Person Pnnted
in&tcf. Dif. Vol 6, No 2. pp. 217-221. in Great Britan. All rights reserved
NATIONAL
Department
1985
0191-8869185 S3.00 + 0.00 a(’ 1985 Pergamon Press Ltd
CopyrIght
DIFFERENCES IN PERSONALITY: INDIA AND ENGLAND
PITTU LAUNCANI of Social Sciences, Polytechnic of the South London SE1 OAA, England
Bank,
Borough
Road,
(Received 10 MUJJ 1984) Summary-This study reports comparisons between results on the Eysenck Personality Questionnaire (EPQ) obtained from 558 male and 636 female Indian Ss; the comparison group consisted of the original standardized English Sample (Eysenck and Eysenck, 1975). Factor analysis of item correlations showed close correspondence between the factors extracted in the two samples, with indices of factor comparisons being well in excess of 0.97. The reliabilities in the Indian sample were only marginally lower than in the original English standardization group. Some items from the original EPQ did not load on the hypothesized factors; they were substituted by the extra items contained in the IO]-item version of the EPQ used in India. The Indian data showed no sex differences with respect to extraversion; females were found to score higher on the Neuroticism and also slightly higher on the Lie scale than the males. The Lie scores for both males and females being quite high suggests a consistent trait of conformity in operation. It was concluded that the organization of personality in India is sufficiently similar to that in England to make national comparisons feasible since identical dimensions underlie the personalities of Indian and English Ss.
INTRODUCTION
The literature on the subject of national differences in personality is quite extensive. The earlier studies in this area have, however, tended to use standard questionnaires of personality (the assumed-etic-approach) which have been administered to carefully selected large samples of residents of those countries. Their scores have then been compared with those given by the original standardization groups. Several such studies using the above methodology have been reported (e.g. Devadasan, 1964; Honess and Kline, 1974; Hosseini, Mehryar and Razavieh, 1973; Kline, 1967; Mehryar, Khajavi and Hekmant, 1975; Orpen, 1972; Sen, 1966). These methods are simplistic and are open to several criticisms, the major one being that the meanings of certain actions or attitudes canvassed in personality questionnaires may change completely as one passes from one country, and one culture, to another. The measures become even more vulnerable to criticism when they are translated into local languages. Brislin (1976) has pointed out that literal translations are virtually meaningless when idioms are involved; also, it is hard to be certain whether the translated test is conceptually equivalent and whether the concepts which the Ss are instructed to rate are perceived in the same manner and have the same meaning in that culture. As Eysenck and Eysenck (1975) point out, psychometric measures are easy enough to obtain, but the meaning of those scores is by no means easy to decipher. They suggest that differences in test scores between two cultural or national groups cannot really be taken as evidence of genuine personality differences between the two groups. Over the past few years they have experimented with a method which seems capable of overcoming this difficulty. The neat solution which they provide is based on an extension of the psychometric method used to provide internal validation for trait or type questionnaires within a given culture. The method which they advocate is an extension of the technique which they have used for the construction of measures of extraversion, neuroticism and psychoticism. To start with, having postulated a given trait or type, items are written which express the putative nature of the trait/type, as expressed in a variety of situations. The next step is to administer questionnaires containing numbers of such items to suitable populations, correlate the item answers and carry out a factor analysis. The results of the factor analysis would allow the investigator to decide whether to reject the postulated factor as being non-existent (it is of course possible that the items selected for the measurement of the postulated factor are unsuitable, in which case, new items would have to be written), or that something similar but not identical to the postulated factor has emerged from the 217
data. Factor analysis would also be able to pick up psychometrically ‘bad’ items which could then be excluded. It would then be possible to write new items and repeat the whole process as many times as it is necessary to achieve a satisfactory scale. Eysenck and Eysenck (1975) suggest that the above method could easily be used in cross-cultural or cross-national investigations of personality. With the use of this method it would be possible to determine whether certain personality factors can be measured meaningfully in a different country or culture. Obviously a set of rules would have to be followed to ensure scientific objectivity. Firstly, the constructed scale would have to be administered to residents in the country of origin (in our case England) and to the residents of the country to be compared. Secondly, care must be taken to ensure that suitable populations have been selected for testing (selected for age, sex, social status, education etc.). They recommend that the scores on the tests given to the two national or cultural groups are comparable, “if and only IY, the correlation matrices between items are identical (or closely