Chapter 9 Survey Methodology Basic survey language Types
Chapter 9 Survey Methodology • • • Basic survey language Types of surveys The survey schedule Survey designs Sampling error, confidence levels, and sample size © 2017 Taylor & Francis
Basic Survey Language • In research, the term survey refers to a sample survey • A sample survey has scientific utility © 2017 Taylor & Francis
Types of Surveys Four basic types of ways to collect survey data: 1. 2. 3. 4. In person On the phone Through the mail On the internet © 2017 Taylor & Francis
The Survey Schedule (Questionnaire) • The interviewer/survey script • Length • Writing the schedule © 2017 Taylor & Francis
Survey Designs There are three basic survey designs: 1. Cross sectional design survey 2. Panel study 3. Multiple cross sections © 2017 Taylor & Francis
Cross Sectional Survey Design • Analogous to a photograph a snapshot in time • Generalizable only to the sample population during that period of time © 2017 Taylor & Francis
Panel Study • A single sample is interviewed at multiple points in time • Allows us to see change in a particular group • Introduces threats to validity (testing and sensitivity) © 2017 Taylor & Francis
Multiple Cross Sections or Continuous Polling • Multiple cross sections: different samples are pulled from a particular population over time for comparison • Continuous polling: interviews take place for weeks or months at a time; the researcher combines the respondents’ data for time periods © 2017 Taylor & Francis
Mortality • A potential threat to validity; the death of a sample element • If a respondent in a panel designs stops being interviewed • Multiple cross sections avoid mortality problems, but present sampling challenges © 2017 Taylor & Francis
Interpreting Statistical Numbers Three basic interdependent quantities must be known: • Sampling error • Confidence level • Sampling size © 2017 Taylor & Francis
Sampling Error • The error rate is how far off the sample statistic is likely to be from the population parameter that is being estimated (the population’s true score) © 2017 Taylor & Francis
Confidence Level • The probability / likelihood that the actual population parameter is in the range of the sampling error • The wider the range, the more certain we are that the range encompasses that actual population parameter; the narrower the range, the less certain © 2017 Taylor & Francis
Sample Size (N) • The number of people or things in the sample © 2017 Taylor & Francis
- Slides: 13