Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm confused by the description of the intraclass correlation (ICC) for a linear mixed model with longitudinal data from this material. A screen capture is shown below.

What I'm confused about is that the ICC describes the amount of variation in $Y$ (the dependent variable) BETWEEN individuals, in this context (longitudinal data) but then also is the correlation between observations (which I think of as the correlation between an observation for a given individual at time $t$ and that same individual at time $t+k$). This correlation matrix seems to describe correlation within an individual.

Why does this make sense?

enter image description here

share|improve this question
up vote 5 down vote accepted

Note: the discussion below is largely copy/pasted from my answers in the following threads:

Our model for the $i$th response by the $j$th individual is $$ y_{ij} = \beta_0 + u_{0j} + e_{ij}, $$ where the random intercepts $u_{0j}$ have variance $\sigma^2_{u_0}$ and the residuals $e_{ij}$ have variance $\sigma^2_e$.

Now, the correlation between two random variables $x$ and $y$ is defined as $$ corr = \frac{cov(x, y)}{\sqrt{var(x)var(y)}}. $$

So to find the formula for intra-class correlation, we use the correlation formula and let our two random variables be two observations drawn from the same individual $j$, $$ ICC = \frac{cov(\beta_0 + u_{0j} + e_{1j}, \beta_0 + u_{0j} + e_{2j})}{\sqrt{var(\beta_0 + u_{0j} + e_{1j})var(\beta_0 + u_{0j} + e_{2j})}}, $$ and if you simplify this using the definitions given above and the properties of variances/covariances, you end up with $$ ICC = \frac{\sigma^2_{u_0}}{\sigma^2_{u_0} + \sigma^2_e}. $$ So in this simple case, the ICC works out to be a simple proportion of variance. However it is important to note that this is not always true in general. But what will always be true is that the ICC can be interpreted as the correlation between two (appropriately defined) observations in the dataset. That is the interpretation that is primary.

share|improve this answer
Very nice answer. So, just to confirm, the ICC is always interpreted as the correlation between two (appropriately defined) observations (from the same cluster or individual) in the dataset and works for the partitioning of variation in the simple case. Is that right? – B_Miner Jan 7 '14 at 4:12
@B_Miner Yes, right. I mainly add the "appropriately defined" qualification just to point out that, in cases of more complicated datasets and correspondingly more complicated models, there are many ways to define an ICC depending on what pairing of observations you are interested in. – Jake Westfall Jan 7 '14 at 4:15

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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