17
This article was downloaded by: [York University Libraries] On: 10 November 2014, At: 20:30 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Multivariate Behavioral Research Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hmbr20 Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales Anne-Sofie Rosen Published online: 10 Jun 2010. To cite this article: Anne-Sofie Rosen (1976) Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales, Multivariate Behavioral Research, 11:1, 47-61, DOI: 10.1207/s15327906mbr1101_3 To link to this article: http://dx.doi.org/10.1207/s15327906mbr1101_3 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Embed Size (px)

Citation preview

Page 1: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

This article was downloaded by: [York University Libraries]On: 10 November 2014, At: 20:30Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

Multivariate BehavioralResearchPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/hmbr20

Personality Traits as Relatedto Socialization: Test of aThree-Factor Model for SomePersonality ScalesAnne-Sofie RosenPublished online: 10 Jun 2010.

To cite this article: Anne-Sofie Rosen (1976) Personality Traits as Related toSocialization: Test of a Three-Factor Model for Some Personality Scales, MultivariateBehavioral Research, 11:1, 47-61, DOI: 10.1207/s15327906mbr1101_3

To link to this article: http://dx.doi.org/10.1207/s15327906mbr1101_3

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness,or suitability for any purpose of the Content. Any opinions and viewsexpressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of theContent should not be relied upon and should be independently verified withprimary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of theContent.

Page 2: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan,sub-licensing, systematic supply, or distribution in any form to anyone isexpressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 3: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

PERSONALITY TRAITS AS RELATED TO SOCIALIZATION: TEST OF A THREE-FACTOIR

MODEL FOR SOME PERSONALITY SCALES

ANNE-SOFIE ROSEN University of Stockholm

A B S T R A C T

A role-taking factor, representing a theoretically important persc~nality variable in socialization, was assumed to load the PEN Psychoticism scale (Eysenck & Eysenck, 196833) and three scales constructed by Ros6n and Schal- ling (1974) of itelms from the CPI Socialization (So) scale (Gough, 1969). A model of a hypothesized factor structure underlying scores on these scales and some anxiety and extraversion scales was tested in a group of sftudents (N = 225). Estimates of the free parameters of the factor model were ob- tained in a restricted maximum likelihood factor analysis, but the model showed poor fit. Some restrictions imposed by the model were found less probable in an explorative analysis of reasoins for poor fit. Relaxing those restrictions led to a good-fitting three-factor model. The Role-taking factor was separate from and negatively oblique to orthogonal Anxiety and Extra- version factors.

Many attempts have been made to demonstrate a relationship of personality variables to the acquisition and maintenance of so- cialized behavior. For example, Gough (1948, 1960) assumed1 that individual differences along a hypothetical dimension sf socializa- tion would reflect differences in role-taking ability and role-taking experiences on which socialized behavior rests. The role taking concept was central in the theoretical social psychology of 1G. H. Mead (1934). According to Mead role-taking ability develops from childhood during social interactions andl is a prerequisite for the emergence of the self. A system of shared or communicable mean- ings, "the generalized other", is built up from role-taking experi- ences. Taking t'he role of the other implies that a person vic, 3 ~ s a social event from a common perspective of a society or group and that he is able to produce symbolically the meaning his own be- havior will have for others. A person controls his own bek~avior and that of others by anticipating the effects of a social act.

Gough (1948, 1960) s~rggested that low role-taking ability is the core problem in psychopathy. Several of the criteria listed by Cleckley (1964) for defining psychopathy as a syndrome charac- terizing some individuals, criminal or not, seem to f i t the role- taking hypothesis well. These criteria, conservatively applied, have been found useful for grouping criminal subjects in experimental

This study was snpported by a grant from the Swedish Council for Social Science Research. The valuable help of Dr. D. Sorbom, Department of Statis- tics, University of Uppsala, is gratefully acknowledged.

JANUARY, 1976 47

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 4: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Solfie Ros4n

research on psychopathy (Hare, 1970; RosCn & Schalling, 1971; Schalling & RosBn, 1968; Trasler, 1973). According to many clini- cal writers, the concept of low role-taking ability covers the appar- ent lack of nonsuperficial affection, empathy, and guilt and the high impulsiveness and acting out tendencies descriptive of psy- chopathy (Hare, 1970). In order to assess the level of role-taking as a factor in predisposition to crime and delinquency, Gough con- structed an inventory scale (Gough & Peterson, 1952), later in- cluded as the Socialization (So) scale in the CPI (Gough, 1969). The findings of validity, empirical and predictive, for the So scale and several translations have been quite impressive. (For a review, see Megargee (1972).) Recently some studies have reported on the dimensionality of the So scale and on the validity of and correlates to subscale scores (Butt, 1973; RosCn & Schalling, 1974; Stein, Gough & Sarbin, 1966).

