Which kind of inference method is preferable?

[ccepeda10]

Hello there,
I was wondering about the criteria for preferring one inference method over another (parametric/nonparametric bootstrap or jackknife). The package recommends using parametric bootstraps or jackknife for small samples, but there isn't much guidance for bigger samples. I also read the paper, but I couldn't find much information about this either, except from a brief comment about the validity of parametric bootstrapping under some conditions (which are not specified). I know this is a niche subject, but is there any reference about this matter?
Thanks

[xuyiqing]

The nonparametric bootstrap is infeasible when the number of treated units is small -- they simply would not appear in some of the bootstrapped samples.

…

On Fri, Feb 4, 2022 at 9:44 AM ccepeda10 ***@***.***> wrote: Hello there, I was wondering about the criteria for preferring one inference method over another (parametric/nonparametric bootstrap or jackknife). The package recommends using parametric bootstraps or jackknife for small samples, but there isn't much guidance for bigger samples. I also read the paper, but I couldn't find much information about this either, except from a brief comment about the validity of parametric bootstrapping under some conditions (which are not specified). I know this is a niche subject, but is there any reference about this matter? Thanks — Reply to this email directly, view it on GitHub <#76>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2PKGFBQANCNV5ANLUWMKDUZQGANANCNFSM5NSM3X7Q> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Yiqing Xu Assistant Professor Department of Political Science Stanford University https://yiqingxu.org/

[ccepeda10]

Thanks for the answer!
What about bigger samples, though? Which one would be preferable? Or it's relatively indifferent?

[xuyiqing]

Inference using the nonparametric bootstraps is unconditional on factors and loadings and thus is preferable in my view. It is also easier to interpret.

…

On Fri, Feb 4, 2022 at 9:50 AM ccepeda10 ***@***.***> wrote: Thanks for the answer! What about bigger samples, though? Which one would be preferable? Or it's relatively indifferent? — Reply to this email directly, view it on GitHub <#76 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2PKGFJFHTBVUXXM73P4L3UZQGWHANCNFSM5NSM3X7Q> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.***>

-- Yiqing Xu Assistant Professor Department of Political Science Stanford University https://yiqingxu.org/

[ccepeda10]

Thanks a lot for the help and the prompt answer! I'm going to use nonparametric bootstraps then.
Best regards.