Random (classical) measurement error in a regressor (X) tends to result in... O the estimated slope being biased toward zero. O the estimated slope being too large. O the estimated slope always being too negative. O the estimated slope being unbiased. In a study with a sample of actors who have appeared in films in the last ten years, a research team showed the faces of the actors to many individuals and then constructed an average "handsome/beauty" score for each actor and actress. They then determined the typical salary of each actor. Using that sample, they estimated a regression with salary as the outcome and "handsome/beauty" score as the explanatory variable. To their surprise, they found that the coefficient was small and not statistically different from zero. They reported their findings to acting schools and fine arts departments, telling them encourage all students to become actors since it did not matter much what they looked like. The questionable internal validity issue in this regression analysis is: O Sample selection O Measurement error O Simultaneous equations O Misspecification of functional form A simple regression of earnings on a binary (dummy) variable for smoking marijuana obtains a negative coefficient on the dummy. If marijuana smokers (the dummy equals one) tend to be less educated than nonsmokers (dummy equals zero), then adding a regressor for education to this regression should make the coefficient on the marijuana dummy variable: O smaller in absolute value O larger in absolute value O unchanged O the coefficient will no longer be well-defined Consider a regression with test score (TS) as the outcome variable, and student-teacher ratio (STR) as the explanatory variable, with school districts as the unit of observation. The estimated coefficient of the single variable regression is -4.5. Someone suggests to you that there might be omitted variable bias, because many rural school districts have small class sizes and also many residents of rural school districts are poor and recent migrants, and so the schoolchildren do not perform well on standardized tests in English. You find data on the percent of the population rura in each school district, and you run the new regression. Relative to the original estimate of the coefficient on STR, of -4.5, you expect that the new estimated coefficient on STR would be more like. (choose the best among the possible answers).. O -2.3 O -.6.3 O unchanged O +4.3