Strategy research and panel data: Evidence and implications

A number of studies in strategic management rely on panel (longitudinal) data to test theory. The advantages of panel data notwithstanding, such data introduce analytic problems (e.g., autocorrelation, heteroskedasticity, contemporaneous correlation) that make traditional estimators (e.g., ordinary least squares) inappropriate. This study highlights the influence of contemporaneous correlation, a statistical problem that affects the analysis of panel data. Using Monte Carlo simulations, the authors find that contemporaneous correlation is particularly problematic when analyzing data sets typically used in strategic management research. They suggest straightforward techniques to mitigate the harmful effects of contemporaneous correlation.