41 Pages Posted: 9 Aug 2019
Date Written: August 6, 2019
We investigate whether outliers in cross-country samples and the common methods we use to address them affect the trustworthiness of our empirical results. Our analysis begins by documenting recent international business (IB) research practices in the identification and treatment of outliers. We then explore the bias and error from using sample-wide univariate outlier detection and treatment methods in studies that use cross-country datasets. Additional analysis, using cross-country data on profitability, executive compensation, and R&D, demonstrates the magnitude of the bias and error from common outlier approaches in IB research. We propose outlier mitigation strategies at both the sample and country levels.
Keywords: Outliers in International Data, Winsorizing, Trimming, Multivariate Identification
JEL Classification: C15, C18, F3, G30
Suggested Citation: Suggested Citation