Eysenck (1970a), theorizing from a biological view of per- sonality as compared with the essentially sociological one of Mead and Gough, has advanced a set of hypotheses to show how indi- vidual differences along three structural dimensions of personality will contribute in socialization and be important for, among other things, the aetiology of crime. These dimensions, labeled neuroti- cism, extraversion-introversion, and psychoticism, have been found to separate various clinical subgroups from normals in experimen- tal and clinical research (Eysenck, 1970b). The same dimensions are implied by factor analysis of personality inventory items, and sets of scales have been developed to be used in research (e.g., Eysenck & Eysenck, 1968a, b; 1970). A critical review of the theory of socialization proposed by Eysenck and of the evidence for i t when applied to crime and delinquency has recently been made by Passingham (1972).

The Eysenckian theory of personality hypothetically relates personality differences to innate functional differences in the ner- vous system. Neuroticism, involving habitual mood states and feel- ings of a mainly negative quality such as apprehension and ten- sions has its psychological substrate in the level of activation, or the autonomic arousal level controlled by the limbic structures (Eysenck, 1970a, b). This dimension may correspond to what ath- ers identify as an essential component of the anxiety concept (e.g., Cattell, 1972; Spielberger, 1972). Extraversion is assumed to relate to differences in cortical arousal, due to different arousal thresh- olds in the reticular system. Such differences may influence be-

48 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 5: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

havior in various ways, e.g., determining the rate of conditioning of avoidance behavior, assumed to underlie the impulsiveness com- ponent of extraversion (Eysenck & Levey, 1972), or the threshold for sensory intake underlying sensation seeking as another com- ponent (Schalling, 1970).

The physiological basis of the third dimension of psycholticism is incompletely outlined at present. As a descriptive dimension (Eysenck & Eysenck, 1970) it refers to being insensitive or hostile to others, not fitting in, liking oddities, and being cruel or. fool- hardy, characteristics which appear with a raised incidence in

El lolls. psychotics and criminals according to some clinicaI observ t' The psychoticism factor, interpreted from the high-loading items in the third-order solution by Eysenck and Eysenck (1!368a), seems to have been extracted from a domain with a considerable overlap in content to that of the Socialization scale of the CPI. Many items from the two scales refer to qualities of interpersonal responsivity, consideration for the feelings of others and positive experiences from interactions with people, and showing self-dis- cipline and adherence to accepted societal standards for evaluating own and others7 behavior, or to opposites to these. The item con- tent suggests that the same bipolar dimension may underlie them. A factor representing this latent dimension may be similar to the factor called superego by Royce (1973) and placed by him together with the anxiety and extraversion factors in the "strong :invari- ance" category for factors implied by many years of taxonomic factor analytic research on personality scale scores.

Not many factor studies have reported on the relationship of the So wale scores to measures of other personality variables out- side the CPI battery of scales. From reported validity of the So scale and of the Psychoticism scale from the PEN (Eysenck & Eysenck, 1968b; 1971) and from the judged similarity of content it may be assumed that a common factor contributes to the scores on the Psychoticism scale and three of the So subscales, construct- ed by Ros6n and Schalling (1974). Tliese subscales, denoted Posi- tive Interpersonal Experiences, Superego Strength, and Conform- ity, correlated .882, .798, and .570, respectively, with C;PI So scale scores, in the group of criminal and noncriminal young males of that study. Scores on these scales discriminated significantly between groups of subjects with high vs. low probabilities for socialization responses to So items. Items of the subscales as well

JANUARY, 1976 49

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 6: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

as details about the construction and validation procedure are given in Ros6n and Schalling (1974).

The present study was made to test if a common factor, tenta- tively identified as role-taking, will be found to load psychoticism and So subscale scores, and if this factor will be separable from factors representing the anxiety and extraversion constructs.

