The Nature of Crossover Voters

Authored by: Barbara Norrander

Routledge Handbook of Primary Elections

Print publication date:  February  2018
Online publication date:  February  2018

Print ISBN: 9781138684089
eBook ISBN: 9781315544182
Adobe ISBN:

10.4324/9781315544182-9

 

Abstract

Crossover voters are either the heroes or villains of primary electorates. To those concerned that partisan voters in primary electorates are too ideologically extreme, crossover voters are viewed as a moderating force. This assumption leads many primary reformers to seek out participation formats that downplay the importance of partisan voters and open up participation to more independents. On the other hand, to party officials crossover voters distort the will of the party’s most loyal voters. Indeed, the Democratic Party in its late twentieth-century reform movement emphasized that participation in their presidential nominations should be open to all Democrats, rather than all voters. In the 1980s the Democratic Party even sued the state of Wisconsin to keep it from holding an open presidential primary, though the party subsequently relented and allowed states which traditionally used the format to continue to do so. 1 Arguments for and against the presence of crossover voters in partisan primaries are many and varied.

 Add to shortlist  Cite

The Nature of Crossover Voters

Crossover voters are either the heroes or villains of primary electorates. To those concerned that partisan voters in primary electorates are too ideologically extreme, crossover voters are viewed as a moderating force. This assumption leads many primary reformers to seek out participation formats that downplay the importance of partisan voters and open up participation to more independents. On the other hand, to party officials crossover voters distort the will of the party’s most loyal voters. Indeed, the Democratic Party in its late twentieth-century reform movement emphasized that participation in their presidential nominations should be open to all Democrats, rather than all voters. In the 1980s the Democratic Party even sued the state of Wisconsin to keep it from holding an open presidential primary, though the party subsequently relented and allowed states which traditionally used the format to continue to do so. 1 Arguments for and against the presence of crossover voters in partisan primaries are many and varied.

Positive views of crossover voters include normative concerns, such that all Americans should be able to participate in a primary without having to register as a partisan. The participation of independent voters in primary elections also may be beneficial to the party. The party’s nominee eventually needs to win the general election, and the voice of crossover voters may help to choose a candidate who is more electable. Participation of crossover voters in a primary also may help recruit new voters to join the party’s core supporters (Fowler, Spiliotes and Vavareck 2003; Norrander 1992).

Negative views of crossover voters depict them as less knowledgeable than partisan primary voters about candidates and issues, and as less likely to consider factors such as viability or electability (Fowler, Spiliotes and Vavareck 2003). Party leaders fear that crossover voters will engage in raiding behavior, supporting the weakest candidate to ensure the victory of their party’s candidate in the fall election (Southwell 1988). Finally, various candidates have belied the role of crossover voters when they perceived that these voters gave additional support to an opposing candidate, such as Humphrey versus McGovern in the 1972 Wisconsin primary (Southwell 1988) and Romney versus Santorum in the 2012 Michigan primary.

Views of Crossover Voters by Social Scientists

The political science literature on crossover voters tackles two questions: (1) who should be considered a crossover voter, and (2) does the presence of crossover voters affect the primary outcome. Various answers have been given to the first question. Some political scientists count as crossover voters anyone who is not affiliated with the party holding the primary (Adamany 1976; Ranney 1972). However, other political scientists argue whether independent voters should be included as crossover voters. Some posit that all independents should be excluded from the ranks of crossover voters (Hedlund 1978). Others assert that independents leaning toward the party holding the primary should not be counted as crossover voters (Wekkin 1988). This latter position is often adopted because of the similarities in behavior between leaning independents and weak partisans (Petrocik 1974, 2009). Finally, while crossover voters are most frequently assumed to be a factor in open primaries, political scientists have found crossover voters across all formats, including both open and closed primaries (Alvarez and Nagler 1997; Norrander 2010; Norrander and Wendland 2016). Crossover voting can occur in closed primaries when a person’s party identification does not match their party registration. While for most Americans their party identification matches their party registration (Finkel and Scarrow 1985), for some a discrepancy occurs. Thus, while they are a registered partisan, attitudinally, in terms of party identification, they think of themselves as an independent. Of course, this is not a complication in the 40 percent of states (e.g., open primary states) which do not ask for party preferences on their voter registration forms. Thus, while there are attitudinal partisans across all 50 states, registered partisans exist in only 60 percent of U.S. states.

