Using Quotas with Survey Software II
The most basic quotas look at one question and let you specify how many people giving each answer to that question you want to include in your final sample. For example, you might decide you want 480 men and 520 women. More sophisticated survey software lets you base a quota on a combination of the answers to two or more questions. For example, you could use gender and age questions to say that you want 37 men aged 18-24, 42 men 25-34, 41 men 35-49, 39 women aged 18-24 and so on.
The above examples discuss quotas based on questions people answer while completing your questionnaire. If you have information in advance about people taking your survey, more capable programs will also let you base quotas on that information. One situation in which you will often have information in advance is when you are doing telephone surveys. Sometimes people do telephone surveys only having the telephone number, but other times people do surveys having other information associated with each number. For example, you might have people’s snail mail address. That would let you establish quotas based on their state of residence or their ZIP Code. You might also have codes representing other characteristics of the people in your sample, such as gender, products purchased and so on. You can base quotas on these as well.
Your program should then be able to check if quotas are still open before they present a particular phone number to an interviewer. If someone is a male 35-49, and you already have as many of that kind of people as you need, the program can look for another telephone number that belongs to someone in a different gender/age group and present that to your interviewers. This kind of checking in advance can save your interviewers significant time.
Similarly, you may have information in advance about people you invite to take a Web page survey. If you associate a password with each person, your software should let you bring that information into a data record at the start of a survey. That lets you use that information for quotas, just like you can use questions people see. You should also be able to base a quota on information you have in advance about people combined with survey answers.
For web surveys in particular, your program should be able to distinguish between people who are qualified to take your survey, but are in a quota group that is full, and people who are not qualified to take your survey. For example, someone might not be qualified to take your survey if you only are interested in the opinions of working people, and someone is currently retired. Your program should be able to present different messages to people excluded from your study for each reason. The ability to distinguish between over quota and not qualified can be especially file important when using a third-party sample provider. These providers typically want you to send them a message at the end of each questionnaire indicating whether someone completed it, was over quota or was not qualified to take the survey.