Subjects

The subjects were 240 first-year students in psychology, par- ticipating in partial fulfillment of course requirements. They were between 19 and 34 years old, with a mean age of 25.76. Nearly two-thirds were females. The scores from 225 students who had responded to all items of the included scales were used for the analysis.

The testing session lasted about 1.5 hours and consisted of groups of 20-25 students. The personality scales were mixed with other tasks, not involving self reports, and the protocols returned anonymously.

Procedure

The following scales were included to let an anxiety factor appear: the Neuroticism scale from the PEN inventory (Eysenck & Eysenck, 1968b), the items of which reflect emotional instabil- ity (e.g., moodiness, depressiveness, and worrying). Other aspects of anxiety are tapped by the items of the Psychic Anxiety scale and the Somatic Anxiety scale from the Multi-Component Anxiety inventory (Schalling, Cronholm & Asberg, 1975). The Psychic An- xiety scale includes items relating to muscular tensions, ruminations, and anticipatory worrying. The Somatic Anxiety scale includes items for autonomic disturbances, somatic complaints, and vague feelings of uneasiness and distress. This division of anxiety compo- nents has been used in psychiatric ratings by Buss and other re- searchers (Schalling, Cronholm, Asberg & Espmark, 1973). The three scales have no identical items, but there is some overlap in content between a couple of items from the Neuroticism scale and the two other scales.

The extraversion factor was represented by three scales: the Extraversion scale from the PEN inventory (Eysenck & Eysenck,

50 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 7: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Rosen

1968b), consisting mainly of sociability items, referring to ]liking man. people and social activities and to being a lively and active p t

The Impubiv i ty scale (Schalling, Note 1) contains items relating to quick actions and decisions and to being a not too meticulous person. The Monotony Avoidance scale (Schalling, Note 1) covers aspects as thrill-seeking, tiring easily and wanting change. 'There is no overlap in items between these scales.

The following four scales were assumed to be loaded by a bi- polar factor of role-taking. A negative loading was expected for the Psychoticism scale from the PEN inventory (Eysenck tSz Ey- senck, 1968b; 1970). The items refer to a lack of feeling for and being hostile toward other people, liking oddities and not fitting in. Three So subscales from the study by B o s h and Sck~alling (1974) were assumed to load the factor positively. Included here were the scales Positive Interpersonal Ezperiences (main items revealing harmonious relations with family members and other people), Superego S trength (following internal standards of be- havior, resisting temptations), and Conformity (observing con- ventions). Two items from the Psychoticism scale are similar in content but not identical to two items from the first So subscale.

The means, SDs, and correlations for the ten scales obtained in the group of students are given in Table 1, together with the number of items in each scale.

The analyses started with an unrestricted maximum likelihood factor analysis to check that a hypothesis of three common fac- tors was statistically reasonable. An introduction to the factor analytical method is given by Harman (1967). The program UFABY3 by Joreskog and van Thillo (1971) was used. Hypotheses of one to five factors were tested successively. Prom the outcomes of the tests of fit only solutions for three or four factors could be considered.

The three factor hypothesis was rejected by the likelihood ratio test statistic (Joreskog, 1967). The obtained x2 wars 36.83 d f = 18, which is significant a t a probability level of .006, Tuck- er's reliability coefficient (Tucker & Lewis, 1973) was modlerately high, or .942. (This coefficient indicates how well a factor model with the hypothesized number of factors represents the covari-

JANUARY, 1976 51

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 8: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Table 1 Means, SDs, and Correlations of the Ten Personality Scales (N = 225)

N PA S A E I M P PIE C SS Neuroticism (N) .700b .617b -.181a -.012 .I98 .223b -.362b -.456b -.375b Psychic Anxiety (PA) .666b -.432b -.218b -.019 .152a -.249b -.332b -.233b Somatic Anxiety (SA) -.I498 -.079 .016 -245" -.341b -.387b -.335b Extraversion (E) .558b .405b .077 -.011 -.I27 -.I28 Impulsivity (I) .388b .017 -.I37 -.209b -.251b Monotony Avoidance (M) .051 -.246b -,315b -.306b Psychoticism (P) -.321b -.173a -.287b Positive Interpersonal

Experiences (PIE) Conformity (C)

.516b .583b

Superego Strength (SS) .520b

Mean 8.34 7.58 7.01 11.20 5.16 5.45 SD

3.25 8.39 8.00 7.74 4.53 4.65 3.91 4.09 2.42 2.07 2.09 1.64

n items 2.02

20 20 1.53

20 20 10 10 20 11 12 9 <.05

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 9: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Roskn

ances among the traits in the population from which the ok~serva- tions were sampled.)

