I have a question regarding mixed effect models. Is it possible that the degrees of freedom differ when I use the same model with different dependent variables? Or does that indicate that something went wrong?

I analyzed my data using the MIXED procedure in SPSS. There are two random factors (participants and stimuli) and one fixed (continuous) factor (PD) in my design.

The number of degrees of freedom is 1201,436 for my first dependent variable "perf". When I use the same model and only choose a different dependent variable ("wr") the number of degrees of freedom is 1194,166.

The number of data sets and the model dimensions are identical for both dependent variables. The information criteria differ (e.g., AIC for "perf" = 3975 and AIC for "wr" = 9942). The tests of fixed effects are significant for both dependent variables.

Does it make sense that the degrees of freedom differ?

  • $\begingroup$ Have you checked missing data? $\endgroup$ Jan 22, 2014 at 13:25
  • $\begingroup$ yes, it's the same for both dependent variables $\endgroup$
    – Josephine
    Jan 22, 2014 at 13:32


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.