Suppose x is uniformly distributed between 0 and 100, and y is binary with log[x;/(1 - x;)] = -2.0 + 0.04x;. Randomly generate n = 25 indepen dent observations from this model. Fit the model, and find corr(y - , /). Do the same for n = 100, n = 1000, and n = 10, 000, and summarize how the correlation seems to depend on n