# Random effects IVW analysis

**URL:** <https://isgw-forum.colorado.edu/t/random-effects-ivw-analysis/892>\
**Category:** Day 7\
**Created:** [June 10, 2026, 3:20pm UTC](https://isgw-forum.colorado.edu/t/random-effects-ivw-analysis/892 "2026-06-10T15:20:32Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![bastien.rioux](https://isgw-forum.colorado.edu/letter_avatar_proxy/v4/letter/b/f9ae1b/32.png) [@bastien.rioux](https://isgw-forum.colorado.edu/u/bastien.rioux)\
**Post date:** [June 10, 2026, 3:20pm UTC](https://isgw-forum.colorado.edu/t/random-effects-ivw-analysis/892/1 "2026-06-10T15:20:32Z")

</div>

Is there an advantage to use the random-effects version of IVW in MR? Would you use it as a primary analysis vs the fixed effect?

---

<div class="post-metadata">

**Author:** ![davide](https://isgw-forum.colorado.edu/letter_avatar_proxy/v4/letter/d/258eb7/32.png) [@davide](https://isgw-forum.colorado.edu/u/davide)\
**Post date:** [June 11, 2026, 11:14am UTC](https://isgw-forum.colorado.edu/t/random-effects-ivw-analysis/892/2 "2026-06-11T11:14:58Z")

</div>

Yes- and some folks actually recommend this as the primary analysis. This will Inflate the standard error on the causal estimate when heterogeneity exceeds expectation and can be useful when there is (balanced) horizontal pleiotropy in the data.