I follow (with interest) questions and answers that talk about pre-post experimental or quasi-experimental research designs. I have noted quite a few questions that ask about this particular research design (not surprising, it is a common design), although IMO our current set of tags are lacking in uniformity, and so here I want to talk about what tags should we assign to related pre-post design questions? Example questions that I am aware of are;
- Best practice when analysing pre-post treatment-control designs
- How to correct for correlation at baseline between predictor and “DV”?
- How should one control for group and individual differences in pre-treatment scores in a randomised controlled trial?
- Help with change scores and/or ANCOVA design
- Is it valid to include a baseline measure as control variable when testing the effect of an independent variable on change scores?
- How to assess differences between three groups in a pretest - posttest design using SPSS?
- Should the difference between control and treatment be modelled explicitly or implicitly?
- What analysis to use with two groups, a pre- and post-test, and three outcomes?
- Pre, Post, and Change predictors all in the same model
I'm sure others exist and would appreciate comments/answers pointing to other examples.
Current (relevant) tags in the list of questions are;
- repeated measures
- change scores
- clinical trial
- least squares
- linear model
I bold repeated-measures and change-scores because they may serve as reasonable tags already in place that could be added to all of the questions (the others are included to highlight how little the current set of tags connect the related questions). I say could because I don't necessarily think either is a great idea. The change scores tag implies a particular type of analysis (which is not always appropriate nor endorsed). Repeated measures is a terminology not frequently associated with fields I am more familiar with (sociology, criminology, economics). For instance, economists difference in difference models (e.g. Card & Krueger, 1994) are really one type of change score models. I don't think I've ever heard repeated measures used by an economist though (much related to the panel data terminology rants @Stask has mentioned in some of his answers).
In addition to this (my opinions aside) I would suggest we reference this answer by chl as a canonical response to related questions. It is not too much of a stretch to say it is still an open debate, and as such references to other literature with discussion about nuances would almost always be appropriate. As such this question in particular should be given the tag(s) we expect such questions to have in the future.
So before getting ahead of myself, should these questions all have at least one tag in common? How about a tag for
pre-post-design? In additon should we cut some of the tags from these questions? Any thoughts on the subject (and how we should proceed in the future) are welcome.