A SHORT HISTORY

Between 1924 and 1968 state and national party elites controlled how delegates to the national convention were selected. This endeared most delegates to state leaders rather than to the individual candidates who were seeking the presidential nomination. Often, this leader was the state governor who bestowed delegate status based on past party work and loyalty. State party delegates, in turn, bargained at the national convention to select the presidential nominee. Therefore, potential presidential candidates pursued an “inside strategy” in their quest for their party's nomination by working with and through party leaders. It was not necessary for them to actively participate in primary contests or appeal to the public for support.

Indeed, nomination rules and procedures made participation by rank-and-file party members quite difficult in many states. During this period caucuses, not primaries, dominated the nomination process, which made it easier for state-level party elites to maintain control, especially in the delegate-selection process. Many states used secret caucuses, closed slate-making, and extensive proxy voting for determining delegate selection. Some states had no formal rules for delegate selection or they changed their rules after delegates had been selected. In other cases, the delegates to the national convention were selected prior to the presidential election year before the slate of presidential candidates was even known (see the Commission on Party Structure and Delegate Selection 1970). As a result, candidate enthusiasts found it difficult to participate in any meaningful way.

The procedural impediments to general party participation led to a clamor for change at the 1968 convention by reformers who envisioned a more open and democratic nominating system. Delegates inside and protesters outside the 1968 Democratic National Convention were angered by their perception that Hubert Humphrey was an illegitimate party nominee; he had received the party's nomination without entering a single primary. Yet party rules and procedures fully allowed for such an outcome. Reformers believed that too much power rested in the hands of the party elites and too little power rested in the hands of the rank-and-file members of the party. Therefore, many convention delegates explicitly called for direct democracy in the selection of the party nominee.

In response, the party elite resolved to appoint a commission to make recommendations to ensure that all Democrats receive a “full, meaningful, and timely” opportunity to participate in the delegate selection process (Commission on Party Structure and Delegate Selection 1970). The McGovern-Fraser Commission, named for its chairs senator George McGovern and representative Donald Fraser, recommended rule changes to the nomination process that were adopted by the DNC in 1971 and applied for the first time during the 1972 presidential nominating campaign. The new rules provided for procedural fairness, proportional representation of presidential candidate supporters, and affirmative action policies to force delegates to represent their underlying demographic groups in the electorate. The purpose of these reforms was to legitimize the party's selection of the presidential party nominee by increasing participation and by opening the party to underrepresented constituencies.

The rule changes transferred the responsibility of selecting a party nominee from the party professionals to the party rank and file, allowing for greater participation in the selection process. They also ensured that the outcomes of primaries and caucuses would influence the collective outcome of the party convention. Prior to the McGovern-Fraser reforms party primary campaigns were primarily beauty contests in which candidates could demonstrate to party leaders their broad appeal and ability to win votes. The results of the primaries did not determine a state delegation's support for the candidate at the convention. Post McGovern-Fraser, however, primary and caucus outcomes were directly linked to the number of delegates each candidate received, especially within the Democratic Party where proportional allocation of delegates is mandated. The thrust of these changes placed the decision-making power for the party's nomination in the hands of primary and caucus participants and away from party leaders.

DOES THE CURRENT SYSTEM MEET THE STANDARD OF MEANINGFUL PARTICIPATION?

The purpose of the reforms was to create opportunities for meaningful participation for party members. Indeed, the mandate for reform mentions the word meaningful in relation to participation 22 times and repeatedly emphasizes its importance as the basis for its recommendations for change. Within the report, meaningful participation implies two major standards to judge the reforms. First, it implies a standard of broad access to participation in the selection process of the party nominee. Second, it implies that nomination participants' preferences and choices should have value in selecting the nominee. Senator Fred Harris, the chair of the Democratic National Committee (DNC) that appointed the McGovern-Fraser commission, said that, “We have no greater task than assuring that ours will be an open party, encouraging the widest possible participation in all of our affairs. The Democratic Party must serve, not be served, it must facilitate choice, not deny it; it must invite diversity, not discourage it” (Commission on Party Structure and Delegate Selection 1970, 4).

The current process is now in its tenth iteration but it has been subjected to continual tinkering throughout the years as candidates and political elites look for ways to manipulate the process to their advantage. Thus, an examination of whether the outcomes of the reforms have effectively served the goal of meaningful participation is in order.

PRIMARIES OR CAUCUSES?

One change that came as a surprise to many was the quick switch from caucuses to presidential preference primaries as the preferred device for delegate selection. State leaders saw the primary as the easiest way to assure that state delegates were selected in a way that was consistent with the new rules. Interestingly, it also served to keep candidate enthusiasts out of other party business that typically takes places within the precinct caucuses as well as the county and state party conventions. From 1912 through 1968, the average number of primaries per presidential year was 15.6, but has averaged 33.4 (p < 0.001, two-tailed test) since the rule changes.

This shift to primaries can be judged positively based on our standard of meaningful participation. Primaries promote open and meaningful political participation because the costs of participating are similar to any other state election. Needless to say, the costs are much lower than for caucuses, where participants discuss issues and candidate preferences over an extended period of time. As a result, a larger numbers of voters go to the polls than go to the caucuses, thereby enhancing broad participation in the process (Cook and Kaplan Reference Cook and Kaplan1988).

Primaries also come closer than caucuses to meeting the second standard—the translation of rank-and-file preferences into binding outcomes. Caucuses have multiple layers of voting that begin at the precinct level and progressively aggregate up to the congressional district and state conventions. Thus, the events of the precinct caucus votes do not necessarily carry over to the later county, district, and state conventions. Votes cast by delegates at these later events are not bound by the outcomes of the precinct caucuses and delegate totals in these circumstances do change. Therefore, we argue they provide less meaningful participation than primary contests.

THE CONSEQUENCES OF FRONTLOADING FOR MEANINGFUL PARTICIPATION

The second significant feature of the nomination process that affects meaningful participation is the timing of the state-level events. States select the dates of their nomination contests independently, though party rules provide some guidelines, and states have incentives to schedule them early. The collective result has been an increasingly frontloaded nomination calendar with many states vying to hold primaries early in the season. Frontloading began in earnest in 1988 and has continued to increase. Frontloading, in fact, was seen as so important in 2008 that two states moved their nominating events to dates where they were openly defying national party rules. Figure 1 compares the proportion of Democrat and Republican primaries that were completed, by sequence, in 1976, 1988, and 2008. In 1976, the primary season was largely drawn out with less than 50% of primaries conducted half way through the time period, and this same pattern holds for all competitive early primaries prior to 1988. In 1988, however, frontloading is quite evident, with more than 50% of primaries conducted by week seven. The 2008 pattern shows an even greater rise in the number of primaries completed quickly.

Figure 1.

Percentage of Primaries Complete By Primary Sequence

Frontloading is a tremendous change to the electoral context in the sequential system of delegate selection. The compression of the primary system influences candidate behavior and ultimately voter behavior. Frontloading has led to an increase in the importance of events in the invisible primary (Flowers, Haynes, and Crespin Reference Flowers, Haynes and Crespin2003; Haynes and Murray Reference Haynes and Murray1998; Mayer and Hagen Reference Mayer, Hagen and Mayer2000) and the advantage given to the frontrunner (Mayer and Busch Reference Mayer and Busch2004), and it has increased the problem of information to the voter (Norrander Reference Norrander1992).

Perhaps the most important way that frontloading influences participation is by determining the eventual nominee earlier in the process, leaving many voters to participate in a fictional contest of preferences when the outcome of the race is already known. As a result, frontloading enhances the importance of the earliest contests and reduces the importance of later contests. This happens because delegates accumulate more quickly for the eventual winner and because the race winnows more quickly. The forces of momentum send clear signals of viability to voters and campaign donors (Mayer and Hagen Reference Mayer, Hagen and Mayer2000; Mayer and Busch Reference Mayer and Busch2004; Atkeson Reference Atkeson, Patton and Robinson2009), so many candidates expend the bulk of their resources quickly in the hopes that a big win will propel their candidacy to the next set of nominating events. Candidates withdraw more quickly when their campaigns run out of steam because they are not anointed with “big mo.”