The four factor solution could not be rejected on statistical grounds as the obtained x2 was 12.62, df = 11, p = .319. Tucker's reliability coefficient was .992. However, the statistical fit, could be considered too close. The solution also contained a Hebywood case (Harman, 1967), evidenced by the unique variance of the Somatic Anxiety scale of .005. In a varimax rotation only this scale had a high loading on the fourth factor. A rotation of the three factor solution, however, was easily interpretable in terms of the hypothesis for the present study, as the highest (lowest) loadings were obtained in expected positions. Three factors thus seemed to be a good approximation.

A specified hypothesis concerrling the factor structure under- lying the ten scales was then tested, assuming that each scale was only loaded bjr one of the three factors of anxiety, extraversion and role-taking. This results in the hypothetical factor-loading matrix in Table 2, where loadings have either been fixed <at zero or left to be estimated. No assumption was made about the corre- lations among the factors. The unique factor dispersion matrix was assumed to be a diagonal matrix. A solution for the hypothe- sized factor structure will identify a unique decomposition,, which is directly interpretable. Rules for obtaining unique solutions are discussed by Joreskog (1969). Because of the number of fixed parameters in the model (Table 2) restrictions are imposed on the common fa,ctor space.

Table 2 The Hypothesized Matrix of Free and Restricted Factor Loadings.

All Restricted Parameters Were Set Equal to Zero.

Factor Anxiety Extraversion Role-Taking

Neuroticism 0 0 Psychic Anxiety 0 0 Somatic Anxiety 0 0 Extraversion 0 0 Impulsivity 0 0 Monotony Avoidance 0 0 Psychoticism 0 0 Positive Interpersonal Experiences 0 0 Conformity 0 0 Superego Strength 0 0

Maximum likelihood estimates for the free parameters in the

JANUARY, 1976 53

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 10: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Ros6n

model were obtained in a run of the program FASCOE (Sorbom, Note 2) together with a test of the fit. The X2 statistic was 137.1, df = 32, p < .0001, and thus the fit was poor.

In order to find a better fitting model some of the parameters restricted to be zero may be set free to vary. If the restriction for a zero element in the factor loading matrix is relaxed, this implies that the corresponding scale is no longer assumed to be a pure trait measure. If one lets the unique dispersion matrix be a full matrix, an off-diagonal element may be left free to be estimated. This means that all unique parts of the measures are no longer assumed to be uncorrelated. The problem, however, is to find that restriction in the rejected model which is least probable, in order to test exploratively for a better fitting model. In maximum-like- lihood factor analysis, estimates of the free parameters are ob- tained by minimizing a function, equivalent to the negative log of the likelihood function, with respect to the free parameters and with the others fixed a t hypothesized values. The likelihood ratio test statistic is used to test for fit, when the minimum for the function has been found (Joreskog, 1989). Sorbom (1974) suggests that fixed element be set free for estimation which has the greatest absolute value among the derivatives of the function a t its minimum in either the common factor loading or dispersion matrices, or the unique dispersion matrix. By this procedure the largest decrease in the metric for the goodness-of-fit test will be obtained. Test of fit of a model with one more parameter free for estimation may indicate that a further release of restrictions is appropriate.

By this procedure five successive models were tested, using the FASCOE program (Sorbom, Note 2). Results from the good- ness-of-fit tests are given in Table 3. The model fitting was stop-

Table 3 Results from the Goodness-of-Fit Tests for Successively Tested Models. Model I is Given in Table 2. Fit of Models I1 to VI after Releasing the Parameter x~~ in the Factor Loading Matrix and pij in the Unique Dispersion Matrix

Model Release of x2 df P

I 137.1 32 .OOOO

I1 A22 92.5 31 .OOOO

I11 A22 + A41 69.1 30 .0001

Iv A22 + A41 + *63 52.3 29 .0051

V A22 +- A41 + *63 + A61 43.8 28 .0293 VI A22 + A41 + p 6 3 + A61 + *84 36.8 27 .lo21

54 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 11: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Rosen

ped when it could be concluded that none of the fixed eleiments in the factor loading or unique dispersion matrices differed sig- nificantly from zero.