Often overlooked in the debate on who is a crossover voter is that primary participation rules, and party registration, shape people’s party identification (Burden and Greene 2000; Campbell et al. 1960; Finkel and Scarrow 1985; Gerber, Huber and Washington 2010; Norrander 1989; Thornburg 2014). In states where primary participation rules encourage people to register as a partisan, e.g. closed primary states, there are more partisans and fewer independents. In open primary states without party registration and without a need to have a partisan identity in order to participate in a primary election, there are more independents and fewer partisans. Semi-closed primary rules influence partisanship close to the pattern found in open primary states, as the independent identity is favored by participation rules that provide registered independents with more flexibility in primary participation. Semi-open primary rules, where individuals need to tell an election official which party’s ballot they wish to use, function more like closed primaries. This format was actually the original format for a closed primary (Berdahl 1942). Semi-open primary states also have used a variety of techniques that cement more partisan identities, such as oaths of party loyalty, voter challenges and maintaining records of which party’s primary a voter participated in (Holbrook and La Raja 2008). As the sizes of the partisan and independent groups vary across the states, so does their ideological orientation. In closed primary states the larger group of partisans are more moderate, while in open primary states the smaller group of partisans are more ideologically extreme (Norrander and Wendland 2016).

Another complication in the study of crossover voters, noted in the political science literature, is that voters may maintain multiple partisan identities. In the mid to late twentieth century the slow secular realignment in the South muddied the partisanship waters. Some Americans appeared to have dual partisanship (Jennings and Niemi 1966) or were “segmented-partisans” (Wekkin 1991), identifying as Republicans for national races and as Democrats for state and local politics. Dual partisanship may be less of an explanation today, after the slow secular realignment in the South solidified attachments of conservative voters to the Republican Party at all levels of government by the 1990s. Yet, some journalistic accounts of Donald Trump’s supporters find a segment of voters who identify as Republican but are registered as Democrats (Cohn 2015). Other voters may maintain split partisan preferences across different issue areas, preferring one party on economic issues and the other on social issues. Depending on the central issues in a primary campaign, these voters may crossover to support a candidate supporting their preferred position on a specific issue (Ranney 1972; Wekkin 1991). The intertwining of partisanship and primary participation rules and some potential for lingering dual partisanship means that crossover voting can occur in all primary settings.

The second question tackles whether crossover voters influence the outcome of primary elections, with most social scientists downplaying this possibility. Fowler, Spiliotes and Vavreck (2003) in their New Hampshire study found no difference in candidate preferences for those registered as undeclared (i.e., independents) versus partisans. Studies of the open primary in Wisconsin likewise find few effects for the presence of independent voters in the electorate (Adamany 1976; Hedlund 1978; Ranney 1972), though Wekkin (1991) argues crossover voters helped Hart best Mondale in the 1984 Wisconsin primary. Other political scientists examining a wider variety of races also find some differences between the preferences of partisan and crossover voters but few instances where this significantly changed the outcome of the primary (Geer 1986; Hedlund, Watts and Hedge 1982; Lengle 1981; Southwell 1988, 1991). Thus, at best, social scientists report only a few isolated cases where crossover voters altered the outcome of a presidential primary.

