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 generalisation he has repeatedly used to illustrate that capability. When tested against cross-sectional and longitudinal empirical data about conflicts in industrial relations, the generalisation is shown not to have predictive power. A second generalisation is then tested, which also fails to demonstrate predictive capability. This paper discusses some characteristics of valid cross-national research in the light of these predictive failures.

No abstract