Sampling

1
Sampling Study Design

• Lecture Notes
• February 14, 2008
• Dr. Sinn

2
Types of Studies

• Observational
• None of the study variables are altered by the
  researcher
• Experimental
• Some variable is altered
• Ethical Experiments with Human Subjects can be
  difficult (smoking).

3
Individuals

• The individuals in a survey study are the
  people or things the survey is asking about.
• Examples
• Researchers call 500 SUV owners and ask them
  about maintenance costs for the SUV.
• Researchers call 250 homeowners and ask about
  their refrigerators (reliability, satisfaction,
  etc)
• Political pollsters call 800 registered voters to
  ask about social security reform.
• Dr. Sinns research (parent-student pairs)

4
Individuals

• In survey research, individuals may be
• Groups (families, couples, households, etc)
• People
• Things (refrigerators, cars, lawns, etc)
• Communities (best places to live in USA)
• The individual in a research study is the thing,
  person or idea being studied, not necessarily the
  person asked the questions.

5
Types of Sampling

• Random
• Numbered Frame Random Number Generator
• Systematic
• Every nth individual
• Stratified
• Randomly sample strata (democrats, republicans,
  independents, etc)
• Cluster
• Sample 100 of sub-group(s)
• Convenience
• Volunteers (all research done with human
  populations)

6
Representative

• It matters not HOW the data were collected
• Is your sample representative of the population?
• If not, what sub-population is your sample
  representative of?
• Example Battle-ground state pre-election polling
  in October, 2004.

7
Generating Random Samples

• Random is a mathematical term (use with care!)
• To have a truly random sample, we must
• Use a Randomization Machine (TI-83)
• Have a complete list of population (the Frame)
• All selected participants must participate
• Ex The mayor of a town wants to poll the 5,832
  residents and gives you a list.
• Obtain a random sample of size 20

8
Example

• You are planning a report on apartment living in
  a college town. You decide to select three
  apartment complexes at random for in-depth
  interviews with residents. Select a simple random
  sample of three of the following.

9
Non-sampling Error (and ways to combat)

• Problem Incorrect Frame
• Fix Insure representative sample
• Problem Non-Response
• Fix Use call-backs, rewards
• Problem Poor Interviewer Skills
• Fix Training
• For your projects, ALL team members must ask
  exact same questions!!
• Problem Faulty Data Entry
• Fix Data checks

10
Non-sampling Error (and ways to combat)

• Problem Survey Problems
• Fix Never use a 1st draft survey!!
• Problem Questionnaire Design
• Fix Thoughtfully choose open vs. closed
  questions to get useable data (categories)
• Problem Question Wording
• Fix Plain language, no heat
• Problem Order of questions/words/responses
• Fix Ask guilt questions last
• Problem Sensitive Subjects/Privacy
• Fix Anonymity, careful wording of questions,
  voluntary!