Question: Kahneman and Tversky (1973) postulate that the concept of representativeness explains why people are prone to underestimate regression to the mean. For example, in a regression of the height of a son on the height of a father, the estimated slope coefficient tends to be positive but less than one. Nevertheless, people are prone to predict that sons will be as tall as their fathers, in effect treating the slope coefficient as one. Explain how representativeness might account for this prediction error and then extend the analysis to explain how representativeness might underlie the De Bondt-Thaler winner-loser effect.