How Not to Lie with Statistics: Avoiding Common Mistakes in Quantitative Political Science
Year: 1986
Type: article
Abstract: This article identifies a set of serious theoretical mistakes appearing with troublingly high frequency throughout the quantitative political science literature. These mistakes are all based on faulty statistical theory or on erroneous statistical analysis. Through algebraic and interpretive proofs, some of the most commonly made mistakes are explicated and illustrated. The theoretical problem underlying each is highlighted, and suggested solutions are provided throughout. It is argued that closer attention to these problems and solutions will result in more reliable quantitative analyses and more useful theoretical contributions. One of the most glaring problems with much quantitative political science is its uneven sophistication and quality. Mistakes are often made but rarely noticed. In journal submissions, conference presentations, and student papers, problems occur with even more frequency. Having observed this situation for a few years, I noticed several patterns. First, the same mistakes are being made or "invented " over and over. Second, to refer a substantively orientated political scientist to an article in Econometrica, The Journal of the American Statistical Association, or even Political Methodology is to give advice that either is not helpful or is not followed. These problems are more than technical flaws; they often represent important theoretical and conceptual misunderstandings. ' However, in most cases, there are relatively simple solutions that can reduce or eliminate bias and other statistical problems, improve conceptualization, make the analysis easier to interpret, and make the results more general. In order to address these concerns, this paper presents proofs and illustrations of some of the most common statistical mistakes in the political science literature, along with theoretical arguments and suggested *An earlier version of this article first appeared at the annual Political Science Methodology
DOI: 10.2307/2111095
Author: Gary King
Institution: Harvard University
Language: English
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W2103958867
