Evaluation of University's Failure Indexes using Confidence Intervals

Abstract

The objective of this paper is to evaluate the failing index of the 2014s second semester in a northern Mexican university, as well to propose a new methodology to analyze data in social sciences. This new methodology presents the use of confidence intervals with the median to detect atypical/extreme cases because it is less sensitive to biased data. Examples included showing how to calculate confidence intervals using the median. The data included all the student's grades (n = 130,452). The backbone of the data is comprised of five institutes, 16 departments and 123 degrees (63 undergraduates and 41 graduates). The results showed atypical cases at an undergraduate level (one case); departments (two cases), graduate (three cases). These results could help the decision makers to channel their efforts into diminishing the number of failing grades. Beyond the results, the new methodology could not only be applied in failing indexes but into deserting, graduating, student satisfaction, etc.