suppose that y is independent of explanatory variables and it takes on the values -2, -1, 0, 1, 1 with equal probability. does this violate the gauss- markov assumption