| Abstrakt: |
Abstract: Many empirical researchers yearn for an econometric model that better explains their data. Yet these researchers rarely pursue this objective for fear of the statistical complexities involved in specifying that model. This book is intended to alleviate those anxieties by providing a practical methodology that anyone familiar with regression analysis can employ--a methodology that will yield a model that is both more informative and is a better representation of the data. Most empirical researchers have been taught in their undergraduate econometrics courses about statistical misspecification testing and respecification. But the impact these techniques can have on the inference that is drawn from their results is often overlooked. In academia, students are typically expected to explore their research hypotheses within the context of theoretical model specification while ignoring the underlying statistics. Company executives and managers, by contrast, seek results that are immediately comprehensible and applicable, while remaining indifferent to the underlying properties and econometric calculations that lead to these results. This book outlines simple, practical procedures that can be used to specify a better model; that is to say, a model that better explains the data. Such procedures employ the use of purely statistical techniques performed upon a publicly available data set, which allows readers to follow along at every stage of the procedure. Using the econometric software Stata (though most other statistical software packages can be used as well), this book shows how to test for model misspecification, and how to respecify these models in a practical way that not only enhances the inference drawn from the results, but adds a level of robustness that can increase the confidence a researcher has in the output that has been generated. By following this procedure, researchers will be led to a better, more finely tuned empirical model that yields better results. |
