Professional Documents
Culture Documents
Response Style in Cross-Cultural Persoanlity Assessment
Response Style in Cross-Cultural Persoanlity Assessment
(2008) CROSS-CULTURAL
PERSOANLITY ASSESSMENT IN Boyle, G.J., Matthews, G., Saklofske, D.H., The SAGE
Handbook of Personality Theory and Assessment: Personality Measurement and Testing, VoL.2,
SAGE, pp.55-73
RESPONSE STYLES
Greece, showed higher acquies-cent and extreme responding than north-western European
countries in surveys on consumer research.
They explained these differences in terms of the individualism versus collectivism
dimension, at the same time acknowledging that the causal relationship might well be reversed,
so that response styles are responsible for country scores on individualism and collectivism (see
also Berry et al., 2002). Two classes of country variables have been shown to be associated with
response styles in cross-cultural studies, namely socioeconomic context and cultural values.
Socioeconomic and political indicators such as gross national product (GNP), politi-cal rights,
and level of democracy seem to be relevant to socially desirable and acquiescent responding.
Van Hemert et al. (2002) reported significant correlations between these indicators and
social desirability as measured by the Eysenck Personality Questionnaire Lie scale across 31
countries; lower scores were found in wealthier and more democratic countries. Similarly,
Fischer et al. (submitted) correlated ARS scores based on 17 different instruments or scales to a
number of socioeconomic indica-tors and found the level of acquiescence to be negatively
correlated with survey quality, wealth-related indices, democratic rights, and positive
correlations with prevalence rates for psychiatric disorders.
They suggest that survey responses are systematic indica-tors of culturally appropriate
expressiveness that are also reflected in the larger sociopolit-ical climate of the societies. An
explanation of these findings might be found in the con-cept of resources. Ross and Mirowsky
(1984) found that external resources influence cross-cultural differences in social desirability;
less powerful (i.e. less resourceful) social groups often give more socially desirable responses.
Multicountry research on the importance of cultural values for response styles started with Bond
and Smith's (1996) meta-analysis of studies on the Asch conformity paradigm. Conformity can
be expected to tap something similar to social desirability.
They found several measures of cultural values, such as indi-vidualism (several
measures), autonomy and conservatism (Schwartz, 1994), to be strong predictors of conformity.
The authors argue that these measures reflect the individualism versus collectivism dimension in
such a way that more individualistic countries showed less conformity. Another concept close to
social desirability was studied by Triandis et al. (2001) in a 12-country study of deception (both
self-reported and other-reported) in scenarios of negotiation situations. High vertical
(hierarchical) collectivism was posi-tively related to the tendency to deceive. Yet, the pattern was
reversed at the individual level. Van Hemert et al. (2002) also found higher collectivism to be
associated with higher social desirability.
Harzing (2006) found in a study involving 26 countries that acquiescence and extreme
responding are more prevalent in countries with higher scores on Hofstede's collec-tivism and
power distance, and GLOBE's uncertainty avoidance. She also reported that extraversion (at
country level) is a positive predictor of acquiescence and extremity scor-ing. Finally, she found
that English-language questionnaires tend to evoke less extremity scoring and that answering
items in one's native language is associated with more extremity scoring. A convincing case for a
substantive cultural interpretation of response styles was made by Smith (2004). He calculated
acqui-escence scores for variables from five multi-country surveys on values, beliefs, reports of
own behavior, and reports of others' behavior (Hofstede, 1980; House et al., 2004; Leung et al.,
2002; Schwartz, 1994; Smith et al., 2002), and added social desirability scores based on the EPQ
(Van Hemert et al., 2002). The GLOBE study (House et al., 2004) provided indices on society 'as
it should be' and society 'as it is'. All six acquiescent response bias indices were positively
interrelated, and correlations with GLOBE 'as should be' were much stronger than correlations
with GLOBE 'as is'.
This can be taken as evidence for a common factor. Congruent with Harzing's (2006)
findings, acquiescent bias scores were positively related to Hofstede's collectivism and power
distance and GLOBE uncertainty avoidance. In addition, depend-ing on whether 'as is' or 'as
should be' scores were used to predict acquiescent bias, 'as is' family collectivism or 'as should
be' institu-tional collectivism and uncertainty avoidance explained cross-cultural variance. Smith
con eluded that response bias at a country level can be meaningfully interpreted in terms of
cultural values and reflects cultural commu-nication styles and patterns of intergroup ralatinne
This idea features in many recent cross-national studies, but specific correlations sometimes
differ. For example, Johnson et al. (2005) studied country-level correlates of individual-level
extreme responding across respondents from 19 countries and acquies-cent responding across
respondents from 10 countries using hierarchical linear model-ing. Extreme responding was
positively related to power distance and masculinity, and acquiescence was negatively related to
power distance, uncertainty avoidance, indi-vidualism, masculinity, and GNP.
