National culture and personality

Here's another report from the 2012 DOP conference.

If people of different nationalities score differently on a personality test, does this say something about national temperament, or simply that the test is biased? Prof Dave Bartram took us through an interesting approach to unknot this tricky issue: when “national differences” in personality also correlate with other measures, we can be more confident they are the real deal.

Bartram worked with a big data set - one million participants all told – but as the correlations were made between countries, not individuals, they involved just 31 cases, a modest sample in which to detect patterns. Correlating the Big 5 personality factors with the four Hofstede dimensions of national culture, he found that each personality measure correlated with one or more Hofstede dimension; for instance, Emotional Stability tended to be higher in cultures that are less masculine, more individualistic, more tolerant of ambiguity, and have less power distance (meaning less acceptance of unequally distributed power).

The next analysis was neat, correlating the cultural dimensions with the standard deviation of personality scores in each country – whether scores tightly clustered or showed large variation - rather than with their average levels. This made it possible to explore the idea that some countries are culturally “tighter” than others, giving less scope for individual difference. The analysis picked up several such effects. The higher the power distance of a culture, the more uniform its members were in terms of measures like agreeableness, conscientiousness or extroversion; the reverse was true for countries high on another measure, individualism. Even with this small data set (the 31 countries) it was possible to predict large amounts of the variance of Big 5 measures from the Hofstede scores, as much as 76% in the case of Emotional Stability.

Correlation of personality with culture ratings might not strike you as objective enough to produce a verdict; perhaps they are both subject to a common confound. But how about correlations with hard measures such as GDP, life expectancy, UNESCO education index and the UNDP human development index? These measures were all found to correlate with standard deviations of personality scores, for instance high GDP was related to larger ranges of openness to experience in the population.

This study doesn't answer whether national culture shapes typical personality or vice versa, although it's useful in honing hypotheses for investigating such matters. But this cascade of correlations does suggest that personality differences between countries, although they are small, reflect something real, rather than meaningless measurement error.