0
$\begingroup$

I've been searching for a way to approach my problem.

This is a scenario from a Multivariate Statistics Assignment on Confirmatory Factor Analysis.

We've been given only a correlation matrix on 6 variables, the Standard Deviations and sample size. The model assumes that $X1$, $X2$ and $X$3 are indicators of given Factor $F1$, while $X4$, $X5$ and $X6$ are indicators of another Factor $F2$ (we actually know which factors and variables those are, but since my question is theoretical, I am keeping this simpler).

I've already evaluated the measurement model, found it has 7 degrees of freedom and moved on to the method's assumptions.

Now, assuming I want to do the Factor Extraction step using the ML estimator, which assumes data normality: is there any way I can test for my sample normality with the provided information?

I appreciate any small insights and/or directions that may help!

$\endgroup$
  • $\begingroup$ Re the title: adding in information about the means wouldn't help at all; you still have nothing concrete about the shape of the joint distribution. $\endgroup$ – Glen_b -Reinstate Monica Oct 26 '19 at 6:34
3
$\begingroup$

No. You cannot test for normality only knowing the first two moments of a distribution (correlation, covariance, mean, and SDs are all first-and-second including product moment estimators). You need to know the entire sample to test for goodness of fit for a particular distribution. At the very least, you would need to know skewness and kurtosis to exclude some specific violations of normality.

$\endgroup$
  • $\begingroup$ thank you for the input $\endgroup$ – Pedro Alonso Oct 26 '19 at 5:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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