Statistical power in international business research: Study levels and data types

Research efforts in statistical power can improve research design and reporting. This paper examines the impact of study levels and data types on the statistical power of published international business (IB) research. To gauge the statistical power of IB research, I surveyed all empirical articles published in International Business Review and Journal of World Business from 2003 to 2008. The results support the notion that statistical power is influenced by study levels and data types, particularly panel vs. non-panel data. Although statistical power in IB research has been considerably improved, 70% of studies are still not sufficiently powered to detect small effects. The problem of lacking statistical power is more serious for country-level and multilevel studies. Studies using primary and non-panel data are less powered than those using secondary and panel data. Three strategies are offered for achieving sufficient statistical power.