Midterm FAQ, part 1: Generalizability / External Validity

Q: On the midterm practice test, why is (c) the correct answer to #19?

A: The main thing to remember here is that because the sample was not representative, you can't generalize to _any_ larger group.

The methodology of a study isn't the main thing that tells you whether or
not you can generalize the results; rather, it's the _sampling method_ that tells you. That's the reason that results of the survey/correlational method are usually more generalizable than results of other methods -- good correlational studies (surveys) use representative samples. So do good content analyses. Experiments usually use convenience/availability (nonrepresentative) samples. When a study uses a nonrepresentative sample, you technically can't generalize to any larger population (though, as Prof. Mullin and I mentioned, experimenters who are looking at "universal" human processes often assume that they can generalize to everyone).

So (a) is wrong because you can't generalize to all Santa Barbara shoppers, and answer (b) is wrong because you can't generalize to all La Cumbre shoppers. The only correct conclusion you can make is that the people _in the sample_ showed those results, so (c) is correct, because it concludes that _some_ shoppers (those in the sample) gave high scores to questionnaire qu. #1.

Also, remember that generalizability has 2 main parts: sample, and environment. If the sample is not representative, you can't really generalize to a larger population. If the study environment is very artificial, then you can't really generalize to real-life settings.