# DFbetas for identifying influential participants - linear mixed models

**URL:** <https://discourse.julialang.org/t/dfbetas-for-identifying-influential-participants-linear-mixed-models/73399>\
**Category:** Performance\
**Tags:** question\
**Created:** [December 20, 2021, 7:50pm UTC](https://discourse.julialang.org/t/dfbetas-for-identifying-influential-participants-linear-mixed-models/73399 "2021-12-20T19:50:33Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![CatOliveira](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/catoliveira/32/29530_2.png) [@CatOliveira](https://discourse.julialang.org/u/CatOliveira)\
**Post date:** [December 20, 2021, 7:50pm UTC](https://discourse.julialang.org/t/dfbetas-for-identifying-influential-participants-linear-mixed-models/73399/1 "2021-12-20T19:50:34Z")

</div>

Hello everyone,

I have used the package [influence.me](https://cran.r-project.org/web/packages/influence.ME/influence.ME.pdf) to determine whether I had influential participants using dfbetas. Yet, this package takes forever to run and I just can’t wait another month for it to run. I tried to use it in Julia with RCall to see if it would help, but so far it doesn’t seem to be helping. I guess if you use slow code in a different interface it just behaves as it would normally.  
Does anyone know of a package in Julia that runs dfbetas for linear mixed effects models? I know of LinRegOutliers but that one is just for linear regression.

Thank you!

---

<div class="post-metadata">

**Author:** ![dmbates](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/dmbates/32/44_2.png) [@dmbates](https://discourse.julialang.org/u/dmbates)\
**Post date:** [December 20, 2021, 11:23pm UTC](https://discourse.julialang.org/t/dfbetas-for-identifying-influential-participants-linear-mixed-models/73399/2 "2021-12-20T23:23:10Z")

</div>

You are quite correct that calling an R package through a different interface is unlikely to make it run faster.

I don’t know of any Julia packages for evaluating dfbetas for linear mixed models. Generally the `MixedModels.jl` package outperforms `lme4` for fitting linear mixed models. If it was known how `influence.ME`is performing the calculations it may be possible to emulate it with a model fit by `MixedModels`. There is already a method for `StatsModels.influence` for a `LinearMixedModel` object. I think it should be much faster than the R method but still it is not blazingly fast when applied to large models.