These findings were only partly in agreement with other research. The correlations found
for extreme responding fitted the hypotheses and the negative correlations of acquiescence with
individualism and GNP replicated previous findings (Smith, 2004; Van Hemert et al., 2002).
Diverging results included the role of power distance in predicting acquiescence, which was
reversed in Smith's (2004) and Harzing's (2006) studies. This difference might be explained by
the limited number of countries for the acquiescence analyses and the multilevel design of the
study in which individual-level response styles were used rather than country-level scores, and
individual-level control variables such as gender, age, and employment were used. Another
multilevel study on extreme responding was performed by De Jong et al. (2007). They used a
method based on item response theory for measuring extreme responding, allowing for
differential useful-ness of items in the calculation of extreme responding. A multilevel analysis
(26 coun-tries, 12,500 individuals) with individual-level demographic variables (i.e. age, gender,
and education) and Hofstede's measures revealed significant effects of gender and age and
positive correlations with individual-ism, uncertainty avoidance, and masculinity. These results
differ from findings by Johnson et al. (2005), who did report a relationship with power distance
and not with individual-ism and uncertainty avoidance, and findings by Harzing (2006), who
also observed a positive correlation with power distance.
New developments
In our view, the study of response styles slowly gets the attention it deserves in crosscultural research. We see several promising directions in which research on cross-cultural
response styles can advance our insight in cross-cultural differences on self-report measures.
First, Lalwani et al. (2006) distin-guish two distinct ways of socially desirable responding in
individualistic and collectivis-tic cultures rather than focus on the frequency only. Social
desirability has been viewed as both other-deception (i.e. conscious lying or position supposemy
retatea to narcissism. in contrast, persons in more horizontally collec-tivist societies, which focus
on sociability and maintaining good relationships, show more impression management and put
forward a favorable self-image.
Second, work has been done on the effects of questionnaire characteristics in relation to
cross-cultural differences. For example, Van Dijk et al. (submitted) tested whether the domain
(i.e. subject) of a questionnaire has an impact on the extent of acquiescence or extremity
responding in a multicountry International Social Survey Program dataset. They found some
evidence for higher acqui-escence and extreme responding scores in domains with high personal
relevance, such as family, than in domains with low personal relevance, such as government.
Furthermore, culture-level correlations with socioeco-nomic indices and values indicated that
wealthier countries have lower acquiescence scores, and individualist, happy countries and
countries with low uncertainty avoidance have lower acquiescence scores for personally relevant
domains.
Extremity responding turned out to be higher in uncertainty avoidant countries (in
concordance with De Jong et al., 2007, and Harzing, 2006), and countries with low subjective
wellbeing, independent of the domain of the questions. A third development can be found in multilevel studies focusing on the interaction of individual-level and culture-level variables. Smith
and Fischer (2008) performed a multi-level analysis of acquiescence, extreme responding, and
culture. They demonstrated interactions in response styles between indi-viduals' predispositions
and cultural context in such a way that interdependent persons show more acouiescence in more
collectivis-tic cultures and independent persons respond more extremely in more individualistic
cul-tures. They argue that cultural context acts as a 'press' towards different communication
styles, but these effects can be reinforced or mitigated by personal characteristics. Another
example of a multilevel approach that includes interaction effects was pro-vided by De Jong
(2006). He included individual-level sociodemographic and per-sonality variables as well as
country-level value dimensions in a study on social desir-able responding across 25 countries and
more than 12,000 respondents. Main effects were found for individualism (negatively related to
social desirability) and uncertainty avoidance, power distance, and masculinity (positive
correlations). In addition, the effect of individual-level conscientiousness was stronger in
countries high on uncertainty avoidance, the effect of agreeableness was mitigated by countrylevel individualism, and the effect of extraversion was stronger in more masculine countries and
countries high on power distance.
Conclusions
Most research on response styles, across cultures, finds systematic and not just random
variation, indicating a substantive basis of response styles at the country level. Differences can
be interpreted in terms of socioeconomic context as well as values.