The maximum likelihood estimates of the parameters in the final Model VI fitting the observations are given in Table 4.

Table 4 Maximum Likelihood Estimates of the Free Parameters of Model VI in the Factor Loading Matrix, the Factor Variance-Covariance Matrix and the Unique Factor Variance-Covariance Matrix. Standard Error of the Estimates

Are Given within Parentheses.

Factor Loadings lij Unique Anxiety Extraversion Role-Taking Variances qii

Neuroticism .413 .OOO .OOO .330(.041) Psychic Anxiety .434 (.030) -.249 (-038) .OOO .I62 (.033) Somatic Anxiety .395 (.030) .OOO ,000 .388 (.045) Extraversion -.I21 (.036) .548 ("058) .OOO .362 (.057) Impulsivity .OOO .501 .OOO .508 (.061) Monotony Avoidance .I04 (.036) .423 ("053) .000 .609 (.070) Psychoticism .OOO .OOO -.320 367 (.085) Positive Intemersonal

~ x ~ e r i e n c e s .DO0 .OOO .643 (.156) .464 (.061) Conformity .OOO .OOO .632 (.155) .482 (.063) Superego Strength .000 .OOO .650 (.172) .437 (.069)

Factor Variance-Covariance Matrix

I I1 111 Anxiety (I) 3.928 (549) Extraversion (11) .I12 (.298) 1.958 (.357) Role Taking (111) -1.397 (.389) ,671 (.229) 1.297 (.621)

Nonzero Unique Covariances : qeB: Monotony Avoidance : Somatic Anxiety -.I75 (.041) @irs,: Positive Interpersonal Experiences : Extraversion .I03 (.041)

DIscussIori It was the purpose of the present study to test the hypothesis

that a common bipolar factor of role-taking loads scores on the Psychoticism scale from the PEN and three subscales constructed of items from the CPI So scale. Scales for measuring aspects of the anxiety and extraversion constructs were included in oirder to obtain more information about the hypothesized role-taking factor and to separate it from the factors of anxiety and extraversion. These factors may be regarded as representing the best estab- lished dimensions of personality and important in current theoriz- ing and experimental and correlation research on socialjzation,

JANUARY, 1976 55

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 12: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofia Rosen

deviance, and psychopathy (e.g., Eysenck, 1970; Hare, 1970; Schal- ling, 1970; Trasler, 1973; Yates, 1970).

The results of the unrestricted factor analysis indicated that the assumption that three factors are sufficient to explain the covariation was statistically reasonable and that the fit of a three- factor model was good. In this preliminary factor analysis, the factors were considered to be orthogonal. A varimax rotation re- sulted in a solution with highest and lowest loadings in expected places in the factor loading matrix. The theoretical interpretation of such a solution may however be regarded as somewhat arbi- trary as orthogonality has been postulated for the frame of ref- erence and rotation to maximum simplicity is based on a mathe- matical criterion (cf. Harman, 1967). Instead, a hypothesis was set up, based on personality theory and research on socialization, to test the dependence of scores on the ten scales on the three factors, assumed to be Anxiety, Extraversion and Role-taking. Under the hypothesis tested (Table 2) a sufficient number of parameters were specified to obtain a unique solution, interpret- able without rotations, and the factors were allowed to correlate. The fit of the hypothesized three-factor model proved to be poor (Table 3). There are, of course, many reasons, psychological and statistical, for poor fit of a model concerning the relationship among factors representing the assumed relationship of person- ality traits in a population, when observations are made on a sample. Two reasons for poor fit were regarded as potentially im- portant for the present study.

When the classical factor analytical model is regarded as a good approximation to the latent structure of observations and its assumptions of linearity and normality are met, other assumptions of the basic model may have been violated; e.g., the factor model assumes that after partialling out the influence of the common factors the residuals or unique factors contain nothing but spe- cifics and random errors and are independent among themselves, making the covariances of the residuals equal zero, It has been argued (Pawlik, 1973) that when personality scale scores are factor analyzed, this assumption often does not hold true, and that correlated errors, contaminating the interpretation of the factors, are present due to similarities in scale construction, to memory effects on responses, and so forth.

