by Chris Atkins

For all my career, and a bit more besides, I have been on what many Mongolfiere Weekly readers would consider the wrong side of C.P. Snow’s two cultures divide. As I drift into middle age and mellow into dabbling in social science, I start to wonder what, if anything, science has to say about culture. It turns out quite a bit more than I imagined but, unfortunately, the discovery brings out the committed analyst in me once more.

A Dutch researcher, Geert Hofstede, obtained and analyzed questionnaire-based data for 116,000 employees of IBM from more than 70 countries, representing one of the largest organizational studies ever conducted. His research resulted in his initially identifying four cultural dimensions in his sample. The first of these is called “power distance,” which is the extent to which members of organizations accept that power is distributed unequally – I’m reminded of Ronnie Corbett’s “I know my place” in the class sketch on the Frost Report from the mid-1960s in the UK. (The sketch can be found online.) The second dimension is “uncertainty avoidance,” which is the extent to which people are threatened by ambiguity and/or have a need for security. The third is “individualism,” as opposed to collectivism, which is an aspect that was already reasonably well understood by earlier work on cultural (and political) analysis. The final one of Hofstede’s original four dimensions is the, to me, rather inappropriately named “masculinity,” representing a culture that is measured by success, money and ownership of material goods.

Hofstede later introduced “long term orientation,” representing the short-term versus longer-term orientations of cultures, the latter involving a focus on future rewards. Finally, he introduced the rather better named “indulgence,” which is more or less an indication of happiness. Both long term orientation and indulgence remain less well known compared with the other four dimensions.

On learning the conclusions from this extensive and, no doubt, rich data set, the “wrong side of the cultures” scientist in me immediately started to ask a number of questions. First, is a sample taken exclusively from the employees of a single company representative? Hofstede anticipated this issue and argued that, in fact, given IBM’s notorious hiring principles (if you’re wearing a blue suit, white shirt and red tie, you’re hired – there’s a reason it’s called “Big Blue”), nationality and, therefore, culture was about the only difference between the respondents to the questionnaires. I am reasonably convinced that this is the case, although I would like to see the work replicated in a more randomly selected sample. Second, do any of the dimensions “autocorrelate,” that is, if one plots a dimension against another for the sample set, do the points lie on a more-or-less straight line, making one or other of the dimensions redundant? The answer to this is less clear-cut, being it depends on which pair of dimensions you select.

In investigating this, Hofstede and other researchers in the field certainly indulged in one of my favourite “data tricks,” which is plotting various findings against each other to see if anything shows up and, if so, if it can be understood. In many circumstances you find one of two situations: either there is the suggestion of autocorrelation or the data points are scattered more evenly with no discernible trend line, but often with somewhat easily identified clusters of like-minded points. In this case the clusters are what you might expect for cultural analysis: the “Anglo” cultures (Great Britain, USA, Australia, Canada, New Zealand) and the “Asian” cultures (China, Hong Kong, Singapore, Taiwan) tend to cluster together, thus reinforcing the sweeping generalizations that so many of us seem to carry about as mental baggage.

Hofstede’s work has been extended, and expanded upon, by others and his findings remain valid. In particular, “GLOBE” – a multi-participant study featuring about 160 researchers – confirmed the original axes of Hofstede’s findings and introduced other useful cultural dimensions.

Is Hofstede’s approach useful? The answer seems to be that it is. For example, organizations in low power-distance cultures tend to be decentralized, have “flatter” organizational hierarchies and have fewer supervisory staff, and the “worker bees” tend to be more highly qualified than in higher power-distance cultures. For me, the greatest strength of this approach, my data niggles aside, is that it turns out that it is possible to apply scientific-like analytical methodology to questions that are pretty clearly outside the normal scientific arena. It has to be said, however, that Hofstede was lucky in that he was an employee of IBM and was thus able to access the original data set in order to kick-start his analysis.

Photo: dirkb86 – dirks LEGO globe.04