325

Printed in Great Britain Br. J. med. Psychol. (19776). 49, 325-328

A factor analytic study of the Dynamic Personality Inventory using a psychiatric population BY ERIC BROMLEY* A N D LEONARD A. LEWISt

Kline (1%8) and Stringer (1970) have presented evidence pertaining to the factorial validity of Grygier’s (1%1) Dynamic Personality Inventory (DPI).The present paper presents a further factor analysis on a different type of sample than those used by either Kline or Stringer. The sample consisted of 175 (100 female and 75 male) psychiatric patients, who had been given the DPI as part of a clinical psychological assessment. The sample would thus seem to have several advantages over the samples used by Kline or Stringer. Firstly, the sample would be more likely to be similar to ones normally used in the clinical administration of the DPI. Secondly, as the administration was part of a clinical assessment, then the ‘response set’ is similarly more likely to be that met with in the ordinary use of the DPI. The use of a psychiatric sample might also be considered useful as an attempt at ‘factor matching across samples’ (Stringer)and in addition meets one of the criticisms he makes of Kline’s sample - namely that it is too small. The present sample has the same sex ratio as that of Kline’s. The method of analysis was precisely the same as that employed by Kline. Factors with eigenvalues in excess of one were extracted and rotated to Varimax criterion. Loadings greater than or equal to 0.300 were assumed to be salient. As Stringer has pointed out, this method is to some extent arbitrary, but was used in order to provide more clearly comparable data to that provided by Kline. The comparisons presented in this paper are with Kline’s solution,as Stringer in his main analysis produced an oblique factor solution which would be expected to give different, non-comparable results. RESULTS

Eight factors with eigenvalues greater than one emerged. The result of rotating these to Varimax criterion is shown in Table 1. The factor accounting for the maximum variance (Factor I) is unipolar and has negative loadings on all of the ‘phallic’ scales, except Pe (Exhibitionism)and Pn (Narcissism),together with Ws (Seclusion and introspection),OA (Oral aggression), CI (Creative interests),TI (Tactile interests) and M (Masculine identification). It is a clearer cut ‘phallic’ factor than Kline found although similar to Kline’s second factor (which loaded negatively on 0, P, Pa, Pi, Ws and M). Factor I1 is unipolar, having negative loadings on H (Hypocrisy); Ov (Verbal aggression); Oi (Impulsiveness);Pe (Exhibitionism);Pa (Ascension);Pi (Icarian exploits);CI (Creativeinterests); M (Masculine identification);SA (Social activities); EP (Ego persistence); EI (Initiative)and C (Interest in children). This factor seems very difficult to interpret meaningfully, although it might be said to have certain qualities of an ‘Extraversion’factor. Factor I11 is identical to Kline’s third factor, having negative loadings on Wp (Passivity); 0 (Orality); Od (Oral dependence);Pn (Narcissism); F (Feminine identification);SA (Social activities); and C (Interest in children). This factor would seem to have some of the loadings one would expect in an ‘oral dependency* factor. Factor IV is the first bipolar factor to emerge, having positive loadings on Ws (Seclusion);Om (Need for freedom of movement); Oi (Impulsiveness);Ou (Unconventionality);CI (Creative * Principal Clinical Psychologist, Rainhill Hospital. t Principal Clinical Psychologist, Winwick Hospital, now retired.

326

ERIC BROMLEY A N D L E O N A R D A. L E W I S

Table 1. Varimax rotated factors of the DPI with psychiatric patients ~

Factors Scale Aa Ac Ad Ah

Ai As C CI EI EP

F H 0 OA

od

Oi

Om

ou ov P Pa Pe Pf Ph Pi Pn

S SA TI

WP

ws M

I 061 155 076 -096

001 -037 - 107 -336 -201 061 015

-117 - 146 -308 -034 -057 -276 - I80 -093 -806 -670 -119 -712 -722 -746

OOO -298 -099 -437 082 -397 -526

I1

111

IV

V

VI

- 120

-069 - I27 003 -079 210 -084 -514 -229 073 050 -779 - I72 -620 -276 -635 -084 019 -045 243 -201 011 -292 120 - 148 -032 -715 109 -431 -236 -433 058 I59

