Skip to content
Prev 5519 / 20628 Next

false convergence (8) with binary covariate (0, 1) but NOT if covariate is (no, yes)?

Is it possible that you coded the variable as an integer before and as a factor now?  I'm not sure if lmer treats a two-level integer as a factor, but that might have been it.
You might try to apply sum contrasts, i.e., -1/2 and 1/2.  Everybody please correct me if this is nonsense.

Best,
Paul
On 8 Mar 2011, at 14:25, Julia Sommerfeld wrote: