Logistic regression in JASP: Calculation of df
Hi,
I am using JASP and the JASP handbook (one of the best stats books I've ever come acorss!) in my stats class. Last week, we discussed logictic regression, which is unfortunately not yet covered in the JASP handbook. When comparing the JASP output to other software (e.g., SPSS) and checking with other stats books (e.g., Discovering Statistics using R) I noticed that the degrees of freedom for the model summary differed. Is it correct that JASP calculates the df as N-1, and not based on the number of predictors in the H0 and H1 models? Is there a reason for this difference - or did I look up the wrong test in other books?
Thank you,
Laura
Comments
Hi Laura,
Thanks for noting this. I'll pass it on to the relevant team members and they will get back to you soon, I hope.
Cheers,
E.J.
Hello Laura,
we have looked into this and it could indeed be clearer in JASP. Currently, only the DF for the model are displayed which are N - predictors (it should not always be N - 1, however?). The DF for the various test statistics which are in the same table are calculated as DFh0 - DFh1, which is not explicitly displayed, so it might cause confusion.
Greetings,
Jonas
Thanks EJ and Jonas, very helpful!