4
$\begingroup$

What is the difference between Hierarchical GAMs (HGAM) and Mixed GAMs (GAMM), if any?

I am looking to model time series of count data against a range of candidate explanatory variables (hoping to understand which environmental parameters could explain the fluctuations). I have a year of data at 7 sites, quite different from each other in terms of environment, but not completely independent (species can easily move from one to another).

I struggle to identify which of the two frameworks would be the most appropriate. A HGAM with grouping per site, or a GAMM with site as a random variable... Is there any key difference I am missing here?

Thank you for any advice!

$\endgroup$

1 Answer 1

6
$\begingroup$

They are the same thing; we just prefer the terminology "hierarchical" over "mixed", because the salient practical feature of these models is that they can model variation in the response that occurs at multiple levels, rather than the fact that they have fixed and random effects.

$\endgroup$
3
  • $\begingroup$ That clarifies it ! Thank you very much ! Then I can probably specify a model using all the data, allowing the smooths to vary across sites, without the need to fit one model per site... Is that correct ? $\endgroup$
    – Timelate
    Commented Oct 17, 2022 at 8:54
  • 1
    $\begingroup$ Yes, that's exactly the use case we cover in the HGAM paper $\endgroup$ Commented Oct 17, 2022 at 14:47
  • 1
    $\begingroup$ Perfect ! Thank you very much ! $\endgroup$
    – Timelate
    Commented Oct 18, 2022 at 10:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.