Skip to content
Prev 175358 / 398506 Next

Post-hoc and planned comparisons for repeated measures

A couple more thoughts.  If you're using nlme::lme to fit the model
there is a 'contrasts' argument to lme which can be used to structure
the design matrix to produce tests of hypotheses of interest.  Also,
when you pass an lme object to anova you can use the 'L' argument to
specify linear combinations of the coefficients to be tested to be 0.

Kingsford


On Sun, Mar 29, 2009 at 11:57 AM, Kingsford Jones
<kingsfordjones at gmail.com> wrote: