Every state wants to be “the” state that selects the next president. Every state wants the attention of the presidential candidates and wants its voters to have a say in the presidential nomination process. But since the early 1980s, presidential nomination events in more and more states have essentially been meaningless as the parties' nominations have effectively been wrapped up before citizens have cast their primary ballots or attended their local caucuses.
In response to this, states have adopted two strategies in an attempt to secure the recognition of the candidates and be relevant in the selection process: front-loading their nomination events and participating in regional primaries. Front-loading is the moving forward of a nomination event toward the early weeks of the nomination calendar and it has resulted in an increasing concentration of nomination events in those early weeks. Some have written of the potentially negative consequences of this front-loaded calendar, citing a lack of voter learning, poor quality campaigns, and less competition through the process (Mayer and Busch 2004). Despite these potential problems, front-loading continues. Currently, over 20 states have Democratic primaries or caucuses scheduled for February 5, 2008—the earliest date allowed under Democratic Party rules for all but four states. And, recently, both Michigan and Florida decided to defy those national party rules by holding their nomination events in January.
The other approach that states have taken to gain the recognition of the candidates is to join in a regional primary. The most recent attempt has been in the West, where officials in several western and Rocky Mountain states have worked toward the goal of holding presidential primaries on the same date. Their hope is to maximize candidate attention to their states' interests and the issues of the region. Their effort will be at least partially realized on February 5, 2008, when five of the eight states in the region—Arizona, Colorado, Idaho, New Mexico, and Utah—hold nomination events on the same day. And political leaders in other Rocky Mountain states are working to join that group, hoping to create a true western regional primary that could rival “Super Tuesday” in political importance. That, combined with the recent elevation of Nevada's Democratic Party caucuses to January 19 (Barabak 2006) and the Democrats' choice of Denver to host their 2008 National Convention (Cillizza 2007), clearly suggests the emerging importance of the West as a region in presidential political selection.
Despite the importance of both primary front-loading and regional primaries, little is known about the political impacts of these two phenomena. Our investigation therefore focuses on two questions: How much of an impact does the timing of a state's nomination event (how front-loaded it is) have on candidate attention (in the form of advertising and state visits)? Second, how much of an impact does holding a regional primary have on the amount of recognition that states receive from the candidates? In order to answer these questions, we use data from the 2000 and 2004 nomination races on both candidate advertising and candidate travel patterns. We then extrapolate from these data to make predictions about one particular case, asking how much attention the states participating in the western regional primary will receive in 2008.
New Hampshire and the Politics of Front-Loading
As the traditional “first in the nation” primary, the New Hampshire primary has the effect of setting the stage for the presidential contenders early in the primary season. Studies have demonstrated that the “New Hampshire effect” plays a role in determining the final ranking of candidate finishes (Steger, Dowdle, and Adkins 2004). The press coverage from this attention (and the media's “horse-race” coverage of primary campaigns) effectively motivates donors to contribute more money to political candidates—either to make up lost ground if behind or to continue to succeed with positive momentum if ahead (Mutz 1995). Candidates who do well in New Hampshire tend to get their party's nomination (Scala 2003).
The importance of the early primaries (especially New Hampshire) is cemented by recent primary front-loading. Under threat of having their primary date usurped, New Hampshire has taken action several times (beginning in 1971 from Florida) to maintain its “first in the nation” status by moving its primary date from early March to early February and even January (Mayer and Busch 2004). Because of the advantages to the states hosting early primaries, other states have followed New Hampshire's march forward in the election calendar. Front-loading has the clear effect of ensuring candidate attention to the issues of the state (also called “special policy concessions”) and provides an economic boon for the states hosting the primary (Mayer and Busch 2004, 27). Media coverage was also greater for candidates in early primaries; moving one's primary up by one month increased the expected television coverage of a state's campaign by 159 seconds (Mayer and Busch 2004, 35).
Regional Primary History and Implications
The history of front-loading and regional primaries is intertwined, for it was the rise of Super Tuesday as a southern regional primary that caused the front-loading process to take “a huge leap forward” (Mayer and Busch 2004, 13). After sustaining lopsided presidential electoral losses in 1980 and 1984, the Democratic Party hoped to devise a strategy that would minimize a growing Republican advantage in the south and reenergize their electoral primary politics. The creators of the Super Tuesday primary hoped it would provide more clout for the South and that “more candidates, including moderate or conservative candidates, would contest in these primaries” (Norrander 1992, 7). Norrander (1992) notes that the 14-state Super Tuesday primary was a “child of the Democratic Party in the South,” although the eventual effect of the event was to assist Republicans in gaining the support of conservative Democrats (28). The primary may also have assisted “outsider” candidates, such as Al Gore in the 1988 elections (Busch 1997).
The emergence of the southern regional primary, scheduled early in the primary calendar, meant that candidates who won (or placed highly) in the New Hampshire primary could parlay that success immediately into the Super Tuesday contest, which featured a large number of the delegates needed to secure the nomination (see Stanley and Hadley 1987, 94). However, the effect of this front-loading was short lived as several states in the southern regional primary moved their primaries to substantially later dates after the 1988 election (Mayer and Busch 2004), leaving no major regional primary on the political scene.
Since the McGovern-Fraser reforms in 1968, several smaller regional primaries have emerged. Of course, the Super Tuesday southern regional primary was the largest and most visible of these attempts. But, in 1996, a New England regional primary (including every state except New Hampshire) called the “Yankee primary” was scheduled for early March, and four large midwestern states (including Illinois, Michigan, Ohio, and Wisconsin) all scheduled primaries on the same day in mid-March calling their experiment the “Big Ten Primary” (Mayer and Busch 2004, 99). While these were both important attempts to connect regional concerns through a provincial primary, neither of these smaller regional primaries transformed the landscape of presidential nominations.
Moving the Conversation West?
Western states have also recently attempted to alter the nomination calendar by linking their political future. In 1998, representatives from several western states met as the Western Presidential Primary Task Force (participating states included Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming) to investigate the idea of a western primary. However, little came of this effort and a regional primary did not materialize, although Utah, Colorado, and Wyoming all held caucuses or primaries on the same day in 2000 (Massey 2004). This proposal was reinvigorated in 2004 when the Western Governor's Association issued a management directive to invite interested states to join a regional primary in 2008.1
See www.westgov.org/wga/policy/04/caucus.pdf (accessed August 21, 2006).
In practice, however, organizing a western primary has been fraught with logistical and monetary difficulties. One significant obstacle is that the national parties set strict rules about the nature of primaries and the earliest date on which primaries can be held, making systematizing primary rules across several states problematic. Recent rule changes by the Democratic National Committee also may penalize candidates who campaign in states with primaries before the New Hampshire primary by denying candidates delegates from those contests (Nagourney 2006). The Democratic National Committee also prohibits the use of mail ballots before the first Tuesday in March, a rule that may prohibit states from relying on non-traditional election techniques. Further, finding economic motivation from several states to alter their electoral laws to engage in a primary has proved difficult. Costs of between $1 and $5 million to fund separate primaries or to move existing primaries have limited the states that may be involved in a western primary (Vergakis 2006).
Data and Modeling
Our goal in this research, first, was to estimate the impact of both primary timing and regionalism on the amount of candidate attention a state receives. Moreover, given the potential importance of a western states primary, we sought to estimate how much recognition the western states would receive—collectively and individually—should they band together and implement a regional primary in 2008. To do that, we studied patterns of candidate activity in the 2000 and 2004 presidential nomination campaigns, which allowed us to extrapolate to a hypothetical campaign environment in 2008. The dependent variable in our analysis, then, is candidate attention to a state. We measure this concept through both (1) the number of ads candidates aired (per media market)2
This was an important rescaling as different states have different numbers of media markets. Using the total number of ads aired in a state, without taking into account the varying number of media markets, would lead to biased and misleading conclusions.
Ads aired by interest groups on behalf of a candidate are also included in these calculations, though such ads are very small in number. In 2000, for instance, 96% of the ads aired during the primaries were aired by the candidates themselves.
CMAG is now known as TNS Media Intelligence/CMAG. The company employs advertising detectors in the largest media markets in the United States, which record the time, date, and station of airing of each political ad. During the 2000 nomination campaign, the 75 largest media markets in the U.S. were tracked. During the 2004 nomination campaign, the number of media markets tracked expanded to 100, about 86% of the country's homes with television.
Data on candidate travel come from a couple of different sources. The data on visits in 2000 come from an extensive search of the Washington Post archives on Lexis-Nexis. Data from 2004 come from National Journal's Hotline, a political “gossip sheet” that is published each day and reports on the campaign schedule of each candidate.5
The Hotline data are much more exhaustive than the data from the Washington Post, and they cover a longer period of time (beginning January 1, 2003, as opposed to October 1, 1999). As a result, the number of recorded visits during the 2004 campaign is substantially higher (3,380 versus 252 in 2000). To account for this difference, we include a year dummy variable in each of our statistical models. This should resolve any problems of comparing different metrics, assuming data collection was not systematically biased in one year toward any particular state or states, and we have no reason to believe that was the case.
Ad Airings and Candidate Visits by State
How can we explain this variation in candidate attention across states? To predict the number of political spots aired in each state and the number of candidate visits, we considered several possible explanatory factors:
Nomination Event Timing
The closer a state's primary or caucus date to that of New Hampshire (which traditionally holds the first primary), the more attention that state is likely to receive. Using data on the timing of each nomination event obtained from the Federal Election Commission,6
We obtained the dates of all the nomination events from 2000 at www.fec.gov/pages/2kdates.htm and obtained the dates for 2004 from www.fec.gov/pubrec/fe2004/2004pdates.pdf (both accessed on September 7, 2006).
Type of Nomination Event
Traditionally, campaigns have perceived primaries as more important than caucuses or party conventions because the news media give them more coverage. This is a result of (1) the greater number of citizens who generally participate, and (2) the easier task reporters have in explaining the results of primaries. Thus, we created a dummy variable indicating a state held a primary as opposed to a caucus or party convention.
Number of Candidates
The number of candidates seeking their parties' nominations was not equal across the two years or across the two parties. Given this, we controlled for the number of candidates contesting each race, believing that more candidates would lead to more attention for a state.
Number of Delegates
Because the candidate with the most delegates pledged to him or her wins the presidential nomination, one might expect candidates to devote more of their attention to states with large numbers of available delegates.7
Our source for the delegate count for each party and each state was thegreenpapers.com (accessed on September 7, 2006).
Number of Same-Day Events
We hypothesized that states that held isolated primaries or caucuses were likely to receive more candidate attention than those states that held nomination events the same day as many other states. The simple reason is that campaigns only have a limited number of resources, in terms of both time to make campaign appearances and money to buy advertising. Thus, the greater the competition, the less attention any one state is likely to receive.
Number of Events in Prior Week
Our logic was similar here to our previous measure. Even if a state is the only one to hold its nomination event on a particular day, it still may face competition for the resources of the candidates from other states holding nomination events in the previous week.
Regional Events
One way to assess how successful individual states and a new regional primary would be in attracting candidate attention is to examine whether contiguous states holding simultaneous nomination events are able to attract more candidate attention than non-contiguous states holding simultaneous nomination events. Presumably, a candidate who is in a state to campaign would be more likely to pay a visit to a neighboring state than one located across the country. We therefore coded for each state the number of contiguous states holding nomination events on the same day.
Year 2004
We also included in our analysis a dummy variable to indicate the year 2004. This should account for differences in the count of candidate visits owing to our use of different data sources in the two different years.
To estimate the number of ads aired per market in each state and the number of candidate visits in each state, we employed ordinary least squares regression. Importantly, because the dependent variable in both instances was skewed, we took the natural log of both ads per market and visits. This helped to eliminate the heteroskedasticity observed in plots of the model residuals.8
Technically, a Poisson (or negative binomial) count model would have been more appropriate for the data generating process. But given the small sample sizes we were dealing with, we ran into issues of convergence with maximum likelihood estimation and thus decided to report the OLS coefficients.
