Resumen

ANDA PADILLA, Humberto de  y  LOPEZ OLMOS, Rafael. A mathematical model proposal for prediction of academic performance. Nova scientia [online]. 2012, vol.4, n.7, pp.55-84. ISSN 2007-0705.

A mathematical model is proposed for predicting academic performance of undergraduate freshmen at the Universidad De La Salle Bajio measured as GPA the obtained. The model uses as predictors the high school GPA and the individual scales of the two psychometric tests used in the admission test: 16PF and Terman whose values are known at the time of admission. The study was conducted with the data of 3,233 students who entered in 2007 and 2008 and completed their first year of studies. Stepwise regression using MINITAB was used as the modeling technique resulting in 67 multilinear equations, each containing between two and six predictors. Between three and four prediction equations can be applied to each candidate. In addition to a general equation, there is one according to his/her career, another one for his/her area of knowledge and in 57% of cases there is an equation for his/her school of origin. Every equation provides an estimated value of his/her fhesman GPA. After appling all of them, it is suggested to use the one which has the highest correlation. A computer application was developed to show the estimated GPA by the model for each candidate. The predictors most used in the equations were the "high school GPA" and the scales of "Emotional Stability" (16PF), deduction (Terman) and "analysis and synthesis" (Terman). The prediction equations have adjusted coefficients for Pearson correlation ranging from 0.52 to 0.92. The general equation has a correlation of 0.61.

Palabras llave : Predictive Validity; Admission Test; Academic Achievement; Student Performance; Undergraduate Entrance Examinations; Mathematical Model.