I am running subgroup analysis (adult vs. child) and meta-regression on a set of data using Comprehensive Meta Analysis.
(1) The overall effect size using basic stat (Random Effect, log odds ratio) is 0.778.
The meta-regression using intercept only is 0.778.
When I group studies by age, mixed effect analysis produce point estimate of overall effect 0.780.
Overall effect sizes for fixed effects model are the same for three methods.
(2) The subgroup analysis produce Q-value 0.753 for mixed effect analysis.
The regression produces Q value of 0.662 for "test of the model" line.
The total between study variance for the two methods are the same.
The N is same for all analysis.
My understanding is that the three numbers in (1) should be equal and the two numbers in (2) should be equal since they are essentially calculating the same thing when there are only two subgroups.
Could someone explain why the numbers are different ?
Thank you!
Your question is hard to understand without seeing the data and CMA output.
I'm a little confused as why you are simultaneously comparing outcomes of subgroup analysis (categorical) and meta-regression, but may provide some practical advice. I agree that it would be helpful to see the actual data set.
(1) The subgroup analysis in CMA has a few different options in how the subgroups are combined and overall effect size is calculated. i.e. the subgroups can be analyzed by random effect meta-analysis within the subgroup and then combined by fixed effect meta-analysis. Or random followed by random. The final overall effect size may differ a little between the two. You can change which to choose by clicking Computational options --> Mixed and random effects options.
(2) When you do meta-regression, Q-total would be the same as the Q value achieved by your initial random or subgroup analysis. The Q-model is the dispersion explained by the covariates that you used for your meta-regression which assesses whether the correlation is significant or not.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis