Вы находитесь на странице: 1из 27

Sex Roles, VoL 36, Nos.

1/2, 1997
Locus of Control and Affectivity by Gender
and Occupational Status: A 14 Nation Study 1
Peter B. Smi th 2 and Shaun Dugan
University of Sussex
Fons Trompenaars
Centre for International Business Studies
This study investigated the extent to which reported gender differences in values
are attributable to differences in national culture and organizational seniority.
Locus of control and affectivity scores were obtained f rom 4599 managers and
employees in business organizations located in 14 countries. Across nations,
men and those in senior positions were more internal and less affective. Gender
effects were additional to those attributable to status. Respondents in nations
scoring higher on an index of modernity were f ound to be less intemaL ,4
pan-cultural factor analysis of locus of control items yielded f our subscales.
Men and those in senior positions scored higher on Socio-Political Control
and Effort, but no differences were f ound on the Luck and Active Friendship
subscales. The results are contrasted with those obtained from single-nation
samples using student subjects.
The nat ur e and det er mi nant s of gender di fferences in social behavi or s have
sust ai ned t he at t ent i on of r esear cher s f or many years. However , t he con-
cl usi ons whi ch emer ge f r om such st udi es are necessari l y bounde d by t he
cont ext wi t hi n whi ch dat a have been col l ect ed. The pr es ent st udy seeks t o
br oa de n t he r ange of count r i es f r om whi ch dat a ar e dr awn and t o at t end
t o var i abl es of t en conf ounded wi t h gender t hat are t hought likely t o affect
t he gender - cor r el at es t hat ar e f ound.
1We acknowledge receipt of ESRC Grant R 000 22 1552, which aided the completion of this
study, and the helpful comments of Michael Bond and John E. Williams upon an earlier
version.
2"1"o whom correspondence should be addressed at School of Socia/ Sciences, University of
Sussex, Falmer, Brighton BN1 9QN, UK.
51
0360-13025/97/0100--0051512.50/0 1997 Plenum Publishing Corporalion
52 S mi t h et al .
Eagly & Wood (1991) review met a-anal yses of t he largely Nort h
American literature on gender differences and document a number of re-
liable sex differences in a wide range of social behaviors. For example,
women tend to agree mor e with ot her people than do men (Eagly & Carli,
1981), are better at encoding and decoding non-verbal information (Eagly,
1987; Hall, 1984; Hall & Halberstadt, 1986), and are less aggressive, par-
ticularly in terms of physical aggression (Eagly, 1987; Eagly & Steffen,
1986).
These results challenge the widespread view that there are few gender
differences in social behavior, a view which Eagly (1987) attributes to Mac-
coby & Jacklin's (1974) highly influential review of developmental gender
differences. She suggests that t he finding by Maccoby & Jacklin of no clear
evidence of sex differences in nurturance or ot her areas of social behavior
is a function of the experimental research methods employed in the studies
they reviewed. Of particular importance to Eagly's argument is t he idea
that male and female subjects share the same role in experimental studies,
that of 'subject,' whereas in naturalistic settings, men and women are dif-
ferentially distributed into social roles. From this perspective, experimental
research when compared with naturalistic observation has tended to reduce
the impact of gender differences because the experimental situation will
often inhibit the expression of differences which are evident in more natural
settings.
Eagly's (1987) review of the literature on gender roles and stereotypes
(e.g., Bern, 1974; Spence & Helmreich, 1978) concludes that beliefs about
the attributes of men and women can be summarized in terms of two di-
mensions, which she prefers to label communal and agentic after Bakan
(1966). The communal dimension can be described as a concern for the
welfare of others, and includes caring and nurturant qualities, as well as
interpersonal sensitivity and emotional expressiveness. Ot her authors have
labelled this dimension as expressiveness, social orientation and femininity.
The agentic dimension is defined by attributes such as serf-assertion and
independence from others. Women are generally believed to be mor e com-
munal and less agentic than men. This distinction is similar to that made
by earlier writers, including Parsons & Bales (1955) who distinguished mas-
culinity, defined in terms of instrumentality and goal directedness, from
femininity, defined by expressiveness, supportiveness and affectivity.
The five factor model of personality (Digman, 1990), which has within
t he last decade become the dominant paradigm in personality research,
offers a further useful way of conceptualizing the differences outlined by
Eagly and her colleagues. In particular, ' communal' traits seem congruent
with the ' agreeableness' domain of t he five factor model, while ' agentic'
traits such as assertion and dominance are adequately subsumed by the
Loc us o f Cont rol and Affecti vi ty 53
' extraversion' /' surgency' factor. Trapnell & Wiggins (1990) note that gender
differences of around >_ .5 standard deviations have often been found using
measures of agreeableness. Moreover these differences have, in some cases,
been detected with spouse ratings as weI1 as self-reports. However, gender
differences on Big Five measures of extraversion/surgency are not generally
large or reliable. Instead, differences of > .4 standard deviations in ' neuroti-
cism' /' negative affectivity' and depression are regularly found on various
measures of these traits (Jorm, 1987; Trapnell & Wiggins, 1990).
CROSS-CULTURAL WORK ON GENDER DIFFERENCES
The great majority of the studies discussed above relied solely upon
dat a collected in North America. Cross-cultural studies hold great potential
for clarifying the generality of the effects noted. In reviewing the relevant
literature it is crucial to note that some studies follow the same type of
design as that employed in the North American literature, namely testing
for individual-level correlates of gender within a single national culture.
Ot her studies treat the national culture as the unit of analysis and examine
culture-level correlates of gender relations. Each approach has strengths
and weaknesses. Single-country individual-level studies can potentially tell
us whet her the differences obtained in the USA are widely replicable, but
only where there is convincing evidence for the comparability of measures
used in different cultural samples. Multiple-country individual-level studies
based upon rating scale data can yield spurious results unless great care is
taken to control for culture-level differences in response bias (Leung &
Bond, 1989). Culture-level studies can provide a basis for hypotheses as to
why results obtained at different locations may diverge.
Individual-Level Studies
Cross-cultural researchers have been much influenced by the proposi-
tion that a cultural dimension defined as individualism-collectivism can pre-
dict differences in a wide range of social behaviors across cultures (Bond
& Smith, 1996). Members of individualistic cultures typically define their
identity in terms of personal choice and autonomy, while members of col-
lectivist cultures define themselves more in terms of enduring commitments
to the groups in which they are members. Triandis, Leung, Villareal, &
Clack (1985) have proposed that to reduce confusion bet ween individual-
level and culture-level analyses, we should refer to the values of those en-
54 Smi t h et ai.
dorsing cultural individualism as idiocentric and the values of those en-
dorsing cultural collectivism as aUocentric.
There is a possible overlap between t he concepts of agency and idio-
eentrism and between communalism and allocentrism. However no gender
differences in idiocentrism-allocentrism were found among Korean adults
(Cha, 1994), Japanese students (Yamaguchi, 1994), or Polish adults (Reyk-
owski, 1994). Kashima, Yamaguchi, Kim, Choi, Gelfand et al. (1995) suggest
that this is because gender differences are not a function of idiocentrism-
allocentrism but of relative endorsement of the values of relatedness (Gil-
ligan, 1982). Af t e r maki ng dat a st andar di zat i ons appr opr i at e t o a
multiple-country study, they found gender differences on a relatedness
measure but not on measures of idiocentrism-allocentrism among students
in Korea, Japan, Australia, Hawaii and mainland USA. This study suggests
that individual-level gender differences may indeed be replicable across cul-
tures where appropriate measures and data-analysis procedures are used,
but existing studies have used insufficiently broad samples to det ermi ne
whet her this is so.
Culture-Level Studies
Williams & Best (1982, 1990) have provided convincing evidence for
the pan-cultural generality of gender stereotypes and differences in the self-
ascribed identities of men and women. Their 25 nation gender stereotypes
study (Williams & Best, 1982) involved students rating each of the 300
items comprising the Adjective Check List (Gough & Heilbrun, 1980) in
terms of their applicability as descriptors of men and women within their
respective cultures. Analyses revealed that in each culture, adjectives asso-
ciated with males were stronger and more active, (but not more favorable)
than adjectives associated with females, when scored in terms of Osgood' s
factors of affective meaning. When scored to reflect Gough & Heilbrun' s
15 psychological needs, pan-cultural generality was again demonstrated,
with items rat ed as characteristic of males scoring higher on dominance,
autonomy, aggression, exhibition, achievement and endurance, while items
associated with women scored higher on nurturance, affiliation, heterosexu-
ality, succorance, deference and abasement.
