If you can do it

https://www.theanalysisfactor.com/effect-size/

Hello Robert,

You can do this.

1. Combine the samples, but include a variable to indicate group membership (e.g., 0 = group A, 1 = group B).

2. Re-estimate your model for the combined sample, being sure to include group membership as a predictor and an interaction term of group by age (e.g., group * age).

3. The new model is: Y-est = b0 + b1*group + b2*age + b3*interaction.

If coefficient b3 is significant, then the estimated slopes for groups are significantly different (after adjustment for age).

Here's a link that walks you through the set-up and interpretation: https://stats.idre.ucla.edu/spss/faq/how-can-i-compare-regression-coefficients-between-two-groups/

Good luck with your work.

David Morse, thank you for that hint, but this is what I already did (and the interaction term is indeed significant). However, I was asked to compare effect sizes in addition. More precisely, I was asked to explicitly compare the effect sizes of the regression coefficients (i.e., compare b1 in the first regression with b2 in the second regression):

Regression 1: y1 = a1 + b1*x1

Regression 2: y2 = a2 + b2*x2

Also, I was asked to interpret the effect size of the coefficient of the interaction term (i.e., of the regression equation that you proposed).

Do you have an idea, what to do here, David Morse?

Thanks a lot and kind regards,

Robert