A colleague is telling me that you can use more “flexibility” in action research with whether the data fits the assumptions of parametric tests, such as a normal distribution, homogeneity of variance, etc. Does anyone have any thoughts on this?
I cannot see the logic in that argument. In general, you want to use the best information available for determining things such as the kinds of action you want to take, so sloppy analysis would not match that goal.
When the participants are involved in the study and then you are capturing the data either through measurement or observation, then the data may or may not follow any distribution. In action research, in your context, how many variables are quantified and how many are described - based on this you can choose the tests. But as David L Morgan mentions, the case may not always fit parametric tests. There could be non-parametric.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation