Creating Z-Scores for Meta-Analysis

[Kat123]

I will be conducting a meta-analysis that compares post-treatment changes in depression between the control and intervention groups. I have the Mean, SD, and N for depression at both time-points. Since different studies used measures to assess depression, I was planning to convert the change scores into z-scores to decrease heterogeneity.
I understand the standard z-score formula is z = (x - meanx) / SDpool, where x is the pre - post-treatment change, but I'm not sure about the other two variables. Is meanx the mean change, and SDpool the SD across all change scores (including both control and intervention values) for all studies?
Any help would be appreciated, thank you!

 

[hlsmith]

For clarity, you are desiring to convert individual study results to z-scores. Then run the meta-analysis on the z-scores, correct? How are results typically presented in the papers? Per your description, it sounds like differences-in-differences.

 

[Kat123]

For clarity, you are desiring to convert individual study results to z-scores. Then run the meta-analysis on the z-scores, correct? How are results typically presented in the papers? Per your description, it sounds like differences-in-differences.

Yes! The meta-analysis will be run on the z-scores. The results are presented as mean, SD, and n at each time point, and for each group separately. Would greatly appreciate any help!

 

[Karabiner]

I do not understand what z values can do here. Why don't you use some common
effect size statsitics like Cohen's d or Hedges' g?

With kind regards

Karabiner

 

[Kat123]

I do not understand what z values can do here. Why don't you use some common
effect size statsitics like Cohen's d or Hedges' g?

With kind regards

Karabiner

Thank you for the suggestion! I was planning to use z scores because a similar and recent systematic review coverted all change scores to z scores in order to reduce heterogeneity. Since the studies in my analysis also have heterogeneity (there were 3-4 different measures used across studies to measure depression), I was planning to do the same. I'm not so familiar with how to conduct a meta-analysis uisng effect sizes, if you have any resources to direct me towards, I would greatly appreciate it!

 

[hlsmith]

The Borenstein book is always a good place to start. It is free on the web.

 

[Kat123]

The Borenstein book is always a good place to start. It is free on the web.

Thanks so much, I'll check it out!