ABSTRACT
EVEN the most casual look at Cattell's factors will be sufficient to show that here we are dealing with factors located towards the C end of our C—T continuum; few if any of them are tautological. It would appear, therefore, that C.F.S.s would be likely to be appreciably lower than those observed in the E.P.I., and this deduction from the generalization made in the last section appears to be borne out in fact. Table 17.1 shows the number of primary factors in the Eysenck, Cattell and Guilford inventories which exceed 0·8, are below 0·6, or are intermediate; it will be seen that while 8 of Eysenck's factors have C.F.S.s in excess of 0·8, only one of Cattell's and 4 of Guilford's are in this category. When it is realized that 0·8 is a very lenient criterion indeed for factor comparison (we would prefer 0·9 as a minimum criterion for the assertion that two factors are identical), then it will be clear immediately that as far as Cattell's work is concerned there is very little evidence in our data to suggest that the factors which emerge have any marked degree of replicability. It might be objected that we are comparing male factors with female, and that possibly factors within one sex might be more reproducible than factors between sexes. This is of course possible, and we will soon turn to a comparison of our factors in both sexes with those suggested by Cattell on the basis of his own work; nevertheless it must be pointed out that neither Cattell nor Guilford in fact hold that men and women show different factor patterns. Both use the same items to measure their various primary factors in the two sexes, and although they mention occasional sex differences in scores on these questionnaires, there is no suggestion even of any differences in intercorrelation of items, or of factor patterns. As experimental disproof of the position advocated by Cattell and Guilford, therefore, our demonstration of low C.F.S.s between the sexes is perfectly acceptable; hardly any of the factors actually found in our analysis have sufficiently high C.F.S.s to enable one to say that a factor discovered on the male side has a unique and clear-cut, comparable factor on the female side. About 50 per cent of all factors have C.F.S.s which are below 0·6; none, in the case of Cattell, is above 0·9. This is a most disappointing finding when it is considered that both authors have spent thirty years or more in the production of their respective systems, and have carried out hundreds of factor analytic studies in the express hope of discovering invariant and replicable first-order factors.
|
C.F.S. |
Eysenck |
Cattell |
Guilford |
|---|---|---|---|
|
0·80 and above |
8 |
1 |
4 |
|
0·60–0·79 |
5 |
9 |
8 |
|
Below 0·60 |
7 |
10 |
8 |
|
Total |
20 |
20 |
20 |