Interestingly, further analyses showed that socioeconomic development
moderat ed the stereotypical strength and activity differences, which were
found to be greater in countries with lower levels of socioeconomic devel-
opment. Moreover, in countries displaying larger differences between males
and females on the strength and activity dimensions, the male stereotype
was viewed mor e favorably.
Locus of Control and Affectivity 55
Williams & Best (1990) extended this research in a later study in which
male and female students from 14 nations described themselves and their
ideal selves on the Adjective Check List, and also completed the Kalin Sex
Role Ideology measure (Kalin & Titby, 1978). As in the previous study,
national culture moderated the gender difference findings. When scored
in terms of affective meaning, males' self and ideal self ratings tended to
be stronger and more active than those of women in most, though not all
countries. On the sex-role ideology measure cultural variables accounted
for more variance than did gender. Liberal sex-role ideology was found to
be strongest in ' modern' nations, i.e., those endorsing the values of indi-
vidualism and low power distance (Hofstede, 1980), which had greater so-
cioeconomic development and which had predominantly Christian as
opposed to Muslim populations. In most countries, however, women were
found to be less traditional than men, reinforcing previous U.S. findings
(Spence & Helmreich, 1978).
Williams & Best found that women viewed themselves more favorably
than men in economically developed nations, i.e., those scoring low on
power distance and high on individualism, where women are less confined
to domestic roles. This relationship was reversed in less economically de-
veloped nations. Furthermore the differences between men and women on
the affective meaning and strength dimensions were less pronounced in the
more developed countries, and male and female self-concepts were gener-
ally more similar.
Buss (1989) studied mate preferences across 37 nations. He found a
consistent male preference for partners who were seen as young, healthy
and beautiful, whereas women expressed a preference for partners who
were seen as ambitious, industrious and of high earning capacity. However
this consistent divergence by gender was overshadowed by much larger vari-
ations in preferences across nations (Buss & 49 co-authors, 1990).
Schwartz (1994) reports upon gender differences in values among his
samples of students and teachers from 33 nations. Men more often en-
dorsed values related to Power, while women more often endorsed values
related to Benevolence. In some nations men also endorsed Tradition and
Conformity values more than women, and Schwartz suggests that they do
this in nations where men are most strongly associated with positions of
authority.
The work of Williams & Best, Buss and Schwartz clearly documents
the existence of culture-level gender differences, and suggests that a sub-
stantial part of these differences can be summarized in terms of the com-
munal and agentic dimensions. While meta-analytic reviews of gender
differences in values and behavior have been largely concerned with West-
ern data, the cross-cultural literature indicates that economic development
56 Smi t h et ai.
and variables associated with modernization are clear correlates of the mag-
nitude of gender differences.
ECONOMIC DEVELOPMENT AND MODERNITY
We need to consider whether or not the association bet ween economic
development and magnitude of gender differences might be thought of as
historical coincidence. Are relatively equal gender relations found in those
nations that industrialized early because they did so? If so, does it follow
that as ot her nations advance economically, gender roles will also change?
It is possible that economic development might affect gender roles in a
variety of ways, such as by causing expansion of educational opportunities
and improvement of health. Both of these changes are likely to enhance
opportunities for women to move into roles previously mostly filled by men.
However, we may suppose that the most direct way in which economic
devel opment could affect gender roles is that in industrialized cultures
women and men will have greater opportunities to take on less differenti-
ated ,work roles. To the extent that social structures evolve so as to enhance
individuals' capacity to choose what roles they will enter, agentic or idio-
centric values are likely to increase among both men and women and the
gap bet ween the two should decrease. Hofst ede (1980), Bond (1988), and
Smith, Dugan, and Trompenaars (1996) all found value measures related
to individualism higher in economically advanced nations, and Hofst ede' s
data suggested increases in individualism over time. Yang (1988) notes that
many of the values and behaviors associated with cultural modernity are
expressions of enhanced individualism. For instance in modern societies
there is less fatalism and a greater sense of personal efficacy, more open-
ness to change, less integration with relatives and more egalitarian atti-
tudes. However Yang further notes that convergence of modern societies
involves only those behaviors which have specific value in adapting to the
demands of industrialized societies. Ot her aspects of cultural difference are
found t o persist. Williams and Best' s (1990) finding that gender roles were
less differentiated in nations with higher socio-economic devel opment was
based upon an index which weighted equally 15 indicators of income, edu-
cation and health. This index was also found to correlate significantly with
Hofst ede' s scores for individualism in the nations that they sampled.
In currently existing economically-advanced nations the freedom to
choose one' s career pat h and influence one' s promotion prospects are by
no means absolute, and continues to yield a pattern of role occupancy in
which men occupy many more of the senior roles than do women. None-
Locus of Control and Affectivity 57
theless we can predict that within nations having advanced economies there
will be smaller gender differences in values.
As Eagly has hypothesized, role occupancy will elicit gender differ-
ences in circumstances where role assignment is influenced by gender, as
is frequently the case in studies outside of the psychology laboratory. The
present study affords the opportunity to survey the values of men and
women occupying roles of greater or lesser organizational seniority within
each of the nations sampled. We may thus evaluate the degree to which
gender differences are attributable to the fact that men frequently occupy
more senior positions, versus the possibility that differences persist even
when seniority is equated. The study focuses upon gender differences in
locus of control and in affectivity. Both these variables can be related con-
ceptually to the agentic and communal dimensions. We consider in turn
their possible relation to status and to gender.
LOCUS OF CONTROL
The locus of control construct has generated enormous interest over
the past 30 years (Lefcourt, 1981, 1982; Phares, 1976). Locus of control,
as defined by Rot t er (1966), refers to individual differences in the extent
to which peopl e perceive events as contingent upon their own behavior or
enduring characteristics (a belief in internal control) versus the extent to
which they believe that reinforcement is contingent not upon the self, but
upon external factors such as chance, fate, or powerful others (a bel i ef in
external control). By this definition, an internally perceived locus of control
is similar to but not identical with agentic traits. Internality has to do with
control over events, whereas agency is defined in terms of the performance
of assertive or controlling behaviors (Eagly, 1987). One might thus be agen-
tic without being internal, but it is unlikely that persons who believe in
their capacity to internally control the outcome of events will not also be-
have in agentic ways. We should therefore expect internality to be higher
among males.
Locus of Contro~ Status and Achievement-Related Behavior
It has often been found that an external locus of control is associated
with diminished opportunity for social advancement, lower achievement
and lower socioeconomic status. Bar-Tal & Bar-Zohar (1977) reviewed 36
studies, of which 31 report ed a positive relationship bet ween locus of con-
trol and achievement. Dyal (1984) reviews relevant cross-cultural studies.
58 Smith et al.
However, an accurate assessment of the strength of these relationships, and
the impact of moderat or variables such as culture upon t hem will require
a meta-analysis.
There are at least two ways in which this association could be produced
(e.g., O' Bri en, 1984). Lef cour t (1982) in discussing the links bet ween
achievement-related behavior and locus of control, remarks upon the ap-
peal to common sense which suggests that,
. . prolonged achievement effort, will occur only among individuals who believe
they can, through their own efforts, accomplish desired goals. Individuals must
entertain the hope that their efforts can be effective before they can make the
sacrifices that are prerequisites for achievement. (Lefcourt, 1982, p. 81)
An internal locus of control may facilitate selection into higher status
groups and occupations by fostering an increased sense of personal efficacy,
with associated gains in planning, effort and motivation. Additionally, social
roles linked with status may influence locus of control. In particular, the
occupation of social roles associated with diminished opportunities for ad-
vancement might encourage external rather than internal beliefs. For ex-
ample, unemployed persons may become more external, since their efforts
and skills are not leading to rewards, they probably had little say in em-
ployment cessation, and the receipt of welfare benefits is not usually con-
tingent upon effort (O' Brien, 1984).
