Monday, March 4, 2013

Science and Trendability

UPDATE:  Link to the Bones article fixed (I hope) - check it out!

The second point made on this blog post (about the precognition article "Dr. Bones" poked fun of, as I discussed here) is a good illustration of what we could call the equivalent to market research's "trendability" in the science context: exact repetition (or exact replication).

We (now: they?) don't do exact replication in social psychology.  Instead, in a single article, the researchers show "conceptual" replication (demonstration of the same basic effect but not doing the same exact study) in a series of experiments, but usually with extra bells and whistles as the experiments progress.  One cookie-cutter way a 3-study paper (to the extent you can publish 3 study papers anymore!) would look is:

Study 1 -- show the main effect (X causes Y)
Study 2 -- show moderation (X causes Y in group A but not in group B)
Study 3 -- show mediation (X causes Y because X causes Z and Z causes Y).

Notably, the stimuli (i.e., what the participants are responding to in the experiment) are usually different in each study and the measures are often different (e.g., measuring attitude differently across the various studies).

There is nothing wrong with conceptual replication, of course.  But the lack of exact replication is problematic.

No comments: