Journal of Business and Economics, ISSN 2155-7950, USA July 2013, Volume 4, No. 7, pp. 620-633  Academic Star Publishing Company, 2013 http://www.academicstar.us 620 Why Logistic Regression Analyses Are More Reliable Than Multiple Regression Analyses Jianing Fang (Marist College, Poughkeepsie, New York, USA) Abstract: Medical scientists have been using logistic regression analyses for their empirical research for quite some time. However, a majority of the previous accounting or finance-related empirical studies were based entirely on results from certain forms of multivariate regressions analysis. The reliability of these findings is subject to question. By definition, all forms of multiple regressions rely critically on some assumptions of the quality of the test data. One of the major problems is that most of the financial data often violate some, and in many cases, all of these assumptions. This paper will discuss the advantages and disadvantages of both the multiple regression analyses and the logistic regression analyses for empirical research. The main goal of this article is to promote the utilization of logistic regressions in addition to any applicable multivariate analysis to provide the much needed verification and reliability of empirical study results. Key words: multiple regression; logistic regression; efficient market hypothesis JEL code: C35 1. Introduction For a long time, researchers have utilized various forms of multivariate analyses as a statistical tool in most accounting or finance-related empirical studies. Since these studies are based solely on the results of some form of multiple regressions, the reliability of their findings is subject to question. By definition, all forms of multiple regressions rely critically on the assumptions of linearity, constant variance, absence of special causes, normality, and independence of the test data. The problem is that most of the financial data often violate some, and in many cases, all of these assumptions. In the medical field, researchers have utilized logistic regressions to analyze their test data for decades. Medical doctors are dealing with life and death. We accountants and financial analysts are dealing with companies’ or people’s wealth or livelihood—not equally important, but important enough. So, if logistic regression analysis is reliable enough for medical scientists, this statistics tool must be safe enough for counting beans too. Therefore, taking guidance from medical researchers, the main goal of this study is to promote the utilization of logistic regressions, in addition to any applicable multiple regression analysis, to provide the much needed reliability of empirical study results. This study will use an example with actual market closing indices on which the author conducted dual tests on the same set of test data by running both the multivariate regressions and logistic regressions. While the multiple regression results provide the necessary statistics as well as reference or Jianing Fang, DPS, Associate Professor of Accounting, Marist College; research areas: XBRL, IFRS, tax, fair trade, research methods. E-mail: jfang@ambras.com and jianing.fang@marist.edu.