similar) between the two countries” (Eysenck and Eysenck, 1981) (Eysencks’ italics). A related test would be to compare the factor loadings of the items using the factor comparison techniques discussed in Eysenck and Eysenck (1969). They suggest that coefficients of factor comparisons of above 0.95 might be required to accept test results as properly comparable. This method of investigation can lead to three possible outcomes. There might be perfect comparability in which case the coefficient of factor comparison would be 1.OO. On the other hand there might be a fair degree of comparability, with distinct exceptions; and lastly. there might be a total lack of comparability. Eysenck and his coworke, s have found almost perfect comparability in studies which have compared foreign groups both. culturally quite close to the English, e.g. New Zealand (Saklofske and Eysenck, 1978) and also quite distant, e.g. Greece (Dimitriou and Eysenck, 1978) and India (Eysenck, Gupta and Eysenck, in press). Complete lack of comparability has never been found by them. Most of their studies have in fact produced results which show good agreement. From time to time there have been some items which failed to show even reasonably similar factor leadings on what were clearly identical factors. This is because the items concerned have probably been culturally irrelevant. The existence of such items can in fact be seen as an advantage. If they are capable of some rational explanation they might advance our understanding of national differences and similarities in personality. EXPERIMENT
AND
ANALYSIS
The experiment to be reported consists essentially of a detailed comparison of factor loadings between the English standardization of the Eysenck Personality Questionnaire (EPQ; Eysenck and Eysenck, 1975) and a group of Indian male and female students. The experiment was undertaken in Bombay, perhaps the most cosmopolitan city in India. Undergraduate students from the University of Bombay served as Ss for the study. They were given the EPQ in English. The decision to use the EPQ in English was based on the following considerations.
(a)
(b)
Bombay is a large, metropolitan city inhabited by several multilingual, heterogeneous communities. The heterogeneity which extends along several dimensions such as, religion, caste, local customs, language etc.. made it practically impossible (and theoretically undesirable) to single out any particular community for the purpose of the investigation. In the colleges affiliated to the University of Bombay, all the teaching is done in English, thus making English the common language of communication among the student fraternity.
The Indian sample consisted of 1194 Ss of whom 558 were males, and 636 females. They were all undergraduate students randomly selected across all faculties from colleges which were affiliated to the University of Bombay. Care was taken to select colleges which were representative of the entire geographical area of Bombay. The average age for both the groups was 18.3 yr. The lOl-item version of the EPQ was administered to the Indian Ss instead of the standardized 90-item questionnaire. The students were tested in small groups of about 30 each. The English sample used
Differences in personahty: India and England
219
for the factor analysis consisted of 500 men and 500 women, taken from samples 5-7 times larger and used to established means and standard deviations for the scores of the different scales which go to make up the EPQ. These scales are: P (Psychoticism), E (Extraversion), N (Neuroticism) and L (a Lie or dissimulation scale). Of the several different techniques of factor analysis, the one which Eysenck and his associates have followed has been based on factoring the product-moment correlations between items by the principal-components methods, rotated by Varimax (see Kaiser, 1958) and then obliquely, by Promax (see Hendrickson and White, 1964), taking the first four factors only for rotational purposes. The rationale and advantages of following the above approach have been discussed by Eysenck and Eysenck (1969, 1976). The Indian data have been analysed in accordance with the guidelines provided by Eysenck and Eysenck (1969). RESULTS
The factor loadings on the items of the questionnaire for the Indian data for both males and females show a striking similarity with the standardized English loadings. The Indian factor loadings may be obtained from the author. The English loadings are given in Eysenck and Eysenck (1976). The actual items used are given in the Appendix. However, not all the items on all the four scales loaded in the expected direction. Obviously there are differences in the pattern of loadings on the P factor between the Indian and the English data. Rational explanations of such differences would clearly enhance our understanding of national differences and similarities in personality. Let us consider Item 9, which pertains to locking up one’s house carefully at night. In a study in Greece, Dimitriou and Eysenck (1978) found that that particular item did not load on the P factor. They argued that because of the heat in Greece people tend to sleep with their windows wide open, and in that case, there seems hardly any point in keeping doors carefully locked at night. This explanation could be extended to include India where the heat forces people to leave their windows wide open in the night. With regard to Item 19, which is related to insurance schemes, two explanations are readily available. Firstly, in an extended family network, decisions of such a nature are taken by the head of