- I93

-801 -298

- 135

080 - I49 -080 267 - I97 -334 -388 -776 -530 -097 -472 -010 043 -073 -303 -050 -205 -559 - I05 -345 -578 -045 -006 -309

082 -229 -686 -110 035 098 -687

-783 -236 -013 074 -054 - I58 498 138 -01 1 198 - 192 - 137

-471 -026

-009

-590

-021

-823 053 146 -053 - 105

014 -442 039

090

015

-395 473 778 423 249 -012 25 1

-077 -057 054 385

284 200 047 217 226 -064 -078 026 - 125 314 001

-040

-066 -094

173 -029 010 07 1 046 - 148 45 1

-049 141 019 -090 -092

155 048

-377 - 191

-028 025 -245 -257 017 466 033 195 171 560

542 053 039 181 2% 138 -092 137 546 105 -423 135 014 -057

VII 056 159 550

763

084 -005 07 1 325 127 295 26 1 326 -045 -078 134 -313 066

185 091 100 - 105 -077 -030 685 030 -085

040 - 143

398 092 35 1 092

VIII 024 086 - I24 051 413 -072 - 161 092 -035 -367 - 140 -247

187 073 150

274 -072 -037 -003 033 159 03 1 - I07 -261 082 036 254 -027 287 639 253 152

interests) and negative loadings on Od (Oral dependence) and Ac (Conservatism). It is similar to Kline’s eighth factor which has an additional negative loading on Pn (Narcissism) and lacks loadings on Ws and CI. Factor V is bipolar, loading positively on Ou (Unconventionality) and S (Sexuality) and negatively on H (Hypocrisy) and all the anal scales except Ah (if we accept as salient the Ac loading of -0.299). It is similar to Kline’s first factor with perhaps clearer ‘anal’ loadings. The final three factors to emerge are different from any of Kline’s. Factor VI is bipolar with positive loadings on OA (Oral aggression), Ov (Verbal aggression), Ou (Unconventionality) and S (Sexuality) and negative loadings on TI (Tactile impressions) and C (Interest in children). Factor VII is bipolar, with one negative loading on Oi (Impulsiveness) and positive loadings on H (Hypocrisy);Ws (Seclusion); Ah (Hoarding); Ad (Attention to detail);TI (Tactile interests) and CI (Creative interests).

A factor analytic study of the DPI

327

Factor VIII, the final factor to emerge, is bipolar with positive loadings on Wp (Passivity)and Ai (Insularity)and one negative loading on EP (Ego persistence). DISCUSSION OF RESULTS

