/* Compare t-test results to dummy regression results  */
ttest realrinc if year==2006, by(white)
reg realrinc white if year==2006

/* What about if we reverse the dummy variable coding? */
reg realrinc black if year==2006