Gender Differences in Locus of Control
Women are a disadvantaged group vis-a-vis men in the occupational
domain (Kanter, 1977). It has been common in most Western democracies
for women to constitute the majority of employees in clerical and secre-
tarial grades, and a small minority within senior management grades. More-
over, this effect is still more pronounced in more traditional societies. We
might therefore expect that women would develop a more external locus
of control when working in environments that are less responsive to their
efforts at achieving higher status.
Directly relevant to this hypothesis is a longitudinal investigation in
which 13,000 US citizens compl et ed a locus of control scale t hree times
over an eight year period (Dohert y & Baldwin, 1985). Results showed that
the female group became substantially more external, while male scores as
a whole remained stable. Dohert y & Baldwin suggest that over t he course
of the study there had been a cultural shift, in a decade when gender role
issues became increasingly salient. According to this explanation, females
became increasingly aware of the constraints on their occupational and
ot her goals.
Lo c u s of Control and Af f ect i vi t y 59
Dyal (1984) reviewed the cross-cultural literature on gender differ-
ences in locus of control. Women have been found to be more external
than men in a number of studies. For instance, Parsons & Schneider (1974)
found a small overall gender effect across eight nations, as did McGinnies
et aL (1974) in five countries. Ot her studies reporting significant effects in
the same direction have been report ed from Australia (Feather, 1967); Bel-
gium (de Brabander & Boone, 1990); India (Aggarwal & Kumari, 1975;
Bhattacharya & Husain, 1985; Khanna & Khanna, 1979); Israel (Yuchtman
& Shapira, 1981); Taiwan (Lao, Chuang & Yang, 1977); United States (Do-
herty & Baldwin, 1985); and Sweden (Lee & Dengerink, 1992). Cole &
Cole (1977) found Mexican and U.S. business administration women stu-
dents more internal, but this is one of very few studies reversing the overall
pattern.
Hypotheses--Locus of Control
The previous review enables us to make some clear predictions con-
cerning relations bet ween gender, status and locus of control:
Hypothesis 1: Me n wi l l s c o r e mo r e i n t e r n a l l y t ha n wo me n ,
panculturally.
Hypothesis 2: Higher status respondents will score more internally,
panculturally.
Hypothesis 3: Gender and status differences in internality will be less
in more modem societies.
AFFECTIVITY ORIENTATION
In contrast to locus of control, the concept of affectivity, as elaborated
by Parsons & Shils (1951), has not generated an extensive research litera-
ture. The concept refers to value orientations relating to the affective con-
tent of a relationship which influences interpersonal interactions. Persons
with an affective orientation are said to be affiliative, ' natural,' and expres-
sive in their interpersonal relations, while persons with an affectively neutral
orientation tend not to be as expressive. Parsons related these orientations
to temporal differences in how peopl e achieve gratification--persons with
an affectively neutral orientation were said to be more likely to defer grati-
fication in the interests of attaining objectives. These definitions suggest
that Parsons' concepts may be related to the Big Five dimensions of ex-
traversion and agreeableness (affectivity) and conscientiousness (affective
neutrality). It was suggested earlier that agreeableness also overlaps con-
60 Smi t h e t al.
ceptually with Bakan' s dimension of communion. Barrick & Mount (1991)
report a meta-analysis showing that conscientiousness is consistently posi-
tively evaluated within organizations, and is therefore likely to lead to pro-
motion to more senior positions. This provides a basis for predicting that
affectivity will be higher among low status persons. The study by Kashima
et al. (1995) showed that women in six nations scored higher on relatedness,
which may also be equated to affectivity. However the precise degree of
interrelation between extraversion, agreeableness, allocentrism, relatedness
and communion remains a mat t er for speculation. The measures available
for the present study were based upon Parsons' concepts and t he hypothe-
ses are therefore formulated in terms of relative affectivity.
Hypothesis 4:
Hypothesis 5:
Hypothesis 6:
Affectivity Hypotheses
Wo me n wi l l be mo r e a f f e c t i v e t h a n me n ,
pan-culturaUy.
Low st at us r e s ponde nt s will be mo r e af f ect i ve,
pan-culturally.
Gender and status differences in affectivity will be less
in modem societies.
METHOD
Subjects
The locus of control and affectivity items were administered to busi-
ness-employed respondents at varying times between 1982 and 1993. The
items analyzed here were taken from a larger 79 item questionnaire (Trom-
penaars, 1985, 1993) which is principally concerned with measuring cultural
differences in work-related values. With the exception of samples from the
former communist bloc, respondents were mostly recruited from employees
attending short training courses addressing cross-cultural aspects of man-
agement. Questionnaires were completed before the commencement of
training.
This study uses t he responses of 4,599 people from 14 countries se-
lected from a larger database of employees from business organizations
located in 58 nations. A country sample was included in the study if there
were 15 or more respondents in each of the four ceils defined by a cross-
tabulation of gender and occupational status. Respondents were catego-
r i zed as hi gh st at us cat egor y i f t hey r e por t e d bei ng ma na ge r s or
Loc us o f Cont rol and Affecti vi ty
Table I. Sample Sizes for Each Country by Gender and Status
61
Occupational Status
Lower Higher
Country Language Used Female Male Female Male
Be l g i u m Fr ench/ Dut ch 57 71 23 147
Brazil Portuguese 32 26 20 40
Bulgaria Bulgarian 17 42 33 33
China Chinese 24 35 24 47
Czechoslovakia Czech 73 52 67 84
Hungary Hungarian 59 22 18 28
India English 16 18 41 207
Mexico Spanish 35 15 22 18
Netherlands Dutch 93 158 62 755
Poland Polish 43 34 32 89
Rumania Rumanian 18 30 86 61
Sweden Swedish 47 96 27 141
U.K. English 110 65 264 604
Ex-U.S.S.R. Russian 5.._.55 2_..6_6 11._..55 141
679 690 834 2395
professi onal s. The r emai nder , pr edomi nant l y secret ari al , admi ni st r at i ve and
clerical staff, wer e cat egor i zed as low status. No i ndependent check was
avai l abl e as t o t he equi val ence of t he cut - of f poi nt bet ween hi gh and low
st at us in di f f er ent nat i ons. Howe ve r it is likely t hat mos t r es pondent s woul d
have been correct l y assi gned by this sel f - r epor t pr ocedur e. A br eakdown
of t he sampl es f or each count r y by gender and st at us is shown in Table I.
Procedure
The Rot t er locus of cont r ol i t ems wer e t r ansl at ed i nt o t he l anguages
shown in Table I. Al t hough t he me t hod of back- t r ansl at i on (e.g., Brislin,
Lonner , & Thomdi ke, 1973) was in ma ny cases not utilized, t he var i ous
versi ons of t he quest i onnai r e wer e checked by bi l i ngual c ompa ny r epr e-
sent at i ves, usually f r om per s onnel depar t ment s . The locus of cont r ol scale
i ncl uded five filler i t ems, whi ch wer e not t he same as t he six used by Rot t er
(1966).
The affect i vi t y i t ems wer e adapt ed by Tr ompenaar s (1985) f r om De a n
(1981). Re s ponde nt s we r e r e que s t e d t o pr ovi de t hei r fi rst and s econd
choi ces in r esponse t o five scenari os, t wo of which ar e r epr oduced bel ow:
Exampl e 1
He r e ar e f our gener al t ypes of peopl e we see ar ound us in our daily
lives. Look at t hese descr i pt i ons careful l y and deci de whi ch one you
62 Smith et ai.
would most like to resemble, then the one you would second most
like to resemble:
A) A person who is est eemed by others and takes a cont i nuous
interest in human welfare in general.
B) A person who is enj oyed by ot hers and takes their joys and
sorrows as they come, from day to day. *
C) A person who is enjoyed by others and takes a continuous interest
in the personal welfare of all those who are dear to him/her. *
D) A person who is approved of by others and attends to his/her
affairs conscientiously from day to day.
Example 2
Her e is another list of four types of people. Of these four what type
do you most prefer to have as friends?
A) Peopl e who compl et el y accept you the way you are and feel
responsible for your personal problems and welfare. *
B) People who do their work, attend to their affairs and leave you
free to do the same.
C) Peopl e who try to improve themselves and have definite aims in
life.
D) People who are friendly, lively and enjoy getting together to talk
or socialize. *
In the examples above, the asterisked response options denot e items
that were coded as affective rather than affectively neutral. For each re-
spondent an overall affectivity score was computed based on the number
of affectively-oriented responses, with first choices weighted twice as heavily
as second choices.