Prior to frontloading candidates stayed in the race much longer, creating a larger field for voters to consider and providing greater incentives for voters across the electorate to tune in and participate in the selection of their party's nominee. In 1972 for example, nine Democratic candidates ran for the party nomination and by the end of the nomination season two-thirds of them remained. In comparison, in 1988 only two Democratic candidates out of seven remained during the entire delegate selection period; and for Republicans only the eventual nominee was left over two months before the delegate-selection process ended. By 2000 both parties effectively ended their campaigns almost three months before the last primary was held; and, in the Republican Party, at least half the field withdrew before the first delegate selection event was conducted.

Table 1 presents the number of primary elections remaining after the nominee is essentially determined, either because the media called a winner, because one candidate amassed enough delegates to cross the 50% delegate threshold, or because all but one candidate dropped out. Examining competitive, non-incumbent contests (incumbent races shown in bold), there is a clear increase in the number of primaries remaining after the advent in frontloading with the exception of the 2008 Democratic race, a point we will return to later. In 1972 and 1976 nearly all primaries took place before the race was called. By 1988, however, over one-third of the primaries in both parties remained when the respective-party candidates were anointed as winners. Four years later, over half the races remained and by 2000 three in five—a huge majority—remained. Even though John McCain did not cross the 50% threshold until March 4, he was essentially the nominee after Tsunami Tuesday when Mitt Romney, who had the second-highest delegate count, withdrew. Mike Huckabee remained in the race, but momentum would have had to significantly change for him to catch up to McCain, who had more resources and nearly three times as many delegates, particularly given the more common winner-take-all delegate allocation rule that many Republican state parties use.

Table 1.

Number and Percentage of Primaries Remaining after Nominee is Determined, 1972–2004

^a Bold numbers are sitting presidents running for reelection.

^b Romney suspended his campaign on February 7, 2008, leaving little mathematical chance that Huckabee could catch McCain. Huckabee did not concede the race until after the primaries held on March 4 when McCain surpassed 50% of the delegate count. At that point there were 11 primaries left in the process.

The increasing trend to declare the winner quickly affects the type of campaign to which voters in different states are exposed. In the earliest states, voters see a full slate of candidates, intense campaigning, and a disproportionate share of media attention (Bartels Reference Bartels1988). This is because early contests are important to momentum and for the most part candidates place all their campaign chips in the earliest contests. The concentration on advertising in these states reflects candidates' needs to perform better than expected in the first few primaries to stimulate momentum for future races. Not surprisingly, the media responds to the intense campaigning early in the process with almost constant coverage in early races, but a drop off in coverage quickly occurs as the campaign events become less important to the nomination outcome (Robinson and Sheehan Reference Robinson and Sheehan1983; Bartels Reference Bartels1988).

Two other aspects of frontloading are worth considering because they also affect the context of the campaigns to which voters are exposed. The first is that frontloading increases primary compression, leading to a larger number of contests on any given day. For example, in 2008 the Democrats opened up their window of participation on February 5, resulting in Tsunami Tuesday with the largest number of contests ever held on one day. Compression is problematic for candidate strategy because candidates must divide their campaign dollars among numerous states at the same time. Because candidates court different constituencies (Gurian and Haynes Reference Gurian and Haynes1993), each candidate makes different strategic decisions resulting in very different primary contests across states. These differences include different levels of competition, mobilization, and information for nomination participants. These strategic choices result in different levels of turnout and judgments across electorates. We argue that these factors undermine meaningful participation.

The second is that frontloading often leads to separation of the presidential nomination contest from other nomination contests in the state (Norrander Reference Norrander1992). In the era before frontloading was prevalent (pre-1988), most states combined their state and federal ballots. After frontloading, however, a majority of states held their presidential primaries on a different date than their state primaries (Atkeson and Maestas Reference Atkeson and Maestas2008). Placing both elections on the same day enhances voter interest as more races are on the ballot and more candidates across races actively compete. The separation of these contests reduces turnout and functionally reduces the meaningful participation of primary voters in presidential and state politics.

Does the frontloaded system that has evolved since the reforms meet the goals of meaningful participation as sought by reformers? We think not. The current incentive structure privileges voters in some states and penalizes those in others, thereby failing to meet the goal of encouraging broad participation. In the earliest states, voters are exposed to an intense multi-candidate contest, where their decisions are seen as the harbinger of the campaign. Because the race is new and every candidate has a chance to win the nomination (at least in theory), votes in these states “count” more. The outcomes from early state contests provide important cues to future voters about the viability and electability of candidates, and can breathe life and the all-important momentum into the underdog campaign. As each subsequent election takes place, however, the dynamics of the campaign change: the field is winnowed, candidates cease campaigning, the “cues” from any one state become less important to future states, and media attention wanes (Norrander Reference Norrander2000; Haynes and Murray Reference Haynes and Murray1998).

