Factors extracted from identical items administered to samples from similar but not identical populations should be comparable if they are to be used as summary measures of the information in the items. Correlation between duplicate factor scores calculated using weights from identical factor analyses of two samples provides a coefficient of factor comparability, which is a more direct measure than the coefficient of congruence based upon factor loadings. Comparability can be defined in either a strong or a weak sense, according to whether the two sets of factors are required to appear in the same order. A study of the work-related attitudes of academics is used to investigate the comparability obtained under alternative models of factor analysis. Principal component and image analyses are found to give similar results, with only the first unrotated factor having consistently high comparability. Accordingly, using a hypothesis-confirming approach, a principal component multiple group analysis provides a set of oblique factors that are not only more interpretable but also more comparable than factors extracted using standard unrotated and rotated analyses of all the items simultaneously.