|
January 1989
Vol. 2, No. 3
Analysis
TV Advertising and Minor Party Campaigns
R.W. Bradford is publisher of Liberty.
Commercial enterprises have long known that the most efficient (i.e. lowest cost) way to get an advertising message to a broad range of Americans is to use television advertising. In recent years, the people who run the campaigns of America’s major political parties have learned this lesson: nowadays, the majority of the funds spent for political campaigning is spent on television advertising.
Because of the huge numbers of people who watch television, television is not as efficient in delivering messages that are targeted to specific segments of the population. If you want to reach motorcyclists, for example, it is more efficient to buy advertisements in motorcycle magazines, or to mail advertising directly to individuals who own motorcycles. Since the major parties try to appeal to all segments of the population — more precisely, to all voters — television works very well for them.
But the question of whether television works for minor parties, whose appeal may be limited to a smaller segment of the population, has remained unsettled. Only two minor parties have used television advertising in recent history: the Libertarian Party and the Lyndon LaRouche party in its various incarnations. The LaRouche parties, however, have never achieved widespread ballot status, and their appeal has been very limited. In addition, the overwhelming bulk of their television advertising has been in the form of 30 minute programs, usually taking the form of lectures by LaRouche. This form of political advertising is not used by the major parties, who learned long ago that most television viewers (and practically all uncommitted voters) either change channels or turn off their television when confronted with a half-hour political program.
The Libertarian Party’s use of national television advertising in its 1980 campaign is widely credited with its relatively high vote total that year. (The 1980 LP campaign yielded more than five times its 1976 total, about four times its 1984 total, and more than twice its 1988 total.) However, the extent to which the 1980 vote total was the result of television advertising rather than other factors remains conjectural.
Ron Paul, the LP nominee in 1988, decided against buying national television advertising. Instead, his campaign produced a single 30 second commercial that stressed Paul’s hostility toward the IRS, and made it available to local LP groups for use on local television stations. In making this decision, the LP presidential campaign created an opportunity to arrive at a reasonably accurate evaluation of the impact of television on voting.
Some markets were exposed to Ron Paul television advertising while others were not. By comparing the vote totals in areas with television spots to vote totals in similar areas without television advertising, we can arrive at answers to the following questions:
How much, if any, impact on vote totals did television advertising on behalf of the LP ticket have in the 1988 election?
How efficient is television advertising? How much does it cost to obtain a vote by advertising on television?
To answer these questions, we must isolate two areas of the country that are reasonably similar in their past voting habits, in only one of which was television advertising used.
The vote totals from Kansas offer an excellent opportunity. There are 7 different television market areas that are all or partly in Kansas. In two of these, including approximately 58% of the state’s voters, local Libertarian Party groups purchased television advertising. No television advertising was used in the remaining markets.
In addition, the variation in other campaign activities between the two areas was minimal, and such other campaign activities can be isolated and taken into account.
Around population centers, party members were fairly active, but in most rural areas, the only impact of the campaign was via television. Both the advertised area and the non-advertised area consist of a mixture of rural and urban areas.
In addition, the two markets which used television advertising chose different approaches: one used the anti-IRS spot provided by the Paul campaign; the other a more generic libertarian ad produced by the Utah LP.
By dividing the state into three areas corresponding to the amount and type of advertising seen in each area, it is possible to get a good idea of the impact of advertising and the relative merits of the two advertising approaches.
The Data
The following analysis is based on a comparison of what happened in each of the three areas, which are determined (with one minor exception) as the television market areas as defined by TV & Cable Factbook. With a single exception, every county in the state is included. The excluded county had substantial variables present that made it anomalous with each defined area, as I explain below.
Wichita area: The Wichita market includes 63 of Kansas 105 counties, whose citizens cast 438,729 (44.2% of the state total) votes in the presidential election. In addition to metropolitan Wichita, it serves most of the rural hinterland of western Kansas through a series of repeater transmitters.
Topeka area: The Topeka market area includes 12 counties, whose citizens cast 141,213 votes (14.2% of the state total) for president. Topeka, with 115,266 residents, is the only large city in the market area. Even so, this is the most urban of the areas considered, with 34% of its population living in Topeka, and a total of 51% living in cities of 25,000 or more.
