Shaw (1999) uses a regression model to study the effect of TV ads and candidate appearances on votes in the presidential elections of 1988, 1992, and 1996. With three elections and 51 states (DC counts for this purpose), there are 153 data points, i.e., pairs of years and states. Each variable in the model is determined at all 153 points. In a given year and state, the volume TV of television ads is measured in 100s of GRPs

(gross rating points). Rep.TV, for example, is the volume of TV ads placed by the Republicans. AP is the number of campaign appearances by a presidential candidate. UN is the percent undecided according to tracking polls. PE is Perot’s support, also from tracking polls. (Ross Perot was a maverick candidate.) RS is the historical average Republican share of the vote. There is a dummy variable D1992, which is 1 in 1992 and 0 in the other years. There is another dummy D1996 for 1996.

A regression equation is fitted by OLS, and the Republican share of the vote is

− 0.326 − 2.324×D1992 − 5.001×D1996

+ 0.430×(Rep. TV − Dem. TV ) + 0.766×(Rep. AP − Dem. AP )

+ 0.066×(Rep. TV − Dem. TV )×(Rep. AP − Dem. AP )

+ 0.032×(Rep. TV − Dem. TV )×UN + 0.089×(Rep. AP − Dem. AP )×UN

+ 0.006×(Rep. TV − Dem. TV )×RS + 0.017×(Rep. AP − Dem. AP )×RS

+ 0.009×UN + 0.002×PE + 0.014×RS + error.

(a) What are dummy variables, and why might D1992 be included in the equation?

(b) According to the model, if the Republicans buy another 500 GRPs in a state, other things being equal, will that increase their share of the vote in that state by 0.430×5 .
= 2.2 percentage points?
Answer yes or no, and discuss briefly. (The 0.430 is the coefficient of Rep. TV − Dem. TV in the second line of the equation.)