Raiding behavior is especially discounted in the academic literature. Wekkin (1991) summarizing previous findings suggests that less than 5 percent of primary voters engage in activity meant to nominate a weaker candidate in the party opposite of their own (see for example, Adamany 1976; Hedlund 1978; Abramowitz, McGlennon and Rapoport 1981). This evidence of minimal raiding behavior is supported even in cases of orchestrated efforts, such as Rush Limbaugh’s “operation chaos” encouragement of conservatives to vote for Hillary Clinton to prolong the 2008 Democratic primaries and make the nomination race more divisive (Stephenson 2011). Among reasons given for the dearth of raiding behavior is a lack of the political sophistication needed to cast a strategic vote. The uncertainty of the outcome of any primary battle would also dissuade individuals from crossing over to raid another party’s primary or to know which candidate would be the strategically weakest candidate to support. Primary voters, if they cross over, appear to mostly want to hedge their bet and vote for a preferred candidate in the opposing party’s primary (Abramowitz, McGlennon and Rapoport 1981; Southwell 1991; Wolf and Downs 2007). Thus, if their own party’s candidate falters in the general election campaign, these voters still have a good second choice candidate.

How Large Is the Crossover Vote?

Perceptions of how many crossover voters exist are shaped by the existing data on primary voters. This endeavor is complicated by the lack of surveys of actual primary voters across a number of states and primary formats. For presidential primaries, the most typically available surveys are the media exit polls. These exit polls, however, contain only a three-category partisanship question. The independent category in the exit polls includes pure independents and independents that lean toward one party over the other. Independent leaners are distinctive from pure independents and often have more in common with partisan voters. Thus, some political scientists call them “hidden partisans” (Keith et al. 1992). With increasing partisan polarization, independent leaners have come even closer to the characteristics of partisans. Today, the ideology of independent leaners is closer to that of strong partisans than is the ideology of weak partisans to that of strong partisans (Abramowitz 2010: 58–59). Leaning independents are also as loyal as weak partisans in their vote choices (Abramowitz and Webster 2016). While prior researchers have adopted a wide variety of definitions for crossover voters, I support the position that independent leaners are not crossover voters. Leaning independents are participating in the primary of their most preferred party.

While the presidential primary exit polls do not separate primary voters’ partisanship in the academic format of seven categories, the 2010 Cooperative Congressional Election Survey (Ansolabehere 2012) contains the seven-category party identification scale and a validated vote for participation in these off-year state-level primaries. Tables 6.1a and 6.1b show the partisanship of those who participated in the Republican and Democratic primaries recoded to indicate whether their partisanship matched the party of the primary, the breakdown of the independent vote, and how many voted in the primary of the opposite party. Table 6.1a reveals that across all primary types, 51 percent of primary voters are strong partisans, 19 percent are weak partisans, and another 19 percent are independents who lean toward the party holding the primary. Table 6.1b collapses these three categories of primary voters into one and reveals that on average 89 percent of primary participants are not crossover voters. Instead, these voters are the core supporters of the party. The number of pure independents participating in the primaries is quite low, at 6 percent, and the proportion of true crossover voters (e.g., partisans from the opposite party and independents leaning toward the opposite party) is equally low, averaging 5 percent.

Table 6.1a   Partisanship of Voters in 2010 Primaries (in Percentages)

All Primaries

Closed Primaries

Semi-Closed Primaries

Semi-Open Primaries

Open Primaries

Own Party Strong Partisans

51

56

53

48

45

Own Party Weak Partisans

19

23

19

16

13

Own Party Leaners

19

14

18

24

26

Pure Independents

6

4

5

7

9

Opposite Party Leaners

3

2

3

3

3

Opposite Party Weak Partisans

1

1

1

1

2

Opposite Party Strong Partisans

1

1

1

1

2

Number of Cases

14,231

4,433

4,295

4,320

1,184

Table 6.1b   Collapsing Partisanship into Three Categories (in Percentages)

All Primaries

Closed Primaries

Semi-Closed Primaries

Semi-Open Primaries

Open Primaries

Own Party Partisans & Leaners

89

92

90

87

85

Pure Independents

6

4

5

7

9

Opposite Party Partisans & Leaners

5

4

5

5

7

Number of Cases

14,231

4,433

4,295

4,320

1,184

Source: 2010 Cooperative Congressional Election Survey.

Table 6.1b also includes a breakdown