It is well known that ranking of law schools in the US influences starting salary of students. Suppose you want to estimate the effect of ranking on median starting salary. The below output reports the results of a simple OLS regression analysis. In this model, the variable school rank has been divided into four dummy variables (with a fifth variable held out of the model as a reference category): (1) top10: i.e., whether the school is ranked in the top 10 (1=yes, 0=no), (2) r11-25: i.e., ranked in the top 11-25 (1=yes, 0=no), (3)r26-40: i.e., ranked in the top 26-40 (1=yes, 0=no) and 4) r41-60: i.e., ranked in the top 41-60 (1=yes, 0=no). The base/reference category in the regression model is >60 for schools above the 60th rank (1=yes, 0=no). Also, given that median salary is almost always positively skewed (right skewed), the dependent variable above is measured as the natural logof starting median salary (lsalary).

Exhibit 5 . reg Isalary top10 rl1 25 r26 40 r41 60 Source | SS df MS Number of obs = 148 F( 4, 143) = 222.10 Model | 9. 69572381 4 2. 42393095 Prob > F = 0. 0000 Residual | 1. 56068434 143 . 010913876 R-squared = 0 . 8614 -+- Adj R-squared = 0. 8575 Total | 11 . 2564081 147 . 076574205 Root MSE = . 10447 1salary | Coef. Std. Err. t P>Itl [95% Conf. Interval] top10 | . 7567631 . 034767 21 .77 0 . 000 . 6880395 . 8254867 rl1 25 | . 603897 . 0282749 21.36 0 . 000 . 5480062 . 6597878 r26 40 | . 3659125 . 0320444 11 . 42 0 . 000 . 3025706 . 4292544 r41 60 . 2447043 . 0275562 8. 88 0 . 000 . 1902341 . 2991745 cons 10 . 35458 . 010833 955 . 84 0 . 000 10.33317 10. 37599