This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Lecture 8 Nancy Pfenning Stats 1000 Pitfalls of Sample Surveys Examples discussed in the preceding lecture illustrate some of the most common problems in taking samples for surveys: • Using the wrong sampling frame • Not reaching the individuals selected • Nonresponse or volunteer response • Self-selected sample • Convenience or haphazard sample The best survey design uses a sampling frame that matches the population of interest. A probability sampling design is used to make the selection at random, rather than relying on volunteers or convenience. Every effort is made to reach the individuals selected, and follow-up efforts are made to attempt to contact non-respondents. Once a representative sample has been correctly obtained, the respondents must be surveyed in such a way as to elicit honest and accurate responses. The following pitfalls should be avoided: 1. Deliberate bias: A man with a clipboard stopped me on DeSoto Street one day, asking, “Ma’am, do you smoke? No? Good, then would you sign this petition to keep smokers away from non-smokers like you in the workplace?” After I declined, he confronted a man: “Sir, do you smoke? Yes? Good, you can win a free pack of Kools by agreeing to sign this petition to provide for designated smoking areas in the workplace.” Questions should not be worded in such a way as to influence the response. After a backcountry camping trip in Glacier National Park, I was asked to complete a survey by University of Idaho researchers on interactions between backcountry travelers and grizzly bears. One question read, “Because people and cattle live practically everywhere in the United States, and grizzly bears only in Wyoming, Montana, and Alaska, I think Montana should forego some grazing when there is a conflict with a bear: (circle one) strongly agree/agree/neither agree nor disagree/disagree/strongly disagree.” This was a leading question which put pressure on respondents to agree, and should have been more neutrally worded. 2. Unintentional bias: USA Today reported on a survey of 102,263 randomly selected adults in 49 states: 87% of the people rated their health as “good” to “excellent”. No wonder: the question read, “Is your health generally excellent, very good, good, fair, or poor?”! The percentage rating health as good to excellent would almost certainly have been lower if the question had listed the options as “excellent, good, adequate, fair, or poor”. 3. Desire to please: People frequently claim to have voted in a past election, even if they hadn’t, because they want to satisfy the interviewer with a substantial response. To avoid this bias, a Gallup Survey question reads, “In the election in November 19–, did things come up which kept you from voting, or did you happen to vote? For whom?” 4. Asking the uninformed: People may claim to know something because they are embarassed—or not given an opportunity—to admit they don’t. Many pollsters do not include “don’t know” as annot given an opportunity—to admit they don’t....
View Full Document
This note was uploaded on 02/15/2012 for the course STAT 1000 taught by Professor Taeyoungpark during the Fall '06 term at Pittsburgh.
- Fall '06