the family, and other members are seldom consulted over the matter; secondly, only a very small percentage of the population in India could afford to pay the required premiums for insurance policies. Finally, Item 63 which relates to arriving at appointments in plenty of time is also amenable to a rational explanation. Indians in general have a different concept of. and approach to, time from Westerners; time in India is seen in cosmic terms; even one’s existence is seen as one of many that were and are to be; it clearly does not have the same connotation which it does for people in Western countries. On the whole, however, the P-factor loadings bear a striking similarity to the standardized English data, thus suggesting that the factor can be identified without doubt as one of psychoticism. On the E factor only three items are unsatisfactory. Item 22, which does not load on the E factor is, perhaps, not a good measure of extraversion. because: (a) most social occasions are gatherings of family members and close relatives and are often related to the performance of religious and/or social ceremonies in which every member of the household is expected to participate; and (b) overcrowding in large cities combined with having to live in cramped and congested conditions at home does not always make it possible for an individual to keep in the background should the person wish to do so. Item 30 too, is perhaps not a good indicator of extraversion. Firstly, Indian society is organized largely as a community-based society; one is, for the most time, surrounded by people. Secondly, the extended family network does not allow much room for privacy. This is exacerbated by the fact that living conditions in the urban areas are generally extremely congested and it is not uncommon to find 34 persons in a family sharing a single room. Secondly, reading is not normally encouraged as an activity in most families. Thirdly, one does not have the easily available facilities of lending-libraries to promote reading. Lastly, books are expensive in relative terms and the purchase of books is not seen as a priority in most homes. In general, however, it is clear from an examination of the factor loadings that the items do indeed measure extraversion as postulated, and the loadings correspond fairly closely with the standardized English data.
On factor N with the exception of three items the pattern of loadings corresponds fairly closely with the standardized English data. Furthermore, two additional items (one from the P factor and one from the ‘spare’ items of the lOl-item version of the EPQ) load on the N factor. The scale is for all practical purposes identical with the original one. On factor L six items have lost their loadings. This loss, however, is made up for by seven other items (three from the P factor, and four from the ‘spare’ items) which load on the L factor. In genera1 it is clear that the items do indeed measure the L scale as postulated, and for all practical purposes the scale is identical with the original one. We must now turn to the reliabilities of the scales, these are given in Table 1. It will be seen that with the exception of the P scale, the remaining three scales show reliabilities of well above 0.75. This is true of both males and females. The P scale has been rather problematic. In an independent study in India, Eysenck et al. (in press) found relatively low reliabilities for the P scale. The authors confessed that some of the items on the P scale were inappropriate in an Indian setting, and they urged that for an adequate measurement of P it was incumbent upon Indian psychologists working there to suggest hypotheses regarding the choice of appropriate substitute items to those which were found to have low loadings on the Indian sample. The satisfactory reliabilities do suggest that the scales measure something very similar to the factors measured in the English population. This is borne out by the indices of factor comparison, which are given in Table 2; all are above 0.97 thus indicating virtual identity of factors between the two nations under comparison. Table 3 shows the intercorrelations between factors (a) and between scales (b), It can be seen from the table that the NL correlations are rather high. It was pointed out earlier that in so far as the NL correlations are concerned the factor of dissimulation cannot be entirely ruled out. High NL correlations are not unique to this particular study: they have been reported in studies undertaken in Hungary, Hong Kong, India, Spain and Sicily. Finally, we shall examine the means and standard deviations of the Indian sample (derived separately for males and females) on the P, E, N and L scales. They are to be found in Table 4. It should be pointed out that the means and standard deviations of the Indian sample cannot be compared with those of the English standardized data (Eysenck and Eysenck, 1976). Different Table
I. Internal
reliabilities
Male (N = 558) Female (N = 636)
Table 2. lndms
of P. E. N and L scales
P
E
N
L
0 52 0.53
0.78 0.77
0 76 0.79
0.78 0.77
of factor comparwms
Indian M vs English M Indian F YS English F Indian M vs Indian F
Table 3. Intercorrelations (a)
PE PN PL EN EL NL
E
N
L
0 979 0.990 0.999
0.989 0 990 0.992
0979 0.987 0.994
of factors and scales
Factors
M
F
-0.06 0.05 -0.11 PO.26 0.19 -0.29
0.07 - 0.05 0.14 - 0.22 0.13 -0.33
Table 4. Means and standard P
Male FC?llXik
P 0972 0.978 0.983
(b) Scales
PE PN PL EN EL NL
deviatmns
M
F
0.03 0.04 -0 I6 -0.36 0.20 -0.31
0 07 0 02 0.00 -0.26 0.14 PO.32
on P. E. n and L scales
E
N
L
n
SD
,?
SD
y
SD
,?
