search

msrosenberg / MetaWin

GNU General Public License v3.0
7 stars 1 forks source link

Import data #30

Open FernandaItaly opened 5 months ago

FernandaItaly commented 5 months ago

In my specific meta-analysis situation, since not all studies report variances for each effect sizes, I calculate weights using the formula proposed by Adams et al. (1997) (wi= Ne*Nc/Ne+Nc) Cattura

So, if variance of effects sizes variances is required (as reported below in the pciture), is it correct to use the inverse of the weights?

msrosenberg commented 5 months ago

Yes, if you want to use the nonparametric weight estimate from the paper, you can import the "variance" estimate for your studies as the inverse of that weight or (Ne+Nc)/(Ne*Nc). That is exactly what MW2 calculated and output into the "variance" column in the effect size estimator so it would inverse it (1/Vi) to get the weight when performing an analysis.

FernandaItaly commented 5 months ago

Thank you for your response. Additionally, rather than importing the effect size as aresponse ratio (RR=Xe/Xc, where Xe and Xc represent the mean of the experimental and control groups respectively) I have chosen to directly import the individual percentage changes of the response ratio (%change Li= (exp(lnRR)-1)*100) to enhance the presentation with the use of the percentages. Will my choice potentially impact the output results?