Since bot h the affectivity items and the Rot t er items rely upon subjects
ranking a response alternative above one or more other responses, it is
possible to analyze the resulting data set at bot h the culture-level and the
individual-level without the need for within-subject standardization. This
procedure is required to correct for cultural differences in response bias
only when data based on ratings are used.
In addition to the variables so far present ed a measure representing
national ' modernity' was required. This was constructed using one indica-
tor from each of the areas employed by Williams & Best (1990). National
dat a for aggregated income per capita, life expectancy and literacy from
47 nations (Economist Book of Vital World Statistics, 1990) were factor
analyzed. The f ast factor account ed for 81% of the variation among these
measures. Scores for this first factor were used as the modernity measure.
Locus o f Cont rol and Affecti vi ty 63
Interaction terms were comput ed by taking the products of the modernity
measure, which was a continnous variable, with dummy variables repre-
senting the dichotomous gender and status variables, in order to test hy-
pot heses 3 and 6.
Data Analysis
Table I shows that the sample sizes differ considerably, ranging from
755 in the Netherlands high status male cell to only 16 in the Indian lower
status female cell. The hypotheses were first tested at the culture-level. At
this level, the preferred strategy to deal with this imbalance is to use ag-
gregated mean data within each cell. The use of aggregated data in cul-
ture-level studies is widespread (e.g., Hofstede, 1980; Schwartz, 1994).
RESULTS
Country-Level Hypothesis Tests
Means and standard deviations for internality and affectivity for each
country by status and gender are shown in Tables II and III. Cronbach
alpha coefficients for the aggregated data are .81 for the 23 internality items
and .61 for the five affectivity items. The latter figure is below the level
normally regarded as acceptable. The results of hypothesis tests with this
variable therefore need to be interpreted with caution. However, the coun-
try scores for the two dependent variables were found to be substantially
correlated. External locus of control was significantly and strongly linked
to affectivity (r = .66, p < .001, 1 tail). This relationship is greater still if
correction for attenuation is applied. The data are therefore further ana-
lyzed using multivariate analysis of variance.
Table IV shows the multivariate results of the MANOVA. The results
support the hypotheses of main effects for gender (p < .005) and status
(p < .01). A significant multivariate effect is also obtained for modernity
(p = .001), but this is in the reverse direction to that which might have
been expected. However, there is no support in this study for the influence
of modernity as a moderat or of the relationship bet ween the dependent
variables and either gender or status. Nor is there support for a gender by
status interaction effect. The three way interaction was not tested.
Table V shows the univariate tests for internality. Significant main ef-
fects in the expected direction are found for gender (p = .001, ETA square
= .20) and status (t7 < .005, ETA square = .17), with male samples and
6 4 S mi t h e t a l .
Table II. Locus of Cont rol Means and St andar d Deviations by Country, Gender , and
St at us a
Occupational Status
Lower Hi gher
Femal e Male Female Mal e
Count ry Mean SD Mean SD Mean SD Mean SD
Belgium 12.14 3.55 11.59 3.57 10.96 3.76 9.31 4.34
Brazil 10.34 2.76 8.15 3.45 9.80 2.91 8.57 4.14
Bulgaria 10.59 3.30 10.31 2.99 10.14 3.48 9.79 2.85
Chi na 11.96 3.18 10.19 3.74 10.88 3.18 9.99 3.42
Czechoslovakia 10.99 3.59 9.46 3.20 10.07 3.52 8.70 3.71
Hungary 12.15 3.05 11.86 3.11 9.08 3.24 9.14 3.43
Indi a 10.56 4.26 8.41 2.59 10.57 4.00 8.11 4.23
Mexico 8.91 3.67 7.50 2.85 7.49 3.19 7.32 3.02
Net herl ands 13.18 3.51 10.10 3.58 10.46 3.28 9.32 3.52
Poland 13.55 3.35 12.10 3.29 11.64 4.55 9.56 3.75
Rumani a 8.17 4.29 10.68 2.62 7.46 3.52 9.48 3.55
Sweden 12.96 3.69 11.69 3.72 12.70 3.82 9.74 4.00
U.K. 12.88 3.49 11.97 3.67 11.55 3.77 9.52 4.04
Ex-U.S.S.R. 12.41 3.56 11.23 2.93 12.79 3.69 9.23 4.38
aHi gher scores indicate great er externality.
Table I l L Affective Ori ent at i on Means and St andard Deviations by Country, Gender , and
Status a
Occupational Status
Lower Hi gher
Female Mal e Femal e Mal e
Count ry Mean SD Mean SD Mean SD Mean SD
Belgium 3.80 2.20 4.00 1.79 4.90 1.63 5.23 1.60
Brazil 4.73 1.65 5.03 1.84 5.30 1.86 4.89 1.48
Bulgaria 4.21 1.61 3.86 1.66 4.30 1.66 4.16 1.72
Chi na 4.26 1.98 4.17 1.54 4.59 1.40 4.56 2.07
Czechoslovakia 4.04 1.50 4.06 1.56 4.19 1.50 4.50 1.80
Hungary 4.14 1.98 4.55 1.62 5.15 1.86 4.91 2.01
Indi a 3.54 2.48 3.37 1.48 3.82 1.69 4.59 1.89
Mexico 5.56 1.79 5.65 1.82 5.68 1.75 6.21 1.65
Net herl ands 3.13 1.70 3.79 1.72 4.01 1.53 4.61 1.77
Poland 3.77 1.64 4.69 2.00 4.73 1.98 4.95 1.79
Rumani a 4.44 1.24 4.80 1.78 5.21 1.67 4.84 1.60
Sweden 3.88 1.30 4.69 1.75 4.03 1.16 4.67 1.53
U.K. 3.27 1.68 3.72 1.86 3.43 1.47 4.10 1.72
Ex-U.S.S.R. 3.34 1.53 4.57 2.02 3.73 1.67 4.48 1.96
aHigher scores indicate a less affective ori ent at i on.
Loc us o f Control a nd Affecti vi ty
Table I~. Locus of Control and Affectivity by Gender, Occupational Status,
and Modernity: Multivariate Tests of Significance a
Test Wilks's Exact F Significant F Effect Size
65
Status .814 5.50 .007 .186
Gender .797 6.12 .004 .203
Modernity .755 7.79 .001 .245
Gender x Modernity .975 0.62 .543 .025
Status x Modernity .945 1.40 .257 .055
Gender x Status .994 0.15 .862 .006
aAll tests 2, 48 df.
Table V**Univariate F Tests for the Locus of Control
Variable a
F ETA
"l~st F Significance Square
Status 10.20 .002 .172
Gender 12.25 .001 .200
Modernity 13.78 .001 .220
Gender Modernity 0.08 .776 .002
Status x Modernity 1.57 .216 .031
Gender x Status 0.06 .813 .001
aAll tests 1, 49 dr.
hi gher st at us sampl es scori ng mor e i nt ernal l y. Thi s r el at i onshi p is cl earl y
di scerni bl e f r om Table II, f r om which it can be seen t hat in 12 of t he 14
count ri es, l ower st at us f emal es obt ai n t he mos t ext er nal scores, and hi gher
st at us mal es obt ai n t he mos t i nt er nal scores. The r el at i on of i nt ernal i t y t o
moder ni t y is al so significant (p = .001). Sampl es hi gh in moder ni t y score
mor e ext er nal l y on t he locus of cont r ol scale. The i nt er act i on effect s do
not even appr oach significance.
Table VI shows t he uni var i at e t est s f or affectivity. The r e is a si gni fi cant
ef f ect f or status, wi t h l ower st at us gr oups scori ng hi gher in affectivity, as
hypot hesi zed (p < .01, ETA s quar e = .13). The r el at i onshi p bet ween gen-
der and affect i vi t y is significant onl y at t he less st r i ngent .1 l evel ( ETA
s quar e = .06). Exami nat i on of Tabl e I I I shows t hat in 11 of t he 14 count r i es,
l ower st at us f emal es r epor t t he mos t affect i ve or i ent at i on and in 7 cases,
hi gher st at us mal es r epor t t he mos t affect i vel y neut r al or i ent at i on.
Over al l t her e is subst ant i al s uppor t f or Hypot hes es I , 2 and 4, and
mode s t s uppor t f or Hypot hesi s 5. The moder at i ng ef f ect of moder ni t y was
not suppor t ed. The i nsubst ant i al ef f ect sizes suggest t hat this fai l ure is not
a funct i on of l ow statistical power .
66
Table VL Univariate F Tests for the Afffectivity Variable a
F ETA
Test F Significance Square
Status 7.40 .009 .131
Gender 2.87 .097 .055
Modernity 1.11 .298 .022
Gender x Modernity 0.49 .487 .010
Status Modernity 0.02 .875 .001
Gender x Status 0.07 .799 .001
aAll tests 1, 49 df.
Smi t h e t al.
Individual-Level Hypothesis Tests
A further series of hierarchical multiple regression analyses were com-
put ed to test the hypotheses, using the internality and affectivity scores of
individuals, rather than cultural level data. A pan-cultural factor analysis
of the Rot t er scale items also made it possible to analyze scores on four
locus of control subscales.
Pan-Cultural Factor Analysis. Th e larger number of responses available
for individual-level analysis makes it possible to examine different aspects
of locus of control within the overall range spanned by Rotter' s rather het-
erogeneous items. A pan-cultural factor analysis of the internality items
was therefore conducted. Although the factor structure of the Rot t er scale
has already been analyzed extensively (Dyal, 1984), this has occurred within
single nation samples. The present investigation a11ows us to examine the
structure within a diverse cross-cultural sample. The binary nature of Rot t er
scale items precludes the occurrence of systematic cultural differences in
response bias. Dat a standardization before factor analysis is therefore not
required.
Since country samples were unequal, data were randomly discarded
until each of the 14 countries contained 90 cases. The total sample was
thereby reduced to 1,260 cases. The correlation matrix for the 23 items
was analyzed using principal components analysis. After examining factor
loadings for obliquely rot at ed 2, 3, 4, & 5 factor solutions, we decided that
the four factor solution was the most meaningful. Before rotation, the four
factors accounted for 14.4%, 6.6%, 5.9% and 5.6% of common variance
respectively.
As can be seen from Table VII, each of the four factors is clearly in-
terpretable with the highest loading items within a factor cohering in a
meaningful way. The four factors were named ' Luck,' 'Socio-Political Con-
trol,' ' Effort,' and Nctive Friendship.' Factor scores for the four locus of
Lo c us o f Cont rol and Affecti vi ty 67
Table VII. Factor Loadings of Rot t er It ems in t he Pancultural Factor Analysis a
Factor
Rot t er Scale It em 1 2 3
Many of t he unhappy things in peopl e' s lives are partly due to bad .59
luck. ( E) / / People' s misfortunes result from t he mistakes they
ma k e . ( I )
Many times I feel t hat I have little influence over t he things t hat .54
happen t o me. ( E ) / / I t is impossible for me to believe t hat
chance or luck plays an i mport ant role i n my life. (I)
Becomi ng a success is a mat t er of har d work, luck has little or .52
not hi ng t o do with it. ( I ) / / Ge t t i n g a good j ob depends mainly
upon being in t he right place at t he ri ght time. ( E)
Most people don' t realize t he extent t o which t hei r lives are .50
cont rol l ed by accidental happenings. ( E ) / / T h e r e is really no
such thing as luck. (I)
Who gets to be boss oft en depend supon who was lucky enough t o .50
be in t he right place first. (E) / / Get t i ng people t o do the right
t hi ng depends upon ability. Luck has little or not hi ng t o do with
it. (I)
In t he l ong run, t he bad things t hat happen t o us are balanced by
t he good ones. ( E ) / / Mo s t misfortunes are t he result of lack of
ability, ignorance, laziness, or all t hree. (I)
In my ease getting what I want has little or not hi ng to do with luck.
( I ) / / Ma n y t i mes we might as well decide what t o do by flipping
a coin. (E)
When I make plans, I am almost cert ai n t hat I can make t hem
work. ( I ) / / I t is not always wise t o plan t oo far ahead because
many things t ur n out to be a mat t er of good or bad fortune
anyhow. (E)
The average citizen can have an influence in government al
decisions. ( I ) / / Th i s world is run by a few people in power and
t here is not much t he little guy can do about it. (E)
With enough effort we can wipe out political corruption. ( I ) / / I t is
difficult to have much control over t he things politicians do
when in office. ( E)
As far as world events are concerned, most of us are t he victims of
forces we can nei t her underst and, nor control. ( E ) / / B y taking
an active part in political and oscial affairs, people can control
world events. (I)
One of t he maj or reasons why we have wars is because people don' t
take enough interest in politics. ( I ) / / T h e r e will always be wars,
no mat t er how har d people try t o prevent them. (E)
Most of t he time I can' t underst and why politicians behave t he way
t he do. ( E ) / / I n t he long run, t he peopl e are responsible for
bad gover nment on a national as well as on a local level. (I)
In t he case of t he well prepared student, t her e is rarely, if ever,
such a t hi ng as an unfai r test. ( I ) / / Of t e n exam questions t end
t o be so unrel at ed to course work t hat studying becomes really
useless. (E)
.48
.46
.46
.70
.67
.59
.51
.47
.64
68
Table VII. Continued
Smith et aL
Rotter Scale Item
Factor
1 2 3 4
Sometimes I couldn't understand how teachers arrived at the grades
they gave. ( E) / / Ther e is a direct connection between how hard
I studied and the grades I achieved (I)
In the long run people get the respect they deserve in this world.
(I)//Unfortunately, an individual's worth often passes
unrecognised no matter how hard he tries. (E)
The idea that teachers are unfair to students is nonsense. (I)/ /
Most students don' t realize the extent to which their grades are
being influenced by accidental happenings. (E)
It is hard to know whether or not a person really likes you. (E)/ /
How many friends you have depends upon how nice a person
you are. (I)
No matter how hard you try, some people just don't like you. (E)/ /
People who can't get others to like them don't understand how
to get along with others. (I)
People are lonely because they don' t try to be friendly. (I)/ / There' s
not much point in trying too hard to please people. If they like
you, the like you. (E)
.55
.46
.43
.63
.60
.48
altems scored in external direction. (I) Internal; (E) External. Only loadings greater than .40
are shown.
c ont r ol subscal es we r e c o mp u t e d f or each r e s ponde nt , us i ng t he r egr es s i on
me t h o d . Th e me a n s f or t hes e f a c t or s cor es by st at us a n d g e n d e r a r e s hown
i n Tabl e VI I I .
Th e fi rst t wo f act or s ext r act ed, Luc k and Soci o- Pol i t i cal Cont r ol , ap-
p e a r si mi l ar t o f act or s des cr i bed i n n u me r o u s si ngl e c ount r y s t udi es a n d
s u mma r i z e d i n Dya l (1984). Fact or s si mi l ar t o o u r Ef f or t and Act i ve Fr i e nd-
shi p di me ns i ons have al so b e e n d o c u me n t e d , al bei t l ess f r equent l y, wi t hi n
t he l i t er at ur e. Fo r exampl e, Es c o v a r (1981) obt a i ne d a f a c t or l abel l ed as
Ac a d e mi c Co n t r o l f r o m a cr os s - cul t ur al s a mpl e o f 402 La t i n Ame r i c a n ,
U. S. Hi s pani c, a nd U. S. An g l o s t udent s wi t h t he t hr ee hi ghes t l oa di ng i t ems
ma t c h i n g t hos e f o u n d o n o u r Ef f o r t f act or . A si mi l ar f a c t or was i dent i f i ed
by Wa t s on (1981). The s e f act or s c o r r e s p o n d t o t he c o n c e p t u a l c a t e gor i z a -
t i on o f Ro t t e r i t ems or i gi nal l y ma d e by Sc hne i de r & Pa r s ons (1970). Pr e c e -
dent s f or o u r Act i ve Fr i e nds hi p f a c t or can be f ound i n Ab r a h a ms o n et al.
(1973), Li t t l e (1977), a nd Wa t s on (1981).
Wh y s houl d we have o b t a i n e d f our c o h e r e n t f act or s f r o m o u r pa n- c ul -
t ur al anal ysi s r a t h e r t ha n t he t wo t ha t have t e nde d t o e me r g e f r o m t he
f act or i ng o f Ro t t e r scal e i t e ms ? We bel i eve t hat t he ans wer is l i kel y t o
r es i de i n t he i nc r e a s e d s a mpl e he t e r oge ne i t y o f a mul t i pl e c o u n t r y st udy,
Locus of Control and Affectivity
Table VIII. Aggregated Means and Standard Deviations on the Four Locus of
Control Factors by Gender and Status a
69
Occupational Status
Lower Higher
Factor Male Female Male Female
Luck .61 (.21) .65 (.23) .57 (.15) .67 (.23)
Sociopolitical Control .71 (.13) .84 (.21) .59 (.12) .71 (.13)
Effort .24 (.21) .27 (.25) .12 (.20) .13 (.26)
Active Friendship .69 (.25) .73 (.18) .77 (.17) .79 (.21)
aN = 56. Within each of the four gender by occupation cells, each cell entry represents
a mean or standard deviation computed so that each country is given unit weighting.
and the concomitant increase in the variance of scale items. As an illus-
tration, our Effort factor largely reflects the extent to which effort is per-
ceived as leading towards positive academi c out comes. Assuming that
respondents' perceptions are to some extent veridical, then the greater the
variation in objective circumstances in educational systems, the greater the
variation in responses. In a sample such as the present one, which includes
respondents from countries differing widely on important cultural variables
such as modernity, there will be much greater variation in socialization ex-
periences than in mono-cultural studies (especially those in which the sam-
ple has been drawn from a single educational institution). Hence, what was
a weak or non-existent factor in a homogenous sample may be clearly de-
fined in a diverse sample. Similar comments could be made about our Ac-
tive Friendship factor.
Individual Level Stepwise Regressions. A series of stepwise multiple re-
gression analyses tested the hypotheses. In addition to the regressions for
locus of control and affectivity, regressions were calculated using the four
locus of control subscales. In each regression, dummy variables representing
country were entered first. Next, occupational status was entered, followed
by gender, and then a gender by occupational status interaction term. As
gender enters the regression equations after occupational status, the pos-
sibility that a significant gender effect occurs as a consequence of a con-
found of gender and occupational status is minimized.
The results of the stepwise regressions are shown in Table IX. For the
Rot t er scale, significant effects were again obtained for bot h gender and
occupational status Co < .0001). The patterning of results for the locus of
control subscales is more complex. The Luck factor yields a significant ef-
fect for gender (p < .001) but not for occupational status. The results for
Soeio-political Control and Effort show significant effects for bot h gender
and occupational status. However, the status effects are larger than the
70 Smi t h et al .
gender effects (status, p < .0001 on both factors; gender p < .001 on "so-
eiopoliticar' and p < .05 on "effort" factors). No significant effects were
obtained for Active Friendship. Affectivity shows significant relationships
with both status (p < .0001), and gender (p < .05). No significant inter-
action effects were detected on any of the six measures.
It is also of some interest to examine the relation between affectivity
scores and the locus of control subscales at the individual level. Separate
correlations were computed within each country's data set, and mean cor-
relations were then computed, weighting data from each nation equally.
As was found at the country level, affectivity is significantly related to ex-
ternality (mean r = 0.18, standard deviation around this mean, 0.15; this
mean is significantly different from zero, t = 6.5, 13df, p < 0.001). Mean
values of r for the subscales were Luck, 0.19 (SD = 0.15; p < 0.001); So-
cio-political control, 0.10 (SD = .12; p < 0.01); Effort, 0.09 (SD = 0.12;
p < 0.01); and Friendship, 0.02 (SD = .08; ns). Thus at the individual level
affectivity is associated weakly but significantly with most aspects of exter-
nality.
D I S C U S S I O N
The results of our analyses reveal some clear trends. However, it is
noticeable that the effect sizes observed only account for small proportions
of the overall variation in the dependent variables.
Modernity
While the principal purpose of this paper was to explore the interre-
lationships of status and gender in different cultures, one of our most strik-
ing findings was obtained with the modernity index. Internal locus of
control was markedly higher in the less modern nations such as Rumania
and Mexico than it was in more modern nations such as Netherlands and
Sweden. This finding is not an artefact of the particular sampling of the
14 nations in the present study. The same pattern is found within the full
sample of 43 nations for which data are available (Smith, Dugan, & Trom-
penaars, 1995). However, the effect found may well be explained by the
nature of the samples upon which these studies are based. The type of
organizations in developing countries who were included in Trompenaars'
databank would be those which were most in touch with Western manage-
ment theories, and which were probably in process of rapid expansion.
Their employees might well experience a high degree of internality. Or-
Locus of Control and Affectivity 71
Table IX. Stepwise Regressions Predicting Locus of Control, Locus of Control Factors, and
Affectivity a
R Square Beta and
Step and Variable R Square change Significance
Locus of Control scale
1: Country .066 .066 b
2: Status .108 .041 -.212 b
3: Gender .133 .025 .163 b
Status -.179 b
4: Gender x Status interaction .135 .002 .076 NS
Luck factor
1: Country .116 .116 b
2: Status .117 .001 -.026 NS
3: Gender .125 .008 .092 c
Status -.007 NS
4: Gender x Status interaction .127 .002 .089 NS
Sociopolitical factor
1: Country .091 .091 b
2: Status .113 .022 -.154 b
3: Gender .122 .009 .097 c
Status -.134 b
4: Gender x Status interaction .122 .000 -.024 NS
Effort factor
1: Country .255 .255 b
2: Status .283 .028 -.173 b
3: Gender .286 .003 .056 a
Status -.162 b
4: Gender x Status interaction .286 .000 .024 NS
Active Friendship factor
1: Country .157 .157 b
2: Status .158 .002 .043 NS
3: Gender .158 .000 -.001 NS
Status .042 NS
4: Gender x Status interaction .158 .000 .008 NS
Affectivity
1: Country .067 .067 b
2: Status .088 .021 .152 b
3: Gender .092 .003 -.057 '/
Status .140 b
4: Gender Status interaction .092 .000 .018 NS
aln the table there are two entries for status. Status is entered
the equation at step 3. N = 1043.
~p < .0001.
~ p
< .001.
< .05.
at step two, and remains in
ganizations in more developed economies which provided data for Trom-
penaars' databank were predominantly well-established large business or-
gani zat i ons, wi t hi n whi ch mi ddl e and j uni or-l evel empl oyees mi ght
72 Smi t h et al .
experience greater externality. Our results concerning modernity and locus
of control might therefore best be read as reminding us that different sub-
samples within national cultures will not always reproduce differences to
be expected from broader characterizations of those cultures. Hofst ede' s
characterization of cultures was based upon comparisons of matched sam-
ples. Williams and Best sampled student populations. Our samples were
less well matched, but this does not detract from the tests of our principal
hypotheses, since the impact of modernity was considered separately within
our analyses.
Locus of Control
There is clear evidence from this dataset, whether analyzed at the in-
dividual or the cultural level, that locus of control scores differ according
to two fundamental social categories, gender, and status. Females and lower
status employees tend to score more externally on Rotter' s locus of control
scale. Our individual level analyses show that, at least in the present sam-
pies, the association with status is largely due to differences on Socio-po-
litical Control and Effort. The association with gender was significant on
three for the four subscales, most particularly Socio-political Control. These
findings are consistent with bot h theoretical predictions and the not incon-
siderable body of existing research on locus of control. With regard to the
status results, our cross-sectional dat a does not permit any conclusions con-
cerning causal factors generating the findings. The dat a does, however,
point to a phenomenon in need of explanation. Whet her internal locus of
control facilitates recruitment into higher status occupations, whet her locus
of control changes in line with status, or whether reciprocal causation oc-
curs, remains open to further investigation.
The significant main effect for gender is interesting. Women at all lev-
els lay more emphasis on luck as a causal explanation. They also t end t o
have a naore external locus of control on the Rot t er scale than men within
a given occupational stratum. Only in Romani a was this pattern reversed.
Perhaps this gives some credence to the notion that locus of control is
determined, at least in part, by patterns of social advantage and disadvan-
tage. According to this view, men in higher status (advantaged) occupations
should exhibit the most internal, and women in lower status (disadvan-
taged) occupations, the most external locus of control in accordance with
their self-perceived potential to influence the course of their lives. This is
the pattern that was obtained. Locus of control may also be taken as an
' agentic' trait, as we have previously suggested.
