Is there any established way of reporting approximated degrees of freedom (be it Satterthwaite or Kenward–Roger) for mixed-effects model in psychology (APA 6 ed.)?
Specifically, is rounding to whole numbers considered best practice or should I stick to two decimal places (as in F or t tests)?
The APA manual doesn't specify this. All examples in the manual are integer DF.
But the APA manual does recommend to give all numbers rounded to two decimals, except for the p-value, which is rounded to three decimals.
Therefore I would suggest to round DF to two deximals.
What you can do is to find examples in your field through Google Scholar. E.g. perform a search for some keywords in your paper and add "Satterthwaite" and "degrees of freedom" OR "DF" and see how your colleagues handle this.
