I'm attempting to run an ANCOVA with 1 between-subjects variable and 2 within-subjects variables and I'm running into an error that makes no sense to me. My data looks like this:
| Scan | ID | Region | ALFF | Age | Resp |
|---|---|---|---|---|---|
| 1 | 20 | AID | 0.826 | Adol | 77.25 |
| 2 | 20 | AID | 1.116 | Adol | 73.18 |
| 1 | 22 | AID | 0.362 | Adult | 78.70 |
| 2 | 22 | AID | 0.849 | Adult | 72.58 |
| 1 | 20 | MDM | 0.826 | Adol | 79.25 |
| 2 | 20 | MDM | 1.116 | Adol | 71.18 |
| 1 | 22 | MDM | 0.778 | Adult | 79.70 |
| 2 | 22 | MDM | 0.291 | Adult | 73.58 |
My ANCOVA code is:
Full_Anova_ALFF<- AlFF_Resp %>% group_by(Region) %>% do(fit=aov_car(ALFF ~ AgeScan+Error(ID/ScanResp), data = .))
and I get this error when I run it:
Converting to factor: Age
Error: Empty cells in within-subjects design (i.e., bad data structure). table(data[c("Scan", "Resp")])
Resp
| Scan | X77.25 | X73.1777777766667 | X63.1944444433333 | X70.3333333333333 | X78.7 | X72.5833333333333 |
|---|---|---|---|---|---|---|
| X1 | 1 | 0 | 0 | 0 | 1 | 0 |
| X2 | 0 | 1 | 0 | 0 | 0 | 1 |
| X3 | 0 | 0 | 1 | 0 | 0 | 1 |
| X4 | 0 | 0 | 0 | 1 | 0 | 0 |
Resp
| Scan | X72.4833333333333 | X78.25 | X65.1833333333333 | X71.9166666666667 | X57.333333335 | X65.55 |
|---|---|---|---|---|---|---|
| X1 | 0 | 0 | 1 | 0 | 0 | 1 |
| X2 | 0 | 0 | 0 | 1 | 0 | 0 |
| X3 | 1 | 0 | 0 | 0 | 0 | 0 |
| X4 | 0 | 1 | 0 |
Scan is factor variable and resp is numeric and I have no idea why this error is occurring. There're no empty cells! And this weird table that was outputted as a part of the error message is also quite strange. It appears to be treating respiration as a factor maybe? But I've definitely told it that respiration is numeric. When I take respiration out of the model, it runs completely fine. However, unfortunately, I do need to include respiration.
Anybody have any idea what's going wrong? Or even just a workaround I can use to get this done?
Thanks in advance for your help!!
For those interested, I figured it out! I ended up using a linear mixed-model instead of what I have above because my covariate was at the scan-level rather than a between-subjects variable.
My command ended up being:
Full_Anova_ALFF<- AlFF_Resp %>% group_by(Region) %>% do(fit=mixed(ALFF~Resp+Age*Scan+(1|ID)+(1|Scan),data=.))
Hope this helps someone in the future!