Locus of Control and Affectivity 73
Our results contrast with those obtained by Cooper, Burger, & Good
(1981), from their meta-analysis of gender differences in academic locus
of control among school-aged US children. They found marginally greater
internality among school-aged females. Their results suggest to us that the
patterning of gender differences in internality may differ across the lifespan,
with greater female externality emerging only in young adulthood. The ra-
tionale for this expectation is the assumption that the school and college
environment presents a more 'level playing field' for males and females
than does the adult work environment where various structural variables
favor male rather than female achievement. From this assumption it follows
that in the adult world, females, as a group, experience disproportionate
numbers of non-contingent reinforcements, as compared with males, a state
of affairs differing from that found in school and college environments. As
a result, adult females will exhibit greater externality than males. We would
expect these changes to occur before the age of 30 or so. McCrae & Costa
(1990) have convincingly argued from a firm empirical base that, at least
for US samples, personality traits remain extremely stable in adulthood.
These points, which are consistent with Eagly's social role interpretation
of gender differences, are of course merely speculative. The Cooper et al.
(1981) findings may well be culturally bounded. Certainly, the pancultural
validity of their findings has yet to be demonstrated.
The findings for internality need some further qualification. Firstly,
the samples are drawn predominantly from West and East European busi-
ness organizations, a factor which constrains the types of generalizations
that can be made. However, this can be viewed as a strength, as large num-
bers of ot her studies have used student samples. Fuller samples from ot her
continents are desirable. Nevertheless, the non-European samples that were
included in the present investigation exhibited similar patterns to the Euro-
pean samples. A second limitation is that status differences were only ex-
amined within two rather broad categories. Further differentiation of status
categories would have introduced greater unreliability in categorization and
substantially reduced the number of country samples that had sufficient
cases, limiting the scope for pancultural generalizations.
The dimensionality of the Rot t er scale has been frequently debat ed
(Dyal, 1984). It has appeared from single-country studies that the Rot t er
scale does not adequately represent the various potential locus of control
domains (e.g., academic, socio-political, interpersonal, etc.) Furthermore,
the various ' facets' of the total scale are not weighted in such a way as to
equally represent these domains. As such, the overall scale is at best a
crude measure of generalized expectancies. The four factors which emerged
from our pan-cultural analysis are rather similar to the range of those which
have been found within single-country factor analyses, thereby strengthen-
74 Smi t h et al .
ing considerably the case for using subscale scores rather than the overall
score. A country-level analysis of the full 43-nation dataset of Rot t er re-
sponses from which the present dat a were drawn found three related di-
mensions (Smith, Dugan & Trompenaars, 1995).
The individual-level analysis indicated that the aspects of internality
upon which those in senior positions differed from others were the scales
made up of the items referring to Socio-Political Control and Effort in
academic settings. It is highly plausible that those at more senior levels in
organizations would experience greater internality in these domains.
A f f e c t i v i t y
The results involving affectivity to a large extent parallel those for in-
ternality. As expected, we have found a pancultural relationship bet ween
affectivity and status. This association is of moderat e size and appears quite
robust, especially as the affectivity variable exhibited only moderat e internal
consistency. An improved measure will be required before one could be
fully confident of the relationship found. A marginal effect for gender was
also obtained, and there were no significant interactions with modernity.
The individual level analysis mirrored these findings. A further issue is that
affectivity appears at face value to be linked with two of the ' Big Five'
personality factors, namely agreeableness and conscientiousness. If this is
so, we cannot certainly suggest whet her lower status samples score high on
affectivity because they are higher on communal orientation or because
they are lower on conscientiousness, or some combination of the two. This
conceptual ambiguity clouds the interpretation we can offer for this finding.
Perhaps the most salient finding from this study has been the strong
association bet ween our measures of locus of control and affectivity at the
aggregated level, and more weakly at the individual level. This result no
doubt occurs because both variables are currently associated with gender
and status. Hi gher status employees tend to be both more internal and less
communal/more conscientious. Women tend to be more external and pos-
sibly more communal. There is no theoretical reason to think that women
are any less conscientious than men, nor any indication from the present
results. However there are ample reasons to think that lower status em-
ployees of either gender are bot h less conscientious and more communal.
As we not ed earlier, Barrick and Mount (1991) confrrmed empirically that
conscientious persons tend to get promot ed. It follows that wherever in
the world organizations promot e peopl e to positions of seniority on criteria
which are not gender-related, we may expect gender differences in consci-
entiousness/communality to be attenuated. Some other aspects of gender
Loc us o f Cont rol and Af f ect i vi t y 75
may persist if, as our dat a indicated, senior women cont i nue to experi ence
less i nt ernal i t y t han senior men. Our results underl i ne the ext ent and rep-
licability of these effects across a range of cultures which differ in many
ot her i mpor t ant aspects.
REFERENCES
Abrahamson, D., Schluderman, S., & Schluderman, E. (1973). Replications of dimensions of
locus of control. Journal o f Consulting and Clinical Psychology, 41, 320.
Aggarwal, Y. P., & Kumari, M. (1975). A study of locus of control of boys and girls of various
levels of soci oeconomi c status. Kurukshet ra University Research Journal (Arts and
Humanities), 9, 26-32.
Bakan, D. (1966). The duality o f human existence. Chicago: Rand McNally.
Barrick, M. R., & Mount , M. K. (1991). The Big Five personality dimensions and j ob
performance: A meta-analysis. Personnel Psychology, 44, 1-26.
Bar-Tal, D., & Bar-Zohar, Y. (1977). The relationship between perception of locus of control
and achievement. Contemporary Educational Psychology, 2, 181-199.
Bern, S. L. (1974). The measurement of psychological androgyny. Journal o f Consulting and
Clinical Psychology, 42, 155-162.
Best, D. L., Williams, J. E., & Briggs, S. R. (1980). A further analysis of the affective meanings
associated with male and female sex-trait stereotypes. Sex Roles, 6, 735-746.
Bhattacharya, R., & Husain, M. G. (1985). Openness-closeness of environmental entity, sex
and age: Determinants of locus of control. Psychologia, 28, 166-172.
Bond, M. H. (1988). Fi ndi ng universal dimensions of individual variation in multicultural
studies of values: The Rokeach and Chinese value surveys. Journal o f Personality and
Social Psychology, 55, 1009-1015.
Bond, M. H., & Smith, P. B. (1996). Cross-cultural social and organizational psychology.
Annual Review o f Psychology, 47, 205-235.
Brislin, R., Lonner, W., & Thorndike, R. M. (1973). Cross-cultural research methods. New
York: Wiley.
Buss, D. M. & 49 co-authors (1990). International preferences in selecting mates: A study of
37 cultures. Journal o f Cross-Cultural Psychology, 21, 5-47.
Cha, J. H. (1994). Aspects of individualism and collectivism in Korea. In U. Kim, H. C.
Triandis, C. Kagitcibasi, S.-C. Choi, & G. Yoon (Eds.), Individualism and collectivism_"
Theory, met hod and applications. Thousand Oaks, CA: Sage.
Cole, D. L., & Cole, S. (1977). Cotmternormative behavior and locus of control. Journal o f
Social Psychology, 101, 21-28.
Cooper, H. M., Burger, J. M., & Good, "12 L. (1981). Gender differences in the academic
locus of control beliefs of young children. Journal o f Pe~onality and Social Psychology,
40, 562-572.
Dean, L. R. (1961). The pattern variables: Some empirical operations. American Sociological
Review, 26, 80-90.
De Brabander, B., & Boone, C. (1990). Sex differences in perceived locus of control. Journal
o f Social Psychology, 130, 271-272.
Digman, J. M. (1990). Personality structure: Emergence of the five-factor model. Annual
Review o f Psychology, 41, 417-440.
Doherty, W. J., & Baldwin, C. (1985). Shifts and stability in locus of control during the 1970's:
Divergence of the sexes. Journal o f Personality and Social Psychology, 48, 1048-1053.
Dyal, J. A. (1984). Cross-cultural research with the locus of control construct. In H. M.
Lefcourt (Ed.), Research with the locus o f control construct (Vol. 3): Extensions and
limitations. New York: Academic Press.
7 6 S mi t h e t a l .
Eagly, A. H. (1987). Sex differences in social behavior. A social role interpretatior~ Hillsdale,
N J: Lawrence Erlbaum Associates.
