Home > News > Negative Campaigning in Multi-Party Systems
85 views 7 min 0 Comment

Negative Campaigning in Multi-Party Systems

- April 27, 2011

As a follow up to “yesterday’s call for research”:https://themonkeycage.org/2011/04/any_research_on_attack_ads_in_.html on negative campaigning in multi-party systems (in response to “Matthew Yglesias’s”:http://yglesias.thinkprogress.org/2011/04/multi-party-systems-and-attack-ads/ post on this topic), we are pleased to bring you the following guest post from “Professor Scott Desposato”:http://swd.ucsd.edu/pubwork.html of the University of California-San Diego.

******

A great deal of research investigates the causes and consequences of negative campaigning – but nearly always in the two-candidate, SMD context of the United States. What happens in other countries, under other electoral rules, or in multi party systems?

Several of us are working on this, as mentioned in yesterday’s post. (Matt Golder, Indridi Indridason, and others) My focus is on Latin America. Starting in 2005, I began recording televised advertisements. Thanks to the hard work of some terrific research assistants, I have about 13,000 unique messages at this point from 13+ countries. Details here.

To build a theory of candidate strategy, a few issues have to be addressed. First, what are candidates trying to accomplish with attack messages? The literature on the impact of attack messages suggests that attack ads can demobilize, moblize, inform, confuse, energize, and (insert favorite political verb here). Assume for now that positive messages increase support for the target, and attack messages reduce support for the target of the message. Further assume that any voters swayed by an attack either fail to vote, or that they are distributed equally across all other candidates. Second, let’s assume that the impact of any message (positive or negative) is a random variable, where negative messages are “riskier” than positive messages: they have a higher expected impact and a larger variance in impact.

Some intuitive results come out of a game-theoretic model based on these assumptions, with empirical support from the Latin American cases:

1. Most strategic attacks happen between pairs of candidates on either side of “electoral thresholds” – pairs (A,B) where candidate A has enough support to win, and candidate B does not, BUT, reducing candidate A’s support sufficiently would lead to B’s election.

For example, under single member district rules, there is always one threshold, between the first and second-place candidates. In other words, if #2 could somehow reduce #1’s vote share, then all else equal, #2 will win. And empirically, almost all SMD attacks are between the first and second place candidates. Consistent with game theoretic results, most of the attacking is from the second-place candidate, and front runners only shoot back when the campaigns get close.

Beyond SMD, the number of thresholds where attacks can yield payoffs depend on the electoral formulas. In runoff systems, one can expect attacks between the first, second, and third place candidates, depending on standings. OLPR’s relatively complex electoral rules create lots of candidate pairs divided by thresholds, but whether these options to attack are exercised depends on the number of competitive candidates.

2. As the number of free-agent competitive candidates rises, negative campaigning fades to zero, because of free rider and information problems. If there are 100 of us competing for 20 seats, and I attack one of my competitors, all the rest of you will benefit from that attack. I’ll suffer any backlash against negative campaigning, and you’ll just continue to run your positive ads. I’ll lose and you’ll win.

There’s also an information problem when there are many candidates – how does one identify the right target? In the 2006 election for the local legislature of Brasilia, there were hundreds candidates competing. Not surprisingly, there was not a single case of one deputy candidate attacking another. Here and here are two examples of OLPR ads. In the first, the jingle criticizes corruption – but never names names. In the second, the candidate introduces himself as a (probably religious) singer and notes that his campaign code is easy to remember.

Think about competition in the market for realtors versus operating systems, or colas. Apple attacks the industry leader’s product, Windows. Pepsi goes after Coke. But the thousands and thousands of realtors only offer smiling faces and annoying slogans.

3. The same logic can be extended to strong party systems – where parties are the key agents, not individual politicians. When highly fragmented, we expect less attack campaigning because of the informational and free rider problems. But parliamentary systems can create SMD-like incentives even under pure PR – because the real payoff is forming a government, not winning seats. Where the largest party is allowed (or large enough) to form a government, the electoral environment for parties is very similar to SMD Plurality, and we expect more attacks between the two largest parties. Finland has OLPR, just like Brazil. But parties competing for control of government run attack advertisements against each other.

4. The more voting is constrained by one-dimensional ideology, the more deviations one can expect from these predictions – but the direction of deviations will depend on patterns of competition.

5. There are many exceptions, some of which fit with this simple model, and some of which do not. One laggging candidate spent all his campaign time attacking the front runner – the rumor was that he had “sold” his campaign to the second place candidate. Some legislative candidates ran negative ads – but only against candidates for executive office! I would call such messages forms of position taking, as they are not directed against competing candidates.

6. Just as Geer found in the US, early results suggest that attack messages in other countries have much more information than do positive messages. Compare two messages from the last Mexican election: AMLO and Calderon.

Much of the US literature suggests that campaign effects are relatively small. In many countries, this is not true. We need more research on comparative campaigns.