No television area: The remainder of the state lies within television markets centered in other states. The largest by far is the Kansas City (MO) market, which includes 14 Kansas counties (315,626 votes, 32.9% of state total). Several substantial suburbs of Kansas City lie within this market area. The Joplin (MO) market includes eight counties (58,534 votes, 5.9% of state total). The Lincoln (NB) market includes six counties (20,534 votes, 2.1% of state total) along the Nebraska border. The Tulsa (OK) market includes two counties (16,562 votes, 1.7% of state total). The Amarillo (TX) market includes one county (1,667 votes, 0.2% of state total).
Within these television markets there is one county where there was an extraordinary local campaign on behalf of the LP ticket. In Atchison County, LP activist and former Mayor Doug Merritt purchased a full page newspaper ad plus 27 radio spots at a cost of $600 and campaigned extensively for the ticket in his weekly newspaper column. Because these activities were so much more intense than activities elsewhere in the state, I have excluded Atchison County from the “no television market area” category. (It should be noted that Atchison is a very small county — it cast only 1.6% of the votes in the no advertising area — and including it would change the following analysis only very slightly.)
Each of these areas contains a single city of more than 100,000 people. Wichita, population 279,835, contains about 27% of its area’s population; Topeka, population 115,266, has about 34% of its area’s population; and Kansas City, population 168,213, contains about 17% of its area’s population.
The areas are predominantly rural. In the Wichita area, 35% of the population lives in cities of 40,000 or more; in the Topeka area 44% live in cities; and in the no advertising area 31% are city dwellers.
Does TV advertising win votes?
The following table summarizes the campaign activity in each area:
As we can see, substantially more money on a per capita basis was spent in the Wichita area than in the Topeka area. It should be noted that the radio advertising did not cover the entire market area: in the Wichita area, radio advertising reached about a third of the electorate; in the Topeka area, about a tenth.
The table below summarizes the voter response in each area: [table omitted]
It is apparent that the ticket was strongest by a wide margin in the Wichita area. Nearly twice as many Wichita area voters chose the LP ticket than did Topeka voters; about 2.5 times as many did as those voters not exposed to television advertising. Given the greater expenditure in the Wichita area, it is not surprising that it had the highest voter response.
It is a mistake to generalize from these data. Rather than reflecting variations in television advertising, it might reflect the disposition of voters to choose the LP. That is, it might be that voters in the Wichita area are simply more inclined to vote Libertarian than voters in other areas.
This is in fact the case: in past elections, voters in the Wichita area have tended to give a higher percentage of their votes to LP candidates than voters in other areas. Table 3 summarizes voter response by area in the 1984 election.
Note that the No Advertising area and the Topeka area had approximately the same LP vote percentage in 1984, and that the Wichita area outvoted the others by only about 50%, much less than the margin by which it outvoted them in the 1988 election.
This suggests a way to project how many voters in the Wichita and Topeka areas would have voted LP if television advertising had not been used in those markets. Since voters in the No Advertising area increased their response rate by a factor of 2.58, it is reasonable to conjecture that voters in the other areas would have increased their response rate by a similar factor.
The table below shows what the projected LP vote would have been if television advertising had not been used: [table omitted]
The difference between these projected vote totals and the actual vote totals in these markets is, presumably, the product of the television advertising in those markets: [table omitted]
It is apparent that these relatively small purchases of television spots had a dramatic impact. The schedule of 40 television spots plus 220 radio spots resulted in increasing the LP vote by more than 80%. The 12 television spots run in the Topeka market increased voter response by 37%.
All told, the areas in which television advertising was purchased voted much more strongly for Paul than if no television spots had been used. In fact, the television advertising gained approximately 3,970 votes for the LP ticket. That is an increase of 73.8%.
What does it cost?
What do results like these cost? In terms of raw dollars, the Wichita advertising program was more efficient than the Topeka program, as is evident from Table 6. [table omitted]
As you can see, it cost nearly 3 times as much to obtain additional votes in the Topeka market as in the Wichita market. In part, this may be the result of the relatively large expenditure for radio advertising in Emporia, a small city within the Topeka market, well away from population centers. It also may be the result of the variation in the ads used in each market.
