 Calculate pooled standard deviation f...    Message/Author  Rulph S posted on Monday, December 28, 2015 - 4:03 pm
Hello,

I wanted to ask how I can calculate the pooled standard deviation from a TECH4 output?

I understand I get the standard deviations by using the square root on each of the measures on the diagonal from the covariance matrix - but how does this lead me to the pooled standard deviation? Specifically, I want to calculate Cohen's d, and substracting the mean of T2 from the mean of T1 is not hard, but what is a simple formula to obtain the pooled standard deviation for the division?

For example, I have the following numbers:

Mean of T1: 0.000
Mean of T2: -0.063

And on the diagonal:

Variance (T1): 0.679
Variance (T2): 0.649
Variance (I): 0.679
Variance (S): 0.414

d=(-0.063-0.000)/????

I would be glad if you could help me with this!  Bengt O. Muthen posted on Monday, December 28, 2015 - 4:17 pm
Because you seem to have the T1 mean fixed at zero only the mean of T2 has variability so no need for pooling. I assume that this concerns time 1 to time 2 change. Just divide by -0.063 by the square root of the T1 variance which tells you how big of a change has taken place in the T1 metric.  Rulph S posted on Monday, December 28, 2015 - 4:45 pm
Oh that's great - thanks a lot!

So it's just (meanT2-meanT1)/sdT1

I'll use that, thanks again.    Topics | Tree View | Search | Help/Instructions | Program Credits Administration