Dear SDM expert, 

I have performed a meta-regression with "percentage males" in the sample as continuous predictor. When I extract effect size and bias information for every peak, I get very small Hedges g values (around 0.01) while the peaks have an SDM Z of around 3 (for contrast analyses these SDM-Z values tend to translate into an Hedges' g of about 0.4). Using the conventional rule of thumb this would indicate that the effect is negligibly small, while the effect is highly significant at a relatively small sample size - this seems odd.  

My question is: how is hedges' g calculated for a continuous predictor in meta-regression, and can it be intepreted in the same way as for a contrast test? 

UPDATE: when I run the tutorial with ybocs as continuous predictor, I do not get a very small Hedges g value for ybocs, so it must be something peculiar to my data-set. 

Thanks and best,

Jurriaan 

Dear Jurriaan, 

The Hedges'g value and SDM Z are two different measures that do not always correlate similarly. SDM Z indicates the significance of the results as the p-value, whereas Hedges'g indicates the effect size. In your case, you got a significant peak of small effects for the percentage of males.

Best,

Lydia