Personality scales constructed of dichotomously scored items may not possess the desired property of being unitractic, or meas-

56 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 13: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Rosen

ures of a single trait. It is well known that such items are rarely, if ever, in practical use, loaded by a single factor. It is still pos- sible to construct unitractic scales, the items of which al'l have their highest loading on the same common factor, uniquely rotated to a position fitting a theoretical concept (Cattell, 1973). A subset of items in such a scale may load on another common factor, but in an opposite direction. The summed scores on a well constructed personality scale with the unitrait property is explained by the influence of one common factor and the contribution of other, un- wanted factors is suppressed. This goal for personality measure- ment is not easily obtained, requiring both. substantive clalrity in defining the theoretical concept involved and methodological sophistication in construction. A reason for poor fit of a factor structure hypothesis, assuming unitrait scales, may be that such a property of balance by suppression has not been acquired for some scales. The hypothesis tested will not then reflect the basic structure of the measures.

The method developed by Sorbom (1974) made it possible Do explore reasons for poor fit to the factor structure hypothesis, if correlated errors or lack of a unitrait property were likely. In the model fitting the observations (Nodel VI, Tables 3 :and 4) two unique covariances differed significantly from zero, those of the Extraversion and Positive Interpersonal Experience scalles and of the Monotony Avoidance and Somatic Anxiety scales. Accord- ing to Pawlik (1973) this would indica'te that the error c:ompo- nents are correlated and the assumptions of the factor model vio- lated. A more likely interpretation, in accord with the factor model, is that the estimated unique factors are not "genuinely unique" (Rozeboom, 1973), but that common sources rema~in un- extracted. The contribution of such sources is small enough to re- main indistinguishable from random error. This interpretaction is also plausible ln light of the outcomes of the tests for the number of common factors that preceded the explicit model testing.

Most of the scales of the present study seemed to have a suf- ficient degree of the desired property of being unitractie when analyzed in the present context (Table 4). W e n evaluating the sizes of the factor loadings it should be kept in mind thzut they were estimated under a hypothesis that restricted the common factor space. This will affect both eommunality and uniqjueness estimates, For two scales of Extraversion, the Monotony Avoid- ance and the Extraversion scales, minor, but significant lc,adings

JANUARY, 1976 57

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 14: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Ros6n

were estimated to come from the Anxiety Factor. The Psychic Anxiety scale was estimated to have a considerable negative load- ing from the Extraversion factor. Removing the restriction of a zero loading for this scale in the hypothetical factor loading mat- rix of Table 2 produced the largest decrease in the X2 value. It sug- gests that some changes, e.g., rephrasing some of the items of this scale, may improve i t psychometrically.

Eysenck, basing his arguments on physiological theorizing (e.g., Eysenck, 1970a, b), has always contended that the person- ality dimensions of neuroticism and extraversion in his proposed theory are orthogonal. It seems that he has recently come to ac- cept that oblique factors, appearing in analyses of personality scales or items, may represent theoretically orthogonal dimen- sions, as indicated by his preference for the Promax method of rotation (Eysenck & Eysenck, 1968a, b; 1970). Orthogonality of anxiety and extraversion has also been proposed by several other researchers-often with somewhat differing definitions and de- notations of the constructs-and is also suggested by a great num- ber of factor analytical findings (e.g., Barratt, 1972; Cattell, 1973; Royce, 1973). In the present study no hypothesis was tested con- cerning the correlations among the three factors. It is interesting, in view of earlier findings and recent theorizing about the physio- logical bases of personality (e.g., Barratt, 1972; Gray, 1973), that the estimated covariation of the Anxiety and Extraversion factors was zero (Table 4). This is not an expected finding for common factors of personality and suggests that in future testing of hypo- theses concerning factor structure, when the anxiety and extra- version dimensions are involved the correlation of these two fac- tors may be assumed to be zero.

There was a substantial correlation between Role-taking and the other two factors. The correlation was negative, or -.610 be- tween Role-taking and Anxiety, and -.421 with Extraversion, as estimated from the factor dispersion matrix (Table 4). If the role- taking concept stands for qualities of behavior such as social re- sponsivity and harmonious relations, ability to anticipate one's own and others' reactions and common standards for judging be- havior, i t is to be expected that a high level of role-taking will generally tend not to go with either a high level of anxiety or extraversion, as commonly defined. Data analyzed in the present study came from a group of students and it is not known when other groups are studied, if the organization of personality fac-

58 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 15: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Ros6n

tors, or the way in which they are measured by the ten scales, will be similar. The factor space may be different even if tlie fac- tor loadings are invariant. The obliqueness of the Role-taking fac- tor, loading the Psychoticism scale from the PEN, was not, unex- pected. Eysenck and Eysenck (196810) have reported correlations from .30 to .45 between the neuroticism and psychoticism factors, as measured by the PEN scales in data from several groups of subjects, analyzing sexes separately. The correlation with extra- version, however, was low.

