jamovi

Hello,

I have an issue that is related to an earlier topic (viewtopic.php?f=5&t=1634) but I haven't been able to solve it. When I try to run a man(c)ova on a data set with more that 5000 cases, I get an error "sample size must be between 3 and 5000", even though the Shapiro-Wilk test is not selected (which is supposed to be causing it). Is there a way to get the manova to run? It works fine with smaller samples. I'm using v2.3.13.0 for Mac.

Many thanks,
JR

jerichoreg@gmail.com wrote:I get an error "sample size must be between 3 and 5000", even though the Shapiro-Wilk test is not selected (which is supposed to be causing it).

Hey @JR,
first of all thanks for reporting.

It is the multivariate test of Shapiro-Wilk, which produces this error, because it is known that it accepts a data matrix not exceeding 5000 cases.
I will not go into the merits of this choice, but you can find many explanations for this, for example also here:
https://stats.stackexchange.com/questio ... ng-shapiro

Why does your MANCOVA stop in jamovi when you exceed 5000 cases, even if you have not checked the Shapiro-Wilk test checkbox?
It happens because the tests used are very fast and are recalled all together in a single invisible step, even if their check boxes are not selected.
The selection with the check boxes is used to show or hide the results returned by the various tests.
The passage, in the sequence of execution of the tests, from Shapiro-Wilk when the cases are higher than 5000, returns the error you have reported, stopping everything.
If you know R, you can see some code here:
https://github.com/jamovi/jmv/blob/mast ... ion,%7D%2C

What can be done?
Act on the code to capture the error and allow the process to continue and add a footnote to the test output table for when it will be selected with its check box.
Of course it's a small thing and may be available when there is a more substantial jamovi update from Team Jamovi.

Cheers,
Maurizio

For friend Jth
Right if you pass from here, I did the above in mancova.b.r and relative PR for the Team.
mz