Monday, August 24, 2009
Standardized Testing and Survey Research
I've been thinking lately about the similarity between standardized tests (of the type used in educational assessment) and surveys (of the type used in social science research).
As a student, I was bored by standardized tests. When I had the choice, I declined taking them. The sole benefit was filling admission requirements. Taking standardized tests was a matter of endurance that involved sitting in a room for many hours answering questions in a dry and mostly non-stimulating way. Frequently I felt frustrated that I wasn't allowed to demonstrate what I knew.
I do, however, understand why standardized tests exist. I'm not trying insult the intelligent people who create and maintain those tests. My concern is for the test-taker: Standardized tests are typically to be withstood rather than happily anticipated. Perhaps it seems strange to hope for a test to be engaging, but as we demand ever more standardized testing, would it kill us educators to make the experience a little more enjoyable? Alas, this is not my domain.
But I face a similar prospect as I contemplate how to gather information for a (very small) research study of my own. Basically, I would be asking my subjects to endure a standardized test. Granted, the scale and difficulty would be much less than an academic test, but I can imagine it still being more of a chore than a pleasure. I know that when I'm in the position of filling out a survey, I whip through the questions as quickly as possible and have very little patience for poorly worded or unclear communication. Does this generate genuinely useful research? Probably not.
So what I need to do is engineer a fun survey method. I think there are a few ways of doing this. First, I could try the old 'hide the medicine in the candy' trick, where the survey is concealed in a more exciting activity. Maybe I could embed my survey into one of those quizzes people are always taking on Facebook, for example. Or, I could perform a test that's basically invisible to the subjects. As long as my standards are clear beforehand, and variables are controlled, that might work. I think either of these approaches would be preferable to the 'pity me, please take my survey' technique of data-gathering.