Drawing an analogy, my task can be phrased like this: Evaluate a person's IQ based on several tests and examine how IQ changes over time.

I do $K$ different intelligence tests for each person several times (Day 1, Day 2,...). I need to find the factor responsible for IQ and estimate how it changes over time.

If I just needed to find IQ scores, I could do an exploratory factor analysis. Find $M$ important experiments from $K$ and calculate the loadings $L_1,L_2...L_M$.

For example, I may find these loading for the day 1 and use for other days, then I will be able to estimate the IQ dynamic. But in this case i will loose a lot of information from Day 2, and 3.

I try to formalise the problem: if i have some factor $F$ and Variables $V_1,V_2,..V_K$

$V_1 = L_1*F + N(\mu_1,\sigma_1)$


$V_K = L_K*F + N(\mu_K,\sigma_K)$

this will lead me to a usual factor analysis.

But my factors evolve during experiments

$F = w*DayN+N(\mu_d,\sigma_d)$

Could you help me to estimate $L_{1-K}$ and $w$?

Additional problem: the number of days vary between observers from 3 to 6.

  • $\begingroup$ If the nature of IQ is changing over time, how can you examine how it changes over time? It's not the same thing. $\endgroup$ Jan 20, 2021 at 19:17

1 Answer 1


A multilevel exploratory factor analysis is the solution. You can check an illustration here. This will also account for the varying numbers of days between participants.


Your Answer

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

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