SD
4.02 3.39
2.32 2.17
I2 53 12.68
3.58 3 45
8.94 10.38
3.96 4.13
8 57 8 96
3.54 3.45
Differences
in personality:
India and England
221
scoring keys have been used for the study. The two sets of keys do not contain the same number of items; also some of the items which purport to measure a given factor in one culture belong to another factor in the other culture. If meaningful comparisons are to be made a reduced weight matrix incorporating all the common items to both the groups would have to be used. For the purposes of this study this has not been found necessary. Our interests, in the main, have been to determine if the factor structures postulated by Eysenck emerge in India. This, the study has demonstrated with great clarity. Table 4 shows that there are no sex differences with respect to E; females tend to score higher on N and also slightly higher on the L scale than the males. The L scores for both males and females are quite high. This would suggest that in addition to the factor of dissimulation which would tend to inflate L scores (Michaelis and Eysenck, 1971) there may well be a consistent trait of conformity in operation with respect to the Indian data. In fact, Eysenck et al. (in press), commenting on their own findings in India, suggest that it may be very likely that Indians are more conforming than English people; this finding has been independently corrborated in a study by Laungani (1982) who used a modified version of the Asch (1952) technique to measure conformity. The detailed analysis of the data lends support to the hypothesis that identical dimensions underlie the personalities of Indian and English Ss. The study has also established the robustness of the EQP which, with some minor modifications, can be used effectively with the English-speaking Ss in India. REFERENCES Asch S. (1952) Social Psychology. Prentice-Hall, New York. Brislin R. W. (Ed.) (1976) Translation: Applications and Research. WileyiHalstead, New York. Devadasan K. (1964) Cross-cultural validity of twelve clinical diagnostic tests. J. Ind. Acad. appl. Psychol. 1, 55-57. Dimitriou E. C. and Eysenck S. B. G. (1978) National differences in personality: Greece and England. Int. J. mterrulr. Relar. 2, 26&282. Eysenck H. J. and Eysenck S. B. G. (1969) Peraonalify Structure and Measurement. Routledge & Kegan Paul. London. Eysenck H. J. and Eysenck S. B. G. (1975) Manual of the Eysenck Personality Questionnaire (Junior and Adult). Hodder & Stoughton, London. Eysenck H. J. and Eysenck S. B. G. (1976) Psychoticism as a Dimension of Personality. Hodder & Stoughton. London. Eysenck H. J. and Eysenck S. B. G. (1981) Culture and personality abnormalities. In Culture andpsychoaparholog>, (Edited by Al-Issa). Univ. Park Press, New York. Eysenck H. J., Gupta B. S. and Eysenck S. B. G.-National differences in personality: India and England. Ps~~chol. Res. J. (India) In press. Hendrickson A. E. and White P. 0. (1964) Promax: a quick method for the rotation to oblique simple structure. Br. J. sfatisr.
Honess
Psychol.
11, 65-70.
T. and Kline P. (1974) The use of the EPI and the JEPI with a student
Psychol.
population
in Uganda.
Br. J.
WC. CC/W.
13, 96-98.
Hosseini A. A., Mehryar A. H. and Razavieh A. (1973) Extraversion, neuroticism, and psychoticism as measured by Eysenck’s inventory in Iran. J. genef. Psychol. 122, 197-205. Kaiser H. F. (1958) The varimax criterion for analytic rotation in factor analysis. Psychomefrica 23, 187T200. Kline P. (1967) Extraversion, Neuroticism and academic performance among Ghanaian university students. Br. J. edlrc. Psychol.
70, 381-387.
Laungani P. (1982) A cross-cultural study of personality and conformity in England and India-a theoretical and experimental analysis. Unpublished Ph.D. Thesis, Univ. of London, Mehryar A. H., Khajavi F. and Hekmant H. (1975) Comparison of Eysenck’s PEN and Lanyon’s Psychological Screemng Inventory in a group of American students. J. consult. clin. Psychol. 43, 9-12. Michaelis W. and Eysenck H. J. (1971) The determination of personality inventory factor patterns and mtercorrelations by changes in real life motivation. J. genet. Psychol. 118, 223-234. Orpen C. (1972) The cross-cultural validity of the Eysenck Personality Inventory: a test in Afrikaans speaking South Africa. Br. J. sot. clin. Psychol. 11, 244247. Saklofske D. H. and Eysenck S. B. G. (1978) Cross-cultural comparison of personality: New Zealand children and En&h children. Psychol. Rep. 42, 111 l-l 116. Sen N. N. (1966) Personality Trait Inventory. New Delhi Psychological Foundation, NCERT, India.