[R-sig-ME] Intercept Variance Components Equal 0- Change Score

[Phillip Alday]

Without seeing the rest of your model, it's hard to tell. I suspect you
model is something like:

change_score ~ 1 + some_predictor + (1|ID)

Then you only have a few change_scores by participant and the
by-participant variation is indistinguishable from the residual
variation. Because the random-effect for participant measures the
variation due to participant *beyond* the residual variation, it goes to
zero here.

Phillip

On 14/09/2019 05:03, Daniel P Moriarity wrote:
> Hello all,
>
> I am revising some MLMs using lmer and a reviewer suggested we use change
> scores as our outcome. When I did so, I noticed that the variance
> components for the intercept were both 0 (see below) regardless of the
> variable I was predicting to (all 6 are now change scores). Any ideas as to
> why this might be? The change scores are not person-centered, and we
> confirmed that there is between-person variability using the VAR function
> (see below the random effects ouput). Thank you.
>
>
> Random effects:
>
>   Groups   Name        Variance Std.Dev.
>
> ID       (Intercept)  0.00    0.000
>
> Residual             19.46    4.411
>
> Number of obs: 310, groups:  ID, 130
>
>
>> var(Data_group_by$CDI_Change)
> [1] 6.353349
>
>
> My best,
> Daniel
>
> Doctoral Student, Clinical Area
> Mood and Cognition Lab
> Department of Psychology
> Temple University
> 1701 North 13th Street
> Philadelphia, PA 19122
> daniel.moriarity using temple.edu <hannah.frank using temple.edu>
>
> *WARNING: This e-mail may contain material that is confidential and is for
> the sole use of the intended recipient.  If you are not the intended
> recipient, please contact the sender and delete all copies.  There are no
> guarantees that electronic communications are secure and protected.*
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models