JASP: DFs for ANCOVA with within-subjects design

tomstafford

Hi

I am trying to run an ANCOVA on a within-subjects design. The residual DF reported by JASP is a bit less than the total number of rows in my data (570), whereas I believed that the correct DF should be 54 ( = N - 2 levels of the within subjects factor - 1 covariate).

Either:

1. JASP - is reporting the wrong DF or something other than the DF i need

2. I am mistaken about the correct DF for this analysis

3. Something else, e.g. I have mis-specified the analysis in JASP and need to (somehow) include participant as a random effect or something

Here is my analysis:
N= 57;
1 between subjects factor, 'group' with 2 levels;
1 within subjects factor, 'rotation' with 2 levels;
1 covariate, 'angle', which takes 5 values;
1 DV

Here is the data https://tomstafford.staff.shef.ac.uk/docs/correctrt.csv

Any help much appreciated!

RaviSelker

Hi Tom,

JASP only allows repeated measures data in wide format. Your data is in long format (for the difference, see http://www.theanalysisfactor.com/wide-and-long-data/), so you must reformat the data in order to use it in JASP. Subsequently, to run the mixed ANCOVA you need to select the repeated measures ANOVA in JASP (not the ANCOVA). The following website gives a clear, step-by-step description of how to setup your analysis in JASP (under JASP: Using the Software): http://www4.uwsp.edu/psych/cw/statistics/textbook.htm.

Cheers,
Ravi

tomstafford

Thanks Ravi - that is absolutely clear and useful

coledavis

I've converted a repeated measures file to wide format, but keep getting a message from JASP ANCOVA that "Singular fit encountered; one or more predictor variables are a linear combination of other predictor variables".

Temporarily, on www.coledavis.org/test you will find the original repeated measures ANOVA csv file and the new 'wide format' ANCOVA csv file. Can you tell me what I've done wrong?

RaviSelker

Are you doing the analysis on the "ThreeWayRepeatANOVA.csv" data set? Because that is the one that you should use in JASP as this is in wide format. The "Ancova.csv" is in long format. If I try the analysis I assume you want to do (I assume you want to do a 322 RM ANOVA) using the "ThreeWayRepeatANOVA.csv" data set it seems to work.

coledavis

No, the ThreeWayRepeatANOVA.csv is only there to show you the original data, which does as you say work on ANOVA. The information in this was converted into long format for the purposes of ANCOVA. If you try out the ANCOVA.csv data on JASP's ANCOVA test, you should see the error message I reported.

coledavis

Perhaps you could send me what you deem to be a satisfactory data set which represents a repeated measures set for ANCOVA and a between-subjects data set which would be satisfactory for ANCOVA?

coledavis

Ah, I have just done some more reading. Really, ANCOVA shouldn't be used for Repeated Measures, requiring independent error terms. So, not really relevant. Sorry again.