1
$\begingroup$

My factor analysis using direct oblimin keeps giving me only partial results, stating 'Attempted to extract 2 factors. In iteration 25, the communality of a variable exceeded 1.0. Extraction was terminated.' Having inputted 7 items, two of which were reverse coded, which are together supposed to measure an attitude. However, my SPSS output shows the aforementioned statetement, only the initial communicatielies, and neither the factor nor the pattern matrix. Would anyone know what could be the solution?

$\endgroup$
1
  • 1
    $\begingroup$ This is what is sometimes referred to as a "Haywood case". Good places to start, in terms of guessing what's gone wrong, are: 1) you have too little data and/or 2) your indicators badly violate the assumptions of your estimator (guessing ML, and so multivariate normality). $\endgroup$
    – jsakaluk
    Commented Apr 25, 2019 at 15:59

1 Answer 1

2
$\begingroup$

You don't say what extraction method you're using, though as @jsakaluk implies, it's commonly maximum likelihood estimation where Heywood cases, or negative estimated residual variances, occur. This can occur for various reasons, including not enough data, too few or too many factors specified, and a generally inappropriate model. In some cases the problem may not occur if a different extraction method is used.

$\endgroup$

Your Answer

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

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