Measuring organizational cultures: a qualitative and quantitative study across twenty cases

Administrative Science Quarterly, June, 1990 by Geert Hofstede, Bram Neuijen, Denise Daval Ohayv, Geert Sanders

Previous Research on National Cultures

Our research project into organizational cultures was modelled after an earlier project by the first author that covered differences among national cultures (Hofstede, 1980, 1983a, 1983b, 1983c, 1983d). That study used an existing data bank from a large multinational business corporation (IBM), covering matched populations of employees in national subsidiaries in 64 countries. The data consisted of answers to questionnaires about employee values and perceptions of the work situation that were collected in the context of two worldwide rounds of employee attitude surveys. Their use for studying differences in national cultures was an unintended, serendipitous by-product, for which the corporation opened its files of 116,000 survey questionnaires collected between 1967 and 1973. Twenty different language versions were used. Initially, from the 72 different national subsidiaries for which data were available, only the 40 largest were selected for the analysis (Hofstede, 1980). Subsequent follow-up research showed data from another 24 subsidiaries to be usable, 10 as separate countries and 14 grouped into three historical/geographical regions (Arab-speaking countries, West Africa, and East Africa), thus raising the total number of units in the analysis to 53. In the remaining eight countries the number of native respondents was insufficient to allow statistical use of their data (Hofstede, 1983a).

The questions in the IBM surveys had been composed from initial in-depth interviews with employees in ten countries and from suggestions by frequent travellers in the international headquarters' staffs who reported on value differences they had noticed among subsidiaries. The surveys had been managed by an international team of social scientists (both from inside and outside the corporation) who were participant observers or observing participants in the daily life of one or more of the subsidiaries. During the years devoted to the analysis of the data, the first author and his family lived and worked in four different countries. This background provided a qualitative context to the cross-national study. The possibilities for quantitative analysis of the precoded answer scores were excellent. National idiosyncrasies and nuances of questionnaire translation weigh heavily in a two-, three-, or four-country study, but with the unusually large number of 40 or 53 countries and regions, national patterns start to show a global structure, which the "noise" of the idiosyncrasies of individual countries cannot suppress. The structure revealed by the IBM data consisted of four largely independent dimensions of differences among national value systems. These were labelled "power distance" (large vs. small), "uncertainty avoidance" (strong vs. weak), "individualism" vs. "collectivism," and "masculinity" vs. "femininity." All 53 countries and regions could be scored on all four dimensions; the four together accounted for 49 percent of the variance in country mean scores on 32 values and perceptions questions.