At least three factors are clearly interpretable in psychoanalytic terms and interestingly enough they are three factors which are identical, or very similar, in terms of factor loadings, across the two samples of this paper and Kline’s. Factor I (Kline’s second factor) is a clear-cut ‘Phallic’factor (missing only Pe and Pn among the phallic scales in this sample). The loadings on CI and M are perfectly acceptable in psychoanalytic terms as, for instance, the person who is relatively fixated at a phallic stage is more likely to show masculine, creative interests than somebody relatively fixated at earlier levels. However, it is not immediately obvious why this factor should also have loadings on Ws (Seclusion and introspection) and TI (Tactile interests). It might perhaps be argued that the loading on Ws represents a tendency not to take part in adult, genital interpersonal relationships, but rather to seek satisfaction within oneself (a type of auto-erotic tendency). Some comment must be made at this point on Kline’s suggestions that the loading on Masculine interests (M) would not be expected and that the best label for the factor would be one of Masculine interests. It seems to us a perfectly reasonable step to suggest that ‘phallic attributes. . .lead to the assumption of masculine roles’. Incidentally, we performed a further factor analysis using exactly the same method except that the sex of the subject was added as a variable. The same factor emerged and there was no loading for ‘sex’ on it. This perhaps goes some way to showing that it is not purely an artefact of using a mixed sex sample (a criticism which Stringer makes). Factor I11 (Kline’s factor eight) is to some extent a reasonable ‘oral’ factor at least in that it shows a scale having items on liking milky, creamy foods etc. (0),as being linked with various This seems scales having items relating to passivity, dependence,need for other people, etc. (Od). to us to go some way to giving credibility to Freudian ideas of oral dependency. Again we comment on Kline’s statement that it is best to regard this factor as a femininity factor as otherwise we would have to assume that ‘oral attributes. . .lead to the assumption of feminine roles’. Surely this is just what we would assume. Oral dependency leads to passivity and interpersonal dependency and (socially imposed)feminineroles are surely essentially passive and dependent. The factor analysis including ‘sex’ as a variable gave slightly different factor loadings, incorporating S (Sexuality) and not incorporating F, SA and C. There was no loading of this factor on sex. Factor V (Kline’s first factor) is clearly an ‘Anal’ factor needing only to assume that anal fixation leads to conventionality and the inhibition of sexual expression. In view of the lack of a loading on Ah (hoarding), it might suggest that this particular scale is not ‘valid’ in the sense of sharing factorial similarity with the other anal scales. The other factors which emerge are less clearly linked to psychoanalytic concepts of personality. Factor VI is interesting in that OA (Oral aggression) is linked with Ov (Verbal aggression), giving perhaps some credibility to the psychoanalytic ideas of oral sadistic fixation in the sense of liking biting, etc., being linked with verbal aggression. It would seem then that the factor analysis of the DPI gives rise to some results which are reasonably consistent with psychoanalytic theory. The analysis in no sense ‘proves’ psychoanalytic theory, but then it is hard to see how it could. The DPI as a clinical instrument is perhaps strengthened as a result of the various factor analyses. But glaring deficiencies still remain. One of these deficiencies is the ‘degree of item overlap, which on some scales is considerable. Another deficiency is that the vast majority of the items are scored as addding to a scale score if answered in the ‘like’ direction. It is probable

328

ERIC BROMLEY AND LEONARD A. LEWIS

that this will give rise to spurious factor loading if, as seems at least plausible, there is a tendency to like or dislike things in general. This tendency is illustrated statistically by the fact that, in the correlation matrix, of 496 correlations only 25 are significantly negative ( P < 0.05) compared to 124 significant in the positive direction. It is probable that some of the factors are contaminated by this effect. This might be particularly pertinent with unipolar factors having large numbers of factors loading on them. SUMMARY AND CONCLUSIONS

A factor analysis of the DPI based on a psychiatric sample of 175 patients is presented. Some of the problems of factor validity in relation to scales purporting to measure psychoanalytic variables are discussed. Comparisons are drawn between this factor analysis and that of Wine. It is maintained that at least three factors emerge reasonably consistently and can be labelled ‘Oral dependency ’, ‘Anal’ and ‘Phallic’scales. A factor emerging on this analysis only might reasonably be labelled an ‘Oral aggressive’ factor. The results seem moderately promising in terms of the validity of the DPI in a clinical setting. Development in terms of removing item overlap and reshaping the items to attempt to remove a very strong ‘yea saying’ effect might be worth attempting. REFERENCES

GRYGIER, T. C. (l%l). The Dynamic Personality Inventory. London: NFER. KLINE,P. (1968). The validity of the Dynamic Personality Inventory. Br. j . med. Psychol. 41, 307-3 13.

STRINGER, P. (1970). A noteon thefactorial structure of the Dynamic Personality Inventory. Br. J. med. Psychol. 43,95-103.

A factor analytic study of the Dynamic Personality Inventory using a psychiatric population.

325 Printed in Great Britain Br. J. med. Psychol. (19776). 49, 325-328 A factor analytic study of the Dynamic Personality Inventory using a psychiat...
247KB Sizes 0 Downloads 0 Views