# AIC/BIC for TwoStage Model

**URL:** <https://discourse.pumas.ai/t/aic-bic-for-twostage-model/740>\
**Category:** maximum-likelihood\
**Created:** [August 3, 2022, 8:23pm UTC](https://discourse.pumas.ai/t/aic-bic-for-twostage-model/740 "2022-08-03T20:23:12Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Rahulub3r](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.pumas.ai/rahulub3r/32/182_2.png) [@Rahulub3r](https://discourse.pumas.ai/u/Rahulub3r)\
**Post date:** [August 3, 2022, 8:23pm UTC](https://discourse.pumas.ai/t/aic-bic-for-twostage-model/740/1 "2022-08-03T20:23:12Z")

</div>

For example I have a dataset with 100 patients and I performed a two stage fitting using a) 1-compartment model b) 2-compartment model. To calculate the total loglikelihood (LL), I can just add the LLs for all 100 fits that I get and compare a and b, but I have a feeling that I we cannot add the BICs or AICs like LLs for all fits. Is there a rational way to get a “total BIC/AIC”? Thank you!

---

<div class="post-metadata">

**Author:** ![vijay](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.pumas.ai/vijay/32/96_2.png) [@vijay](https://discourse.pumas.ai/u/vijay)\
**Post date:** [August 10, 2022, 12:05am UTC](https://discourse.pumas.ai/t/aic-bic-for-twostage-model/740/2 "2022-08-10T00:05:01Z")

</div>

I don’t think AIC/BIC add up as the individual log-likelihoods do

---

<div class="post-metadata">

**Author:** ![Rahulub3r](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.pumas.ai/rahulub3r/32/182_2.png) [@Rahulub3r](https://discourse.pumas.ai/u/Rahulub3r)\
**Post date:** [September 17, 2022, 1:08pm UTC](https://discourse.pumas.ai/t/aic-bic-for-twostage-model/740/3 "2022-09-17T13:08:10Z")

</div>

Yes, I figured that we can rather do a paired t-test using individual AIC/BIC to compare between models in a two-stage approach rather than trying to get a total value.  
Thanks