Using Quotas with Survey Software I
When you do a survey you almost always want the answers you receive to reflect those that would be given by a larger group of people, if you are able to ask all of them. For example, if you are trying to do gauge national public opinion about a proposed change to Social Security, you cannot ask everyone who might be affected, so you ask a sample.
If you want the opinions of both current and future retirees, you need to make sure your ample includes adequate numbers of each. If you are doing telephone interviews, especially if you are doing some of them during the daytime, you might find it easier to reach retired people than people currently working. So if you just go with the people who are easiest to reach, you will likely wind up with a much higher ratio of retirees to current workers in your sample then there are in the general population. This difference from the general population could very easily bias your results. It is easy to imagine proposed changes to Social Security that might please current retirees, but not current workers or vice versa.
One way to minimize this risk is to use quotas. Sophisticated survey software supports quotas to let you determine in advance the proportions of your sample that represent different kinds of people. For example, the current ratio of workers to retirees is about 2.8:1. Using this ratio, if you wanted 1000 interviews, you would want your sample to include 737 workers and 263 retirees.
You could have a question at or near the beginning of your interview asking people whether they were currently working or currently retired. Your program should let you specify the numbers of people you want in each group, and when one group fills up, and an interviewer winds up talking to a person from that group, your program should tell the interviewer that the group is full, so that they can terminate the interview without asking the rest of the questions. They can then spend their time trying to reach people in the other group. You can do something similar with a Web page questionnaire that people fill out themselves. If they give an answer that puts them over a quota, the program can show them a message saying something like “Thank you, but we already have enough people similar to you in our sample. Thank you for your time.”
Our next blog will continue this topic and describe more ways in which sophisticated survey software that help ensure you are interviewing the kinds of people you need.