Eagly, A. H., & Carli, L. L. (1981). Sex of researchers and sex-typed communications as
determinants of sex differences in influenceability: A meta-analysis of social influence
studies. Psychological Bulletin, 90, 1-20.
Eagly, A. H., & Crowley, M. (1986). Gender and helping behavior: A meta-analytic review
of the social psychological literature. Psychological Bulletin, 100, 283-308.
Eagly, A. H., & Steffen, V. J. (1986). Gender and aggressive behavior: A meta-analytic review
of the social psychological literature. Psychological Bulletin, 100, 30%330.
Eagly, A. H., & Wood, W. (1991). Explaining sex differences in social behavior: A meta-analytic
perspective. Personality and Social Psychology Bulletin, 17, 306-315.
The Economi st Book o f l, qtal World Statistics: A Complete Guide to the World in Figures. (1990).
London: Hutchinson.
Escovar, L. A. (1981). Di mensi onal i t y of the I-E scale: A cross-cultural comparison.
International Journal o f Group Tensions, 1L 17-33.
Feather, N. T. (1967). Some personality correlates of external locus of control. Australian
Journal o f Psychology, 19, 253-260.
Gilligan, C. (1982). I n a di f f erent voice: Psychological theory and women' s devel opment .
Cambridge, MA: Harvard University Press.
Goldberg, L. R. (1990). An alternative "description of personality." The big-five factor
structure. Journal o f Personality and Social Psychology, 59, 1216-1229.
Gough, H. G., & Heilbrun, A. B. (1980). The adjective checklist manual. Palo Alto, CA:
Consulting Psychologists Press.
Hall, J. A. (1984). Nonverbal sex differences: Communication accuracy and expressive style.
Baltimore, MD: Johns Hopkins University Press.
Hall, J. A., & Halberstadt, A. G. (1986). Smiling and gazing. In J. S. Hyde & M. C. Linn
(Eds.), The psychology o f gender. Advances through meta-analysis. Baltimore, MD: Johns
Hopkins University Press.
Hofstede, G. (1980). Culture's consequences: International differences in work-related values.
Beverly Hills, CA: Sage.
John, O. P. (1989). Towards a taxonomy of personality descriptors. In D. M. Buss & N. Cantor
(Eds. ), Personal i t y psychology." Recent t rends and emergi ng directions. New York:
Springer-Verlag.
Jorm, A. E (1987). Sex differences in neuroticism: A quantitative synthesis of published
research. Australian and Ne w Zealand Journal o f Psychiatry, 21, 501-506.
Kalin, R., & Tilby, P. (1978). Development and validation of a sex-role ideology scale.
Psychological Reports, 42, 731-738.
Kanter, R. M. (1977). Men and women o f the corporation. New York: Basic Books.
Kashima, Y., Yamaguchi, S., Kim, U., Choi, S.-C., Gelfand, M., & Yuki, M. (1995). Culture,
gender and self: A perspective from individualism-collectivism research. Journal o f
Personality and Social Psychology, 69, 925-937.
Khanna, P., & Khanna, J. L. (1979). Locus of control in India: A cross-cultural perspective.
International Journal o f Psychology, 14, 207-214.
Lao, R. C., Chuang, C. J., & Yang, K. S. (1977). Locus of control and Chinese college students.
Journal o f Cross-Cultural Psychology, 8, 299-313.
Lee, V. K., & Dengerink, H. A. (1992). Locus of control in relation to sex and nationality:
A cross-cultural study. Journal o f Cross-Cultural Psychology, 23, 288-397.
Lefcourt, H. M. (1982). Locus o f controL" Current trends in theory and research. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Lefcourt, H. M. (1981). Research with the locus o f control const ruct Assessment met hods (Vol.
1). New York: Academic Press.
Leung, K., & Bond, M. H. (1989). On t he empirical identification of di mensi ons for
cross-cultural comparisons. Journal o f Cross-Cultural Psychology, 20, 133-151.
Little, C. B. (1977). Dimensionality of the internal-external control scale: Some further
evidence. Journal o f Genetic Psychology, 131, 329-330.
Locus o f Cont rol a nd Affecti vi ty 77
Maccoby, E., & Jacldin, C. N. (1974). The psychology o f sex differences. Stanford, CA: Stanford
University Press.
McCrae, R. R., & Costa, P. T, Jr. (1990). Personality in adulthood. New York: The Guilford
Press.
McGinnies, E., Nordholm, L. A., Ward, C. D., & Bhanthumnavin, D. L. (1974). Sex and
cultural differences in perceived locus of control in five countries. Journal o f Consulting
and Clinical Psychology, 42, 451-455.
O' Brien, G. E. (1984). Locus of control, work, and retirement. In H. M. Lefcourt (Ed.),
Research with the locus o f control construct (Vol. 3): Extensions and limitations. New York:
Academic Press.
Parsons, O. A., & Schneider, J. M. (1974). Locus of control in university students from eastern
and western societies. Journal o f Consulting and Clinical Psychology, 42, 456-461.
Parsons, T, & Shils, E. A. (1951). The social syster, t New York: Free Press.
Parsons, T, & Bales, R. E (1955). Family, socialization and interaction process. New York: Free
Press.
Phares, E. J. (1976). Locus o f control in personality. Morristown, N J: General Learning Press.
Reykowski, J. (1994). Collectivism and individualism as dimensions of social change. In U.
Kim, H. C. Triandis, C. Kagitcibasi, S.-C. Choi, & G. Yoon (Eds.), Individualism and
collectivisrru Theory, met hod and applications. Thousand Oaks, CA: Sage.
Rot t er, J. B. (1966). Gener al i zed expect anci es for internal versus external cont rol of
reinforcement. Psychological Monographs, 80(Whole No. 609).
Schneider, J. M., & Parsons, O. A. (1970). Categories of the locus of control scale and
cross-cultural comparisons in Denmark and the United States. Journal o f Cross-Cultural
Psychology, 1, 131-138.
Schwartz, S. H. (1994). Beyond individualism/collectivism: New cultural dimensions of values.
In U. Kim, H. C. Triandis, C. Kagitcibasi, S.-C. Choi, & G. Yoon (Eds.), Individualism
and collectivism." Theory, met hod and applications. Thousand Oaks, CA: Sage.
Smith, P. B., Dugan, S., & Trompenaars, E (1995). The Rotter locus of control scale in 43
countries: A test of cultural relativity. International Journal o f Psychology, 30, 377-400.
Smith, P. B., Trompenaars, E, & Dugan, S. (1996). National culture and the values of
or gani zat i onal empl oyees: A di mensi onal analysis across 43 nat i ons. Jour nal o f
Cross-Cultural Psychology, 27, 231-264.
Spence, J. T., & Helmreich, R. L. (1978). Masculinity and femininity: Their psychological
dimensions, correlates and antecedents. Austin: University of Texas Press.
Trapnell, P. D., & Wiggins, J. S. (1990). Extension of the Interpersonal Adjective Scales to
include the big five dimensions of personality. Journal o f Personality and Social Psychology,
59, 781-790.
Trompenaars, E (1985). The organization o f meani ng and the meani ng o f organization.
Unpublished doctoral dissertation, University of Pennsylvania.
Watson, J. M. (1981). A note on the dimensionality of the Rotter locus of control scale.
Australian Journal o f Psychology, 33, 319-330.
Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes: A thirty nation study. Beverly
Hills, CA: Sage.
Williams, J. E., & Best, D. L (1990). Sex and Psyche: Gender and s e l f viewed cross-culturally.
Newbury Park, CA: Sage.
Yamaguchi, S. (1994). Collectivism among the Japanese: A perspective from the self. In U.
Kim, H. C. Triandis, C. Kagitcibasi, S.-C. Choi, & G. Yoon (Eds.), Individualism and
collectivism." Theory, met hod and applications. Thousand Oaks, CA: Sage.
Yang, IC S. (1988). Will modernization eliminate cross-cultural psychological differences? In
M. H. Bond (Ed.), The cross-cultural challenge to social psychology. Newbury Park, CA:
Sage.
Yuchtman, Y. E., & Shapira, R. (1981). Sex as a status characteristic: An examination of sex
differences in locus of control. Sex Roles, 7, 149-162.

Вам также может понравиться