A very simple extension of the ANOVA model. Instead of analysing the values of some dependent variable as in an ANOVA, the analysis is of the residuals of those values after first regressing them on some other variable.
The classic example in our field was to analyse change in therapy not by analysing the final score on some measure but to analyse that score after regressing out its relationship with the starting score. Given the tragic dominance of controlled trials in our field this would typically have analysed those residuals against the arm of the trial. There were pros and cons to doing this as opposed to analysing the data using a mixed ANOVA with the arm as a between-participants predictor and starting and finishing scores as a repeated, within-participants predictor.
The approach is within the NHST paradigm so gave p values for the predictor effects.
Details #
Hm. I should probably put some simulations here or in an Rblog post some day but for now I think this will do as the approach has been largely replaced by using multilevel models and analysing change across more than just baseline/termination values.
Try also #
- ANOVA
- NHST (Null Hypothesis Significance Testing) paradigm
- Regression and linear regression
Chapters #
Not covered in the OMbook.
Online resources #
None likely from me I think.
Dates #
First created 27.ii.26.
