Why use dfmethod(satterthwaite) in mixed model?

[Brennan RBratton]

Hello Statalist!

Can someone explain in which instance would you use Satterthwaite method of calculating degrees of freedom in a mixed model? I understand that Satterthwaite and Kenward-Roger are preferred for smaller sample sizes. What is "small" in this instance? I have 55 participants with 25 years of data for each. Should I use this instead of the default?

Thank you for your help!

[Marcos Almeida]

I believe so. The example 14 of the Stata Manual will display a sample of 110 individuals.

[Brennan RBratton]

Thank you, Marcos!