Example 9.10 (page 280) used a data set on house sales to regress y = selling price of home (in dollars) to x

= size of house (in square feet). The prediction equation was ˆy = −50,926 + 126.6x. Now, we regard size of house as x1 and also consider x2 = whether the house is new (yes or no). The prediction equation relating ˆy to x1 has slope 161 for new homes and 109 for older homes. This gives evidence:

(a) of interaction between x1 and x2 in their effects on y.

(b) of a spurious association between selling price and size.

(c) of a chain relationship, whereby whether new affects size which affects selling price.

(d) that size of house does not have a causal effect on price.