-Figure One & Table Two here-
III. Model
I treat the advertising budgets of candidates and parties as a function of four
different sets of factors: the structure of the media market in a congressional district, the
amount of money available to the candidate, the actions of other advertisers, and the
nature of the race. I estimate a linear model, since there is little reason to suspect ceiling
effects. There is plenty of anecdotal evidence that campaigns believe there is no such
thing as too much TV. Indeed, the only reasonable stopping point, beyond financial
constraints, is the availability of time slots to purchase (e.g. Magleby 2002).
The salient aspects of market structure for political advertisers have been
measured in different ways. At the bottom is the matter of cost. Almost any effort to
reach all the voters in a particular House districts will result in communicating with
significant numbers of citizens in neighboring districts. This is true both when the
dominant market in an area is huge, such as New York City stations which reach over 7
million households in four different states, and even when the dominant market is small.
In the latter case, the candidates’ effort to speak to everyone will lead them advertise in
surrounding markets. The end result will certainly be less inefficient or wasteful than
advertising in New York City, but what Campbell, Alford, and Henry (1984) call the
“lack of congruence” between market and district boundaries will result in plenty of
waste. Thus authors try to gauge both the underlying cost of communicating with all
voters, and the fragmentation of the market.
9
Convention