Disappointment and pessimism concerning the contribution of factor analysis to personality theory development has frequently been vented and common applications for the purpose of tax:onomy scathingly criticized (ag., Lykken, 1971). Recently, however, i t has become possible to combine the c~ommon factor model of clas- sical factor analysis with statistical inferential procedures, mak- ing possible more adequate testing of hypotheses derived from psychological theorizing in personality. Using maximum likelihood factor analysis methods, efficient eomputational methods have been developed mainly in the work of Joreskog. These nlethods for estimation are more adequate for an explanatory view of fac- tors as determiners of covariation st the phenotypic level (cf. Hakstian & Muller, 1973) than the methods developed for the component model, frequently used in personality research today.

1. Schalling, D. Components of extraversion, Manuscript, 1975. 2. Siirbom, D. FASCOE-A computer program for factor analysis in

several populations with structured means and correlated errors. Manuscript, 1974.

Barratt, E. S. Anxiety and impulsiveness. In C. D. Spielberger, (Eld.). An- xiety: Current trends in theory and fiesearch, Vol. 1. New Y o ~ k : Aca- demic Press, 1972.

Butt, D. S. A factorial facet analysis of Gough's Socialization scale. Social Behavior and Personality, 1973,1, 50-57.

Cattell, R. B. The nature and genesis of mood states. In C. D. Spielberger, (Ed.). Anxiety: Current trends in theory and research, Vol. 1. New Pork: Academic Press, 1972.

Cattell, R. B. Personality and mood by questionnaire. San Francisco: Jossey- Bass. 1973.

Cleckley, H. The mask of sanity. (4th ed.). St. Louis: Mosby, 1964. Eysenck, H. J. Crime and Personality (2nd ed.) London: Paladin, 1970. (a)

JANUARY, 1976 59

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 16: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Rosdn

Eysenck, H. J. The stmcture of human personality (3rd ed.) London: Meth- uen, 1970. (b)

Eysenck, H. J. & Eysenck, S. B. G. A factorial study of psychoticism as a dimension of personality. Multivariate Behavioral Research. Special Issue, 1968,3, 15-31, (a).

Eysenck S. B. G. & Eysenck, H. J. The measurement of psychoticism: A study of factor stability and reliability. British Journal of Social and Clinical Psychology, 1968,7,286-294. (b)

Eysenck, S. B. G. & Eysenck, H. J. Crime and personality: An empirical study of the three factor theory. British Journal of Criminology, 1970, 10, 225-239.

Eysenck, S. B. G. & Eysenck, H. J. A comparative study of criminals and matched controls on three dimensions of personality. British Journal of Social and Clinical Psychology, 1971,10, 362-366.

Eysenck, H. J. & Levey, A. Conditioning, introversion-extraversion and the strength of the nervous system. In V. D. Nebylitsyn & J. A. Gray, (Eds.). Biological bases of individual behavior. New York: Academic Press, 1972.

Gough, H. G. A sociological theory of psychopathy. American Journal of Sociology, 1948, 53, 359-366.

Gough, H. G. Theory and measurement of socialization. Journal of Consulting P~ychology, 1960,24, 23-30.

Gough, H. G. Manual for the California Psychological Inventory (rev. ed.), Palo Alto : Consulting Psychologists Press, 1969.

Gough, H. G. & Peterson, D. R. The identification and measurement of pre- dispositional factors in crime and delinquency. Journal of Consulting Psychology, 1952,16, 207-212.

Gray, J. A. Causal theories of personality and how to test them. In J. R. Royce, (Ed.), Multivariate analysis and psychological theory. New York: Academic Press, 1973.

Hakstian, A. R. & Muller, V. J. Some notes on the number of factor problem. Multivariccte Bqhavioral Research, 1973,8, 461-475.

