Difference between revisions of "Test"
Jump to navigation
Jump to search
m (→How to mathematically represent a mixed effects model?) |
|||
Line 17: | Line 17: | ||
Assume that the distribution of the vector of individual parameters $\psi_i$ depends on a vector of individual covariates $\beta c_i$: | Assume that the distribution of the vector of individual parameters $\psi_i$ depends on a vector of individual covariates $\beta c_i$: | ||
− | + | <math> \label{prior2} | |
− | |||
\psi_i \sim \psi(\psi_i | \beta c_i ; \theta) | \psi_i \sim \psi(\psi_i | \beta c_i ; \theta) | ||
− | + | </math> | |
A mixed effects model assumes here that $\psi_i$ can be decomposed as follows: | A mixed effects model assumes here that $\psi_i$ can be decomposed as follows: |
Revision as of 14:50, 25 March 2013
$ \newcommand{\Re}{\mathrm{Re}\,} \newcommand{\pFq}[5]{{}_{#1}\mathrm{F}_{#2} \left( \genfrac{}{}{0pt}{}{#3}{#4} \bigg| {#5} \right)} $
The population distribution of the individual parameters
Introduction
How to mathematically represent a mixed effects model?
Assume that the distribution of the vector of individual parameters $\psi_i$ depends on a vector of individual covariates $\beta c_i$\[ \label{prior2} \psi_i \sim \psi(\psi_i | \beta c_i ; \theta) \]
A mixed effects model assumes here that $\psi_i$ can be decomposed as follows:
\begin{equation} \label{prior3} \psi_i = H(\psi_{pop},\beta, c_i,\eta_i) \end{equation}
where
- $\psi_{pop}$ is a ``typical value of $\psi$ in the population,
- $c_i$ is a vector of covariates,
- $\beta$ is a vector of fixed effects,
- $\eta_i$ is a vector of random effects, i.e. $(\eta_i, 1 \leq i \leq N)$ are random vectors with mean 0.
In this model, part of the inter-individual variability is explained by the covariates $c_i$. The random effects describe the part of IIV which is not explained by the covariates.