Dear researcher,
I need to compare different linear mixed-effects models. I use the same data set. Sometimes the corporation is between two positive and negative AIC. I do not know if that right.
Note: AIC has some time very high level -1000
thank you in advance
Use conditional AIC and select the model that has the smallest conditional AIC value. The R package "cAIC4" (https://cran.r-project.org/web/packages/cAIC4/cAIC4.pdf) can be used to calculate the conditional AIC.
Article A Note on Conditional AIC for Linear Mixed-Effects Models
Check these articles:
1. Harrison, X. A., Donaldson, L., Correa-Cano, M. E., Evans, J., Fisher, D. N., Goodwin, C., Robinson, B. S., Hodgson, D. J., & Inger, R. (2018). A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ, 6, e4794. https://doi.org/10.7717/peerj.4794
2. 2013, Vol. 28, No. 2, 135–167 DOI: 10.1214/12-STS410 c Institute of Mathematical Statistics, 2013 Model Selection in Linear Mixed Models Samuel M¨uller, J. L. Scealy and A. H. Welsh
3. Heo J, Lee JY, Kim W. Bayesian information criterion accounting for the number of covariance parameters in mixed effects models. CSAM 2020;27:301-311. https://doi.org/10.29220/CSAM.2020.27.3.301
Dima Sabboura normally the lower AIC or AICc (for small n), the better the model and a difference of >5 numerical value is the benchmark, for instance a model with -3 AIC value is preferred over a model with 3 AIC value (numerical difference is 6 , >5) regardless of the quantum of the value.
see
https://www.graphpad.com/guides/prism/latest/curve-fitting/reg_how_the_aicc_computations_work.htm
Oyekola Oluyimika Oloyede Thank you so much. The first study is so useful.
- Badges
- Science topic