Recently, I ran Quade rank ANCOVA in SPSS for the first time. I followed the instructions at https://www-01.ibm.com/support/docview.wss?uid=swg21477497 (I have attached my syntax). I ran the analysis several times (for different outcome variables). For one of the outcome variables, one of the groups has a negative mean rank value (read from the estimates table). I struggle to understand what a negative mean rank can mean? Any help would be much appreciated.
Hello Nicola,
Estimated marginal means (or, so-called "adjusted" means) represent the best estimate of what the means would have been for the groups, had the groups had equal means on the covariate. The resultant values may well be (as you see with your data set), apparently "impossible" values. These can arise anytime you use a prediction equation which imposes no bounds (that's why, for example, logistic regression never yields impossible odds/probabilities of an event whereas ordinary least squares regression can an does when using a dichotomous DV).
How to interpret? It depends on whether the ranks were assigned in ascending (e.g., highest score = rank value of 1) or descending form (highest score = rank value of N). A negative rank estimate would therefore imply that the group was predicted to have either the highest or lowest rank (depending).
Good luck with your work.
Watch this video for step-by-step procedure to perform Non-parametric (Quade’s) ANCOVA,
https://youtu.be/jozx59H5IHw
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS