Why linear probability model (LPM) is not a good modeling approach for binary response models?

a)

Linear regression model assumes that the statistical experiment draws from a normal distribution, whereas statistical experiment for binary response model draws from a half-normal distribution.

b)

Linear regression model assumes that the statistical experiment draws from a normal distribution, whereas statistical experiment for binary response model draws from a Poisson distribution.

c)

With LPM, predicted probabilities may be outside the [-1, 1] interval.

d)

None of the above.