Small I square in TSSEM1

Posted on
Picture of user. jing niu Joined: 10/10/2018
Forums

Dear Mike,

Thanks a lot for the kind help all the time.

I came across with an issue when interpreting my tssem1 model(I restricted re.type="Diag"). I have 15 correlations to be pooled. But for correlation14, I got a Tao square equal to 1 and its I square is much lower than I expected (0). And another correlation with I squares equal to 0.4. I only know that this means the variance is not largely at study-level. I am wondering whether this means that my results are questionable? Can I apply the correlations to my tssem2 model? Also, is this means that my dataset is too small?

I appreciate a lot for the attention a lot!

Best,

Jing

Replied on Sun, 10/21/2018 - 02:44
Picture of user. Mike Cheung Joined: 10/08/2009

Dear Jing,

The small heterogeneity variance (tau^2) means that the population effect sizes seem homogenous. It does not invalidate fitting structural equation models on the average correlation matrix in the stage 2 analysis

Regards,
Mike