Do your Findings Depend on your Data(base)? A Comparative Analysis and Replication Study Using the Three Most Widely Used Databases in International Business Research

Abstract

Theoretical and empirical advances in international business (IB) depend heavily on archival data obtained from databases. As there has been scant research on the implications of database choice in IB research, we analyzed data from the three most-widely used databases in the field: Compustat Global, Osiris and Worldscope. We examined data coverage across several geographic regions and countries, analyzed descriptive statistics and regression results, and replicated a study by O'Brien (2003), who theorized and found a negative firm performance effect of innovation strategy matched with high levels of debt governance. Based on our empirical results, we found the presence of what we call a “database effect” – researchers likely would come to a different conclusion based on the database used – particularly for developing country results. The database effect is confirmed using multiple estimation techniques and also in our replication study. Our replication study also found evidence that O'Brien's (2003) findings do indeed apply to countries outside of the United States. However, in China, we found statistically significant contrasting results in two databases, requiring a need for more theoretical reflection of firm governance of innovation in different institutional contexts.