“…We also include other demographic variables that could affect the production function; examples include gender and number of total credit hours earned. Since letter grade is discrete and ordinal and we assume the error term is normally distributed; an ordered probit model is chosen for the analysis (Spector and Mazzeo, 1980; Becker, 1983; Borg, Mason, and Shapiro, 1989; Borg and Shapiro, 1996; Borooah, 2002; Greene, 2003; Raehsler et al, 2012). The ordered probit differs from the ordered logit in its assumption of the error term being normally distributed as opposed to logistically distributed; however, as Greene (2003) points out, “it is difficult to justify the choice of one distribution over the other on theoretical grounds when in most cases it seems to not make much of a difference.” The ordered probit model takes the form: where y * is the unobserved latent dependent variable, (coded as 0, 1, 2, and 3) for letter grades of D, C, B, and A, respectively.…”