Meta-analysis
Posts  1 - 1  of  1
cacrane
I have several studies that fail to report data necessary to calculate an effect size. Rather, the authors state that the relationship between the variables of interest was not significant. I've seen other analyses calculate d using the degrees of freedom and an estimated p=.50 for non significant results and p=.05 for significant results. I was wondering if this is the conventional method of addressing this issue. In the past, I have eliminated studies that fail to report specific data but I'm currently working with 20 studies and would like to retain as many as possible.

Also, one of my studies has no variability within or between groups (all participants denied violence perpetration). I can think of no way to calculate a reliable effect size using this data. Any suggestions? N=30 with an even split and M=1, SD=0 for both groups.
Save
Cancel
Reply
 
x
OK