As in class, suppose the true underlying model relating tests scores T and class size C is T; = a + BC; + yX; + 6;, where X is unobserved and thus omitted from the sparser model that is actually estimated: T; = a + bC; + uj. Let X represent the proportion of time that a student is non-disruptive in class, as in Lazear (2001). What is the sign of the correlation between C and X (the main finding of Lazear 2001), and, assuming that X is positively correlated with test scores, how is the estimate of b biased? O A. C and X are negatively correlated; the estimate of b is biased downward O B. C and X are positively correlated; the estimate of b is biased upward O C. C and X are negatively correlated; the estimate of b is biased upward O D. C and X are positively correlated; the estimate of b is biased downward