Differences in work values: understanding the role of intra- versus inter-country variation

Abstract

A growing literature emphasizes the need for studies taking a contingency perspective to international HRM to move beyond mean country differences in work values and begin considering intra-country variation (ICV). We use individual-level data on Hofstedeian values – not hitherto available – to infuse this literature with systematic quantitative evidence regarding the importance of ICV vis-à-vis inter- or between-country variation. We begin by estimating various random effects models, discovering that ICV accounts for the bulk, approximately 85%, of total variation in work values. To add a much-lacking comparative dimension and because ICV only has real-life relevance if we know its sources and can observe them, a three-level multilevel analysis provides a novel disentanglement of the importance of country relative to region and socioeconomic stratum as readily observable within-country sources of variation in values. Results show the value for practitioners and scholars of not only focusing on country differences strictly but also considering subnational categorizations when seeking to understand differences in work values. Key contribution of this paper is to take the debate on ICV out of the theoretical and into the practical realm. Implications of our findings are discussed.

Keywords::

• cross-cultural management
• cultural contingency
• Hofstede
• intra-cultural diversity
• values

Notes

^1. Taras, Rowney and Steel (2009) review 50 years of measuring culture. They find that self-report questionnaires are virtually the only tool used for quantitative culture measurement (Taras et al. 2009, p. 364). Culture frameworks differ in the post-collection processing of individuals' responses. All the key frameworks (e.g. Hofstede 1980; Schwartz 1994; Inglehart and Baker 2000; House et al. 2004) ask individuals about their values, however, with aggregated responses subjected to further post-collection processing such as principal components analysis and index construction.

^2. Measures of cultural distance, including the famous index developed by Kogut and Singh (1988), aim to capture the extent to which countries differ from each other in terms of their culture, notably their values.

^3. Gerhart and Fang (2005) report that there is more variation in values between organizations than between countries, which goes some way to address the issue of the real-life relevance of ICV. We are interested in the work values of the whole population, however, and knowing how much variation in values occurs between organizations is information that is too specific to offer much practical guidance for the general issue firms face, namely the contingency of their practices and strategies, notably HRM policies, to the values held by the general population.

^4. Over the years, Geert Hofstede (born in 1928) has added coauthors to various editions of his book Cultures and organizations: Software of the mind, first published in 1991 (Hofstede 1991). The second edition that updated the first edition was written jointly with his son (Hofstede and Hofstede 2005). The third edition, updating the second edition, added Minkov as a third author (Hofstede et al. 2010). There is a great deal of overlap between the three editions and Geert Hofstede remains the main author of all editions.

^5. To elaborate, measurement error adds as much to within-country variation as it does to total variation. In calculating the proportion of total variation that is within countries, the denominator thus increases by exactly as much as the numerator, leading to an inflation of ICV as a percentage of total variation. For calculating the proportion of total variation that is between countries, measurement error does not affect the numerator but increases the denominator, thus leading to a deflation of BCV as a percentage of total variation.

^6. Concerning the relation between socioeconomic class and values, see for example Kohn's (1969) classic work. Côté (2011) surveys the work on social class in relation to organizational behavior. We focus on only two within-country sources of values differences. The reason is that the idea of considering such sources is general, while considering additional within-country sources would lengthen the manuscript considerably without generating much additional insight. We leave it to future research to extend the analysis in the present paper and assess the importance of other easily observable subnational categorizations that could be important sources of values differences, e.g. age and gender.

^7. Many different measures of overlap exist (see Deza and Deza (2009) for a survey). Beugelsdijk, van Hoorn, Maseland, Onrust and Slangen (2012) calculate an index of percentage overlap, comparing countries' fraction of yes/no scores on 0/1 questionnaire items on values, and use this index to explain foreign sales by US multinationals.