This website uses cookies to ensure site visitors get the best experience on our website. By continuing to use this site, you accept our use of cookies and Privacy Statement. To find out more, please visit Southern University's Privacy Statement.

I agree

Current issue

College of Business E-Journal 

Fall 2022 Volume XVII, Issue III

ISSN  number 2158-303X


I. Media Influence through What Is Reported and How It Is Reported


Living in the digital age certainly has its advantages but presents challenges to the basic principles of integrity, honesty, and truthfulness.  Media bias can be detected by reviewing the list of trending and widely discussed topics shared by media outlets. The news presented can often be identified as leaning towards one of the three political spectrums: right-wing, left-wing, or center. The infiltration of political bias within the media influences public opinion. 



Dr. Tammy Johnston

Professor of Economics
University of Louisiana at Monroe



Dr. Veronika Humphries

Assistant Professpr of Business
University of Louisiana at Monroe



Mr. Rabi Tiwari

Graduate Student, MBA Candidate
Univerisity of Louisiana at Monroe


II. Weak Identi cation in Nonlinear Econometric Models


Prof. Adkins discusses the important practical problem of collinearity when estimating a nonlinear model. Nonlinear models require an iterative algorithm to either minimize an objective function, such as the least squares criterion, or maximize an objective function such as the log-likelihood function. Multicollinearity is well understood in linear regression models. In this paper Prof. Atkins shows that commonly used metrics for collinearity in linear models can also be applied to nonlinear models. In Newton-Raphson type algorithms for nonlinear models the gradient matrix, the derivatives of the objective function with respect to the parameter values, plays the role of the matrix X of explanatory variables in the linear regression model. Prof. Adkins shows how to use the open-source software Gretl and its built-in collinearity diagnostics to examine problems that are nonlinear in the parameters. Collinearity in nonlinear models not only affects the estimator variance but also can cause problems in the algorithm converging to a local minimum or maximum of the objective function. The diagnostics Prof. Adkins uses are not only available in Gretl, but also in other commonly used statistical and econometric software packages. Thus, this paper is an important contribution to investigators estimating such common models as probit, ordered probit, and the like.



Lee C. Adkins, PhD

Oklahoma State University