Abstract

Econometric models, especially statistical regression analysis, are among the most useful techniques available to peace scientists for empirical research. These methods are multi-purpose in that they can be used to test theories and hypotheses about war and peace from empirical data. They can also be used for forecasting purposes and, in certain circumstances, they can be used to evaluate alternative policies. There is a rigorous and extensive statistical theory underlying the use of these methods. The credibility of much social science and peace science research depends heavily on the proper use of these methods.

This paper explores some of the more important assumptions underlying the use of ordinary least squares, the most commonly used regression technique, and the significant consequences of not satisfying these statistical assumptions when fitting a model to empirical data. Classical knowledge about these consequences is summarized succinctly in a few tables. Additional, less well-known developments about other assumption violations related to, for example, stationarity and statistical levels of significance are introduced. Finally, suggestions are offered to help provide readers with more information about the actual research process in order to assist them in evaluating the credibility of reported econometric results.

Recommended Citation

Saltzman, Sidney (2002) "Empirical Econometrics in Peace Research," Peace Economics, Peace Science and Public Policy: Vol. 8 : Iss. 3, Article 1.
DOI: 10.2202/1554-8597.1055
Available at: http://www.bepress.com/peps/vol8/iss3/1