But one conclusion is obvious: the marginal cost of obtaining votes by television advertising is less than the average cost of obtaining votes by the means employed by the Paul campaign, which reported it spent approximately $3 million to obtain about 420,000 votes, for an average cost of $7.14 per vote.
National television advertising is much more efficient than buying ads in local markets. That is, the cost to reach each viewer is less for national ads because of economies of scale. It is therefore likely that had the Paul campaign purchased national television spots, it would have acquired additional votes more cheaply than the $1.86 per vote average cost of the Kansas LP.
What about radio advertising?
I attempted to use the same technique to isolate the influence of the radio ads; that is, I isolated the areas where the radio ads were heard, compared the improvement from 1984 to 1988 in both the radio influenced areas and the non-radio areas, projected the vote totals for the radio areas if the radio spots had not been aired. In an effort to maintain the interest of those who have been bored or annoyed by all the tables and projections I have reproduced so far, I shall not reproduce the study step-by-step here.
The study indicated that the cost of votes gained by radio advertising ranged from -$1.05 to $18.25, depending on the definition of radio market area used. The negative vote cost occurs because when using one definition of radio market area, the projected vote total without the spots actually exceeds the actual vote total. (In other words, it showed that the radio ads actually reduced votes for the LP ticket.)
This confusing result is the product, I believe, of two factors:
1) radio market areas do not correlate nicely with the voting data (because radio markets are smaller, they do not correspond so closely to county lines); and
2) radio advertising is not nearly as effective a means of reaching voters as television advertising.
The major parties, which have studied the effect of radio and television advertising, spend only a tiny proportion of advertising funds on radio, presumably because they know that it doesn’t work as well as television advertising. And commercial advertisers have long been convinced that television advertising is the cheapest way to reach a general audience and that radio is better employed for reaching specific target audiences.
But it may be a mistake to conclude that radio advertising is ineffective, or less effective than television advertising, without further study.
Which television spot worked better?
Even when all radio advertising expenditures are excluded from the analysis, the cost per vote gained was much cheaper in Wichita where the generic ad was used than in Topeka where the IRS ad was used. This supports the thesis that the generic ad was more effective.
But in my judgment, it is not conclusive.
The difference in cost per vote received may be entirely the result of the higher cost of advertising in smaller markets, where economies of scale are not available. There are 132,100 television households in the Topeka market area, and the ads there cost an average of $40 each. Thus the average cost per thousand potential households for these ads is 30.3. The Wichita market has 413,000 households, and its ads cost an average of $100 each, so the cost per thousand potential households is 24.2.
In addition, there were 3.33 times as many spots run in the Wichita market as in the Topeka market, yet Wichita’s expenditure was only 22% higher in terms of cost per voter within each market.
A more conclusive answer could be drawn if the ratings of all the ad spots were known, so that cost per exposure of each ad to each voter could be estimated.
What does all this mean?
The experience in Kansas clearly demonstrates that the use of television advertising can substantially increase LP vote totals at a low cost. Television advertising can greatly increase voter response to minor party campaigns, just as it increases voter response to major party campaigns.
Television advertising in Kansas increased votes for the LP ticket by 3,970 votes (73.8%) at a cost of $7,265, or about $1.86 per vote.
Given the fact that Kansas is a fairly average state for the LP (in 1976, the LP got 0.34% in Kansas versus 0.35% nationally; in 1980 [when Kansan David Koch was on the ticket] the LP got 1.48% in Kansas versus 1.06% nationally; in 1984 the LP got 0.35% in Kansas versus 0.34 nationally), it is likely that the effect would be similar in a national campaign.
However, the cost of obtaining the additional votes would likely be lower in a national campaign, since national advertising reaches voters at a lower unit cost.
In addition, the data suggest that radio advertising is not so effective as television advertising and that the generic advertisement produced by the Utah LP was more effective than the IRS spot produced by the Paul campaign.
Further analysis of the relationship between election returns and specific campaign techniques would be very useful for those pursuing electoral politics.
Special thanks to Karl Peterjohn and Douglas Merritt of the Kansas Libertarian Party, who provided the data for this study.
|
|