 |
The Survey Method
How Are You Going to Get Your Respondents?
Sampling Methods:
Sampling methods are a crucial part of survey work. The goal is to get a representational sample. What does that mean? Well, you can't test everyone in the population. Most populations are too big. Therefore, you test a sample and you hope that the smaller group is similar to the larger.
What are the important issues?
Response Rate:
50% to 90% is preferable (but see our comment on the dark side above).
Sample Types:
Haphazard - Chosen by hit and miss methods - man on the street for example - Risk: Can be wildly off! Watch Leno and Letterman for the Man in the Street! May be successful for UFO aliens picking up humans to probe, though.
Purposive Sample - a deliberately chosen nonrandom group with some specific purpose. You'd better hope that the small group is on target. For example, asking college presidents about the best small college may not be the same as asking students. When was the last time these old coots were in a college dorm?
Convenience Samples - You use an intact group like a class. Easy to do - for example, Psychology students - may be OK if you are studying dark adaptation but certainly not representative of people on the minimum wage.
Probability Samples and Random Samples- First, determine the population (Sampling Frame). Those selected are called an element - then use one of these methods. It is hoped that the smaller group will mirror the larger group. Techniques exist to calculate the number of subjects needed to give a reasonable confidence interval for your estimate.
Systematic Sample - Choose people from a list according to a numerical position. For example, every fifth person from an alphabetic list is picked. While this isn't a true random sample, it works pretty well. Look left.
Able Aardvark 555-1212 #1 Joan Ace 555-1212 Bill Atom 555-1212 Kyle Azore 555-1212 Bruce Batman 555-1212 #5 Etc. Clark Zargon 555-1212 #625 Random - Everyone is going to have an equal chance of getting into the sample. The risk is that you had better be sure that you really do give people an equal chance. What if there are subgroups that your solicitation methods don't have a chance of reaching?
Stratified random samples - Your goal is to match relevant subgroup proportions. Then, choose randomly in the subgroup.
I've seen survey firms go a little nuts with this technique. A client was a representative sample and the stratified sample is used. However, due to expense, only 50 folks are sampled. Given the demographics of the area - this required for proportionality only three African Americans in the sample. Can you really generalize about the opinions of African Americans with a sample of three?
You also have to be careful about the variables that you use for the stratification. You might be interested in the role of race in jury selection issues. However, some studies found that income was just as important. Did you stratify on that?
Bottom Line
Survey methods are crucial to the world we live in. We see important policies being driven by the polls. We monitor progress on racial and ethnic harmony using surveys. Our politicians live and die by the polls. Thus, we must pay attention to questionnaire design, sampling and interviewing techniques. A change in a question can bring wildly different answers.
A few years ago, college men were asked whether they thought it was important whether or not their marriage partner was a virgin when they were married. 80% said: NO. Another sample was asked if it was acceptable for the person they would marry would have engaged in premarital sex. 80% said: NO.
Huh? Why was there such a difference? It turned out that "premarital sex meant" many partners to the sample men, while "virgin" implied that the woman would have had sex only with them. The latter was OK.
Whenever, you engage in survey research - keep these kinds of problems in mind.
|
|
