Laserfiche WebLink
APPENDIX B: <br />FURTHER EXPLANATION OF THE <br />PROBIT REGRESSION ANALYSIS <br />Probit regression is a special type of regression analysis. While there are many <br />differences between the underlying estimation techniques used in the probit <br />regression and the standard regression analysis, the main differences from the <br />layperson's point of view lie in the nature of the dependent variable and the <br />interpretation of the coefficients associated with the independent variables. <br />The basic model looks the same: <br />DV = f(D, I, 0), <br />where DV is the dependent variable; D is a set of demographic variables; I is a <br />set of industry and occupation variables; and 0 is a set of other independent <br />variables. <br />The estimation process takes this equation and transforms it into: <br />IDV =C+(p1*D)+(p2*1)+(p3*O)+4, <br />where C is the constant term; 01, 02, and 33 are coefficients, and µ is the ran- <br />dom error term. <br />In the standard regression model, the dependent variable is continuous and <br />can take on many values. In the probit model, the dependent variable is <br />dichotomous and can take on only two values: zero or one. For instance, in the <br />standard regression analysis, we may be exploring the impact of a change in <br />some independent variable on wages. In this case, the value of one's wage <br />might be any non -negative number. In contrast, in the probit regression analy- <br />sis, the exploration might be the impact of a change in some independent vari- <br />able on the probability that some event occurs. For instance, the question <br />might be how an individual's gender impacts the probability of that person <br />forming a business. In this case, the dependent variable has two values: zero, if <br />a business is not formed; one, if a business is formed. <br />The second significant difference—the interpretation of the independent vari- <br />ables' coefficients—is fairly straight -forward in the standard regression model: <br />the unit change in the independent variable impacts the dependent variable <br />0 2020 Colette Holt & Associates, All Rights Reserved. 105 <br />