Claiming too much, delivering too little: testing some of Hofstede’s generalisations

This paper considers Geert Hofstede’s claim that his national cultural ‘dimension scores’ and related rankings of countries enable effective prediction by examining a causal generalization he has repeatedly used to illustrate that capability. When tested against cross-sectional and longitudinal empirical data about industrial relations conflicts the generalization is shown not to have predictive power. A second generalization is then tested. It too fails to demonstrate predictive capability. The paper then discusses some characteristics of valid cross-national research in the light of these predictive failures.