Results
Table 2 shows the predictors of the number of ads aired (logged) in each media market in each state where the unit of analysis is the state-party-year.9
This should leave us with 150 cases: 50 states each for the Democratic race in 2000, the Republican race in 2000 and the Democratic race in 2004. But because we had no advertising data from some states with no large media markets (e.g., Wyoming, North Dakota, South Dakota, Montana), we were left with only 121 observations.
OLS Predictors of Ads Aired (logged)
Additional findings deserve some discussion. The first is the positive relationship between holding a primary and receiving candidate attention. This relationship, however, is only suggestive as the coefficient just failed to reach conventional levels of statistical significance (p = .155, two tailed, p = .078, one tailed). Second, the number of nomination events held in the previous week has a strong negative influence on the amount of advertising a state receives. This suggests that the more crowded the calendar, the less likely any one state will receive candidate attention. Surprisingly, though, the number of simultaneous events had no statistically significant influence on the number of ads aired, and the number of candidates in the race and year had no impact on the amount of advertising received. Finally, the number of delegates a state had to be won had no discernible impact on the amount of advertising it received. We had hypothesized that larger states would receive more attention, but clearly other factors, such as position on the calendar, were more important, allowing small states such as Iowa, New Hampshire, and South Carolina to garner substantial media attention.
We performed a similar analysis predicting the number of candidate visits to each state. Again, the unit of analysis was the state-party-year, so that Democratic visits to Nevada in 2000 were treated separately from Republican visits to Nevada in 2000 and Democratic visits to Nevada in 2004. The results, shown in Table 3, overwhelmingly confirmed our expectations. First, states that held their nomination events longer after New Hampshire were much less likely to receive a candidate visit. Front-loading clearly matters. The number of simultaneous nomination events in contiguous states, however, had no impact on the frequency with which the candidates visited a state. This finding supports our earlier evidence that regional primaries do little to attract the attention of the candidates to individual states.
Predictors of Candidate Visits (logged)
We also found that primary states received more visits than caucus states, and, in contrast to advertising, the size of the state mattered for candidate visits. States with larger numbers of delegates to be won received considerably more candidate visits than states with small numbers of delegates. That size mattered for candidate visits but not for advertising is not surprising. As Shaw (2006) has noted, the cost of a visit is essentially the same whether a candidate visits a large or small state, but the benefits of visiting a large state are greater because free media coverage reaches a larger audience. By contrast, the costs-benefit ratio for advertising does not change between large and small states because the rate charged for advertising rises as the audience increases.
The crowded nature of the calendar around a state's nomination event also helped to predict the number of candidate visits. As the number of same-day events increased, the attention given to a state declined. The number of visits a state received also declined as the number of nomination events in the prior week increased. Moreover, consistent with our change in coding visits between 2000 and 2004, states received more visits in 2004 than in 2000.
Predicting Candidate Attention
Overall, the results of the two models in the previous section reveal that the amount of candidate attention that a state receives is predictable and depends on a variety of factors. But how does moving a primary closer to New Hampshire influence the number of visits a state receives? How does an increase in the number of previous-week nomination events influence the number of ads aired in a state? And what does this all mean for the western states primary? In order to be able to speak more precisely about the attention to political primaries in 2008 and effects of the proposed western states primary, we predicted the number of advertisement airings and candidate visits that each western state would receive based on the coefficients in the previous statistical models and various assumptions about the characteristics of the 2008 nomination calendar.
Table 4 shows the expected number of Democratic ads per market in each of the western states, progressively adding the number of days after New Hampshire that the nomination event is held and altering the number of nominating events in the previous week.10
In the interest of conserving space, we do not report these same calculations for the Republican race, but the numbers are very similar to the Democratic numbers.
Predicted Number of Democratic Ads per Market in 2008
One important message of Table 4 is that the timing of the western states primary is important for how much attention it would receive. For every 10 additional days past New Hampshire, the predicted number of ads aired per market in each of the states fell by almost half. For example, New Mexico is predicted to receive 383 television ads (assuming no primaries in the previous week) if it holds its primary just 10 days after New Hampshire, but that number declines to just 111 predicted ads if it were to hold its primary 30 days after New Hampshire.