Once the winner of the nomination has been declared voters in remaining states lack meaningful choices at the polls. Without the prodding that an active campaign provides and because whatever marginal effect voters' choices would make to an election outcome is lost when the candidate is known, voters have fewer incentives to participate. It is not surprising to find that state primary turnout declines in states that fall late in the sequence of contests. Estimates from an empirical model suggests that between 3.5 to 7% (or between 71,000 and 17,8000 voters) per state are turned off by their position in the process and their lack of meaningful participation (Atkeson and Maestas Reference Atkeson and Maestas2008).

DOES 2008 SUGGEST FRONTLOADING DOESN'T MATTER?

There is no doubt that the 2008 Democratic contest was one of the most dramatic and drawn-out nominations contests in recent times and the level of participation was greater than in previous elections. Does this mean that frontloading is not an impediment to broad participation? Probably it is not. First, frontloading certainly influenced the Republican race in the way we describe above, by winnowing the field to a single viable candidate quite quickly after Tsunami Tuesday. The winner-take-all delegate counts piled up rapidly for McCain and, although Huckabee stayed in the race until March 4, the race was effectively over after Romney suspended his campaign on February 7. Nineteen primaries still remained.

Although the Democrat race winnowed to two candidates quickly, the two candidates remained in a competitive contest to the end of the process. But, the oddities of this race likely make it unique. For example, the status of Michigan and Florida delegates were unclear until a party deal was brokered at the end of the process. As early states in the process, they should have provided momentum to the winning candidate going into Tsunami Tuesday, but the degree to which a win in those states mattered was unclear. Subsequent primary voters were left with confusing signals about momentum. That meant that voters went into Tsunami Tuesday with no clear frontrunner from the earlier contests and the short distance between contests left little time for voters to deliberate their choices. Instead the vote-share differences across races were relatively small and there were two state wins for senators Barack Obama and Hillary Clinton and the rest of the field, except John Edwards who was hanging on by a thread, had conceded. This, combined with rules that proportionally allocated delegates, and candidates' strategic targeting of resources on Tsunami Tuesday, lead to an extremely close split in delegate counts and state wins in what became a two-person contest. The renewed competitiveness of the election and the full campaign coffers of Obama and Clinton created a highly unusual circumstance never seen before and led to an uptick in turnout and a renewed enthusiasm and interest in the campaign. Given that this is one very unusual contest out of 20 since the reforms, it appears to be the odd case out and is not reflective of normal nomination politics.

That being said, it may highlight an important problem in frontloading never considered before because compression had never been as intense as it was on Tsunami Tuesday 2008. Specifically, it suggests that if there is not enough time for momentum to signal to voters the frontrunner, a compressed schedule might actually exacerbate competition and party divisiveness because of candidate strategies, creating a drawn out, but not necessarily meaningful battle. Thus, in this scenario, voters do not have clear signals from previous voters, nor do they have the full campaign to respond to because of candidate incentives to target their campaigning efforts and because of the speed of the process.

Thus, even in the 2008 contest we argue that frontloading was consequential to the outcome. Extreme compression in the race led candidates to employ selective mobilization efforts. Obama concentrated his efforts on the states holding caucuses while ceding some key primary states to Clinton, and Clinton focused on primary elections rather than caucuses. One problem with frontloading is that compression forces candidates to strategically select where to compete and this has consequences for the participation of state electorates. States where one candidate strategically cedes the race and skips campaigning generates less effective participation. Rank-and-file party members are not drawn into the race in the same way they would be in a state where both candidates choose to compete. This means that outcomes are less reflective of the underlying rank-and-file preferences than if the nomination contests had been spread out over time, encouraging two competitive candidates to focus on the same state. The result was an aggregate outcome that increased overall participation throughout the process, but with uneven mobilization of underlying electorates in individual states. Thus, although frontloading did not stymie broad participation in this case, it likely created biases in participation in individual state electorates due to selective state targeting by candidates.