What is a probability-weighted sample for a survey?

When law departments are surveyed at random, and a single lawyer responds from each participating department, any given lawyer in larger department has a smaller probability of responding than a lawyer in a smaller department. More precisely, the probability of a lawyer being included in the survey is inversely proportional to the size of the department in a random sample of law departments, if lawyers within a department are selected with equal probability and there are no non-responses.

However, composition of the sample is affected by non-response. One source of non-response is non-availability: no one answers the phone, no one receives the message on the answering machine, or no one answers the email or letter. It seems reasonable to suppose that in a larger department it is more likely that someone will be in the office.

Statisticians have developed methods to correct for this bias, one of which is to create a probability-weighted sample. With these methods, large law department lawyers have the same influence as smaller law department lawyers.

We welcome comments

Your email address will not be published. Required fields are marked *