Command to obtain effect sizes.

[Alex Xu]

Hello,
I am running two models on a panel data of 50 firms over 15 years (one using the xtnbreg and the other using xtreg). I tried to obtain the effect sizes by using the esize, esizei, and estat esize. However, all of them returned with error messages. Would you please help me to identify a command that will produce the effect sizes?
Many Thanks.

[Wendy Garrard]

Welcome to StataList Alex.

What type of data are you working with -- standardized mean diffs, correlations, other? Have you derived any transformations for your stats -- e.g., Fisher's Z?

Because you have panel data I'm not sure I can help you directly. But if you post more information about your operating system, version of Stata, specific Stata menu commands or user-written modules you have tried, and the exact error message presented with each command I will try to help.

Also, this User Group is a great resource -- but folks are (usually) able to help only if you provide more descriptive information in your subject line and body text. This advise is explained in detail in the StataList FAQ. . . . You are highly likely (p<.001) to get more responses to your posts after you read the FAQ.

Good luck with your meta

[Phil Bromiley]

Welcome Alex to Statalist.

You'll increase your chances of a precise answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. Also, it helps if you simplify things as much as possible while still retaining the basic problem. Not need to include any variables you don't use, etc., maybe even create an artificial data set (or use one of the data sets provided by Stata in the documentation).

You first need to define what you mean by effect size. The kind of effect size that folks use in anova is not necessarily applicable to panel non-linear models. What I often do is to look at the change in predicted value for a given change in the x's. This is easy with the margins command. Look at xtreg postestimation and xtnbreg postestimation in the documentation.