Of course, the “crowdedness” of the nomination calendar matters as well. Increasing the number of events in the previous week from 0 to 10 decreased the predicted ad count by over eight-fold. For example, the predicted number of ads in Utah dropped from 357 to 39. Clearly, the schedulers of a regional primary, assuming they wanted to maximize candidate attention, would want to negotiate between having the primary close to New Hampshire and also having the primary on a relatively isolated part of the calendar. It is the late scheduling problem that likely sunk the “Yankee” (New England) and “Big Ten” (midwestern) regional primaries.
Table 5 further illustrates the relationship between the timing of a primary; its location in a crowded or uncrowded spot on the primary calendar; and the expected amount of candidate attention. As the number of days post-New Hampshire a primary is scheduled increased, the expected number of Democratic candidate visits falls. Moving from 10 days post-New Hampshire to 30 days post-New Hampshire decreased the expected visit count by about a third. This can be seen in Arizona, where the expected number of visits falls from 27 if the primary were held 10 days after New Hampshire to 18 if the primary were held 30 days after New Hampshire. Similarly, moving from an uncrowded spot on the nomination calendar (no events in the previous seven days) to a more crowded spot (10 events in the previous seven days) resulted in a decline in the expected number of candidate visits by almost one-half. Again, a good illustration of this is in Arizona where the predicted number of candidate visits fell from 27 to 14 (assuming the primary was held 10 days after New Hampshire).
Predicted Number of Democratic Candidate Visits by State in 2008
Conclusions
Generally we find that the earlier the primary in the season, the more attention the state will receive from the candidates. Specifically, our analyses have shown that the closer the primary is scheduled to New Hampshire's first-in-the-nation presidential primary, the more likely it is to attract the attention of the candidates. Indeed, if their goal was to attract the attention of the candidates, then the schedulers of a western states primary have chosen wisely in selecting early February 2008 as its date. Of course, all states know that attention comes from scheduling an early primary, and so the calendar is going to be a crowded one in early February 2008. And that bodes poorly for attracting the candidates to the West, as the more crowded the primary calendar is surrounding a state's nomination event, the less likely the candidates are to pay attention.
The good news for the western states is that having a high delegate count—something most western states do not have—is less important than when the primary is scheduled. According to our statistical models, large states were able to attract more candidate visits, but they did no better than small states in attracting the candidates' political advertising. These findings differ slightly from Mayer and Busch's (2004) finding that candidates spend more money in states with larger delegate totals (31). The proliferation of early primaries in 2000 and 2004 seems to have reshaped candidate spending and advertising strategies less towards going after larger delegate pools and more towards participating in early contests.
Perhaps somewhat surprising was our finding that the number of contiguous states holding simultaneous nomination events—our admittedly crude indicator of a regional primary—was in no way related to the amount of candidate attention that a state received. We had expected that the geographic proximity of other nomination events would help draw candidates to campaign in a state, but our data did not support such a conclusion. One possible explanation for this somewhat counterintuitive finding is that there were no true “mega-primaries” in 2000 or 2004 of the type that would attract considerable candidate attention. The largest regional event was a New England primary on March 7, 2000, when five states—Maine, Massachusetts, Rhode Island, New York, and Vermont—held primaries on the same day.
Of course, it also may be the case that this counterintuitive finding was a true finding: that regional primaries attract no more candidate attention than the individual states in a region would attract if they held their primaries on separate dates. This does not, however, discount the other potential benefits of holding a regional primary (such as getting candidates to talk about regional issues or a boost to the regional economy). In addition, despite low numbers of delegates and smaller populations, the regional accumulation of Latino voters (traditionally trending Democratic) and conservative voters (traditionally trending Republican) in the western states may engender additional candidate attention to a western regional primary—especially if a candidate's home state is involved (see Norrander 1993). But our findings do suggest that when it comes to candidate decision-making about where to campaign, timing trumps all else.
Author Bios
Travis N. Ridout is assistant professor of political science at Washington State University. His research focuses on political advertising, media and politics, presidential nominations, and political participation. He is co-author of Campaign Advertising and American Democracy (Temple University Press, 2007).
Brandon Rottinghaus is an assistant professor of political science at the University of Houston. His research interests include presidential leadership, presidential campaigns, public opinion, and media studies.
