# Using probit link for ds.glm

**URL:** <https://datashield.discourse.group/t/using-probit-link-for-ds-glm/798>\
**Category:** Support\
**Created:** [16 June 2023 14:54 UTC](https://datashield.discourse.group/t/using-probit-link-for-ds-glm/798 "2023-06-16T14:54:00Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Devin](https://avatars.discourse-cdn.com/v4/letter/d/278dde/32.png) [@Devin](https://datashield.discourse.group/u/Devin)\
**Post date:** [16 June 2023 14:54 UTC](https://datashield.discourse.group/t/using-probit-link-for-ds-glm/798/1 "2023-06-16T14:54:00Z")

</div>

Hello,

is it possible to specify a link function for the ds.glm function? I would like to do binary regression and use the probit link for it, but apparently this is not possible yet for this function, right? Is there somehow a way to get the regression coefficients for probit modeling of the data?

Thanks a lot

---

<div class="post-metadata">

**Author:** ![demetris.avraam](https://avatars.discourse-cdn.com/v4/letter/d/439d5e/32.png) [@demetris.avraam](https://datashield.discourse.group/u/demetris.avraam)\
**Post date:** [19 June 2023 12:05 UTC](https://datashield.discourse.group/t/using-probit-link-for-ds-glm/798/2 "2023-06-19T12:05:09Z")

</div>

Hi Devin,

The glm functions in DataSHIELD use the canonical link functions for each family (e.g. identity for gaussian, logit for binomial, etc.) and there is not any way to specify a different link function. I am not sure if they are is a way to calculate the coefficients for a probit model if you use a different link. I will have a look and I will let you know if I find anything relevant.

Many thanks, Demetris