Hare, R. D. Psychopathy: Theory and research. New York: Wiley, 1970. Harman, H. H. Modern factor analysis (2nd ed.). Chicago: University of

Chicago Press, 1967. Joreskog, K. G. Some contributions to maximum likelihood factor analysis.

Psychometrika, 1967,32,443-482. Joreskog, K. G. A general approach to confirmatory maximum likelihood fac-

tor analysis. Psychometrika, 1969,34, 183-202. JBreskog, K. G. & Lawley, D. N. New methods in maximum likelihood factor

analysis. Bdtish Journal of Mathematical and Statistical Psychology, 1968,21, 85-96.

JBreskog, K. G. & van Thillo, M. New rapid algorithms for factor analysis by unweighted least squares, generalized least squares and maximum likelihood. Research Memorandum, Educational Testing Service, Prince- ton, New Jersey, 1971.

Lykken, D. T. Multiple factor analysis and personality research. Journal of Ezperintental Research in Personality, 1971,5, 161-170.

Mead, G. H. Mind, self, and society. Chicago: University of Chicago Press, 1934.

Megargee, E. I. The California Psychological Inventory Handbook. San Fran- cisco: Jossey-Bass, 1972.

Passingham, R. E. Crime and personality: A review of Eysenck's theory. In V. D. Nebylitsyn and J. A. Gray, (Eds.). Biological bases of individual behavior. New York: Academic Press, 1972.

Pawlik, K. Right answers to the wrong questions? A re-examination of factor analytic personality research and its contribution to personality theory. In J. R. Royce, (Ed.). Multivariate analysis and psychological theory. New York: Academic Press, 1973.

60 MULTIVARIATE BEHAVIORAL RESEARCH

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14

Page 17: Personality Traits as Related to Socialization: Test of a Three-Factor Model for Some Personality Scales

Anne-Sofie Rosen

Rodn, A.-S. & Schalling, D. Probability learning in psychopathic and non- psychopathic criminals. Journal of Experimental Research i n Personality, isii, 5,191-198.

Ros6n, A.-S. & Schalling, D. On the validity of the CPI Socialization scale: A multivariate approach. Journal of Consulting and Clinical Psychology, 1974,.42,757-765.

Royce, J. R. A conceptual framework for a multi-factor theory of individual- ity. In J. R. Royce, (Ed.). Multivariate analysis and psychological theom. ~ k w York: ~cademic Press, 1973.

Rozeboom, Wm. W. Comment on Pawlik's paper. In J. R. Royce, (Ed.). Mul- tivariate analysis and psychological theogy. New York: Academic Press, 1973.

Schalling, D. Contributions to the validation of some personality co~xcepts. Reports from the Psychological Laboratories, University of Stockholm. Supplement Series Nr. 1, 1970.

Schalling, D. & R I D S ~ ~ , A.-S. Porteus Maze differences between psychc~pathic and non-psychopathic criminals. British Journal of Social and C:linical Psychology, 1968,7,224-228.

Schalling, D., Cronholm, B. & Asberg, M. Components of state ancl trait anxiety as related to personality and arousal. In L. Levi (Ed.), Emo- tions-Their parameters and measurement. New York: Raven Press, 1975.

Schalling, D., Cronholm, B., Asberg, M. & Espmark, S. Ratings of psychic and somatic anxiety indicants. Acta Pqchiatrica Scandinavica, 1973, 49, 353-368.

Spielberger, C. D. Anxiety as an emotional state. In C. D. Spielberger, (Ed.). Anxiety: Current trends in theow and research. Vol. 1. New York: Aca- demic Press, 1972.

Stein, K. B., Gough, H. G., & Sarbin, T. R. The dimensionality of the CPI Socialization scale and an empirically derived typology among delinquent and nondelinquent boys. Multivariate Behavioral Resea~ch, 1066. 1. . . 198-208.

SGrbom, D. Detection of correlated errors in longitudinal data. Research Re- port 74-8, Department of Statistics, University of Uppsala, 1974.

Trasler, G. Criminal behavior (2nd ed.). In Eysenck, H. J. (Ed.). nandbook of abnormal psychology. London: Pitman, 1973.

Tucker, L. R. & Lewis, C. A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 1973,38, 1-10.

Yates, A. J. Behavior therapy (Chap. 11). New York: Wiley, 1970.

JANUARY, 1976 61

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 2

0:30

10

Nov

embe

r 20

14