Sunday, February 27, 2011

De Vaus: Chapter 6: Finding a Sample


 

Some Sampling Concepts

-          Fundamental goal of research à generalize from sample to wider population

-          Statistical generalization – use probability theory to estimate likelihood that patterns observed in sample will hold in population (relies on random samples)

-          Replication – test generalisability – test in different situations

-          Census

-          Sample

-          Sampling frame  - list of population elements

-          Representative sample – profile of sample same as population

-          Weighting

-          Sampling error

 

 

Types of Probability Samples

 

-          Simple random sampling

o   Complete sampling frame

o   Requires good sampling frame, and population is geographically concentrated or data collection method does not require travelling

o   Cost

-          Systematic sampling

o   Similar to simple random sampling

o   Randomly chose start point, then choose nth unit

o   Problem – periodicity of sampling frame – eg, husband & wife

-          Stratified sampling

o   Select stratifying variable

o   Divide sampling frame into separate lists

o   Randomly close sample from each list

-          Multistage cluster sampling

o   Final sample involves several different samples

§  Eg

·         Divide city into areas (clusters)

·         Select sample of areas

·         Divide clusters into smaller units – sample

·         For each smaller unit – list addresses – sample

·         At each selected address – select individual

o   Maximize number of initial clusters à increase chance of representativeness – but this increases sampling costs

o   Use stratification techniques

 

 

Internet Samples

 

-          E-mail surveys

-          Web page based surveys

o   Pop ups

o   Advertising on other sites

o   E-mail invitations

o   Panels / representative panels

 

http://statisticsandastudent.blogspot.com/2011/02/aapor-report-on-on-line-panels.html

 

-          Internet samples and representativeness – 3 ways to use internet to gain representative general population samples

o   Connect a random sample to internet

o   Multimode methods of questionnaire administration

§  Complex

§  Mode effects

o   Using quota internet samples

§  Make a sample that is representative in specific respects

o   Use of internet samples

 

Sample sizes

 

-          Degree of accuracy

-          Extent of variation in population

-          Relationship between sample size and accuracy

o   Small samples à small increase in sample size can lead to substantial increase in accuracy

o   Size of population from which we draw sample is irrelevant

o   Heterogeneous population – smaller sample à ie 90% voting one way vs 50/50 split

§  But allow flexibility in survey purpose, and go for larger sample sizes

o   Sample sizes of sub groups

§  Sample size and variation within each group should determine sample size of each group

 

Non Response

 

-          Problems

o   Unacceptable reduction of sample size

o   Bias

o   Techniques to reduce non response may not avoid bias problem

§  Often non responders are different in crucial respects from responders

·         Use available information on non responders

·         Some sampling frames can provide useful information

o   Official records

·         Compare with known characteristics of population – ie census balanced

 

http://statisticsandastudent.blogspot.com/2011/02/chapter-1-introduction-to-survey-errors.html

 

Weighting Samples

 

-          Due to

o   Non response

o   Deliberate oversampling

o   Inadequate sampling frames

-          Purpose – adjust sample so that sample profile on key variables reflects that of population

-          Statistically increasing / decreasing 'numbers' of cases with particular characteristics so proportion of cases in sample is adjusted to population proportion.

-          How to weight a sample on single characteristic

o   Example

§  Select variable – gender

§  Obtain population and sample percentages

§  Calculate weight for each category

·         Ie male

o   Sample 35%

o   Population 50%

·         Male weight = 50 / 355 – 1.43  

-          How to weight a sample on two / more characteristics

 

 

Secondary Analysis

 

 

 

Non Probability Sampling

 

-          Eg, to sample gays

-          Purposive sampling – cases selected by researcher

-          Quota sampling

-          Availability samples

 

 

Sampling Checklist

 

 

 

 

 

GW Comments

 

-          How does stratified sampling differ from blocking

-          Cluster sampling – see article on fathers & un wed births.

http://statisticsandastudent.blogspot.com/2011/02/fragile-families-sample-design.html

 

-          Get notes from Anova course on sample size

-          Sources for : Heterogeneous population – smaller sample à ie 90% voting one way vs 50/50 split

-          Non response – see notes

 

http://statisticsandastudent.blogspot.com/2011/02/chapter-1-introduction-to-survey-errors.html

 

 

 

No comments: