probability Samples

Convenience Purposive

Quota

8/10/2019 Lecture NotesNota 7 Sampling

17/22

Convenience Sample

Subjects selected because it is easy toaccess them.

No reason tied to purposes of research. Students in your class, people on State,

Street, friends

8/10/2019 Lecture NotesNota 7 Sampling

18/22

Purposive Samples

Subjects selected for a good reason tied topurposes of research

Small samples < 30, not large enough for powerof probability sampling. Nature of research requires small sample

Choose subjects with appropriate variability in what

you are studying Hard-to-get populations that cannot be found

through screening general population

8/10/2019 Lecture NotesNota 7 Sampling

19/22

Quota Sampling

Pre-plan number of subjects in specifiedcategories (e.g. 100 men, 100 women)

In uncontrolled quota sampling, the subjects

chosen for those categories are a conveniencesample, selected any way the interviewerchooses

In controlled quota sampling, restrictions are

imposed to limit interviewers choice No call-backs or other features to eliminateconvenience factors in sample selection

8/10/2019 Lecture NotesNota 7 Sampling

20/22

8/10/2019 Lecture NotesNota 7 Sampling

21/22

Sample Size

Heterogeneity: need larger sample to studymore diverse population

Desired precision: need larger sample to get

smaller error Sampling design: smaller if stratified, larger if

cluster Nature of analysis: complex multivariate

statistics need larger samples Accuracy of sample depends upon sample size,not ratio of sample to population

8/10/2019 Lecture NotesNota 7 Sampling

22/22

Sampling in Practice

Often a non-random selection of basic samplingframe (city, organization etc.)

Fit between sampling frame and research goals

must be evaluated Sampling frame as a concept is relevant to all

kinds of research (including nonprobability) Nonprobability sampling means you cannot

generalize beyond the sample Probability sampling means you can generalizeto the population defined by the sampling frame