RMSEA
Message posted by kathleen ashton (via 216.67.38.67) on September 9, 2001 at 3:37 PM (ET)

Could you explain to me what the RMSEA (root mean square error of approximation) exactly is and how it works as an indicator of fit in factor analysis? Thanks.

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: RMSEA
Message posted by Tomi (via 154.32.143.62) on September 10, 2001 at 3:09 AM (ET)

First, you calculate the values predicted by your model (the fitted values).
Then you calculate the difference between the observed values and the fitted values. This difference is the error.

Now, you have some positive errors and some negative errors, so simply adding them up to find the total error is a bad idea, because they will cancel themselves out. So we square them, add them all together and divide by the number that there are. This gives the mean square error (MSE). RMSE is the square root of the MSE.

In any comparison of models, the model with the lowest MSE or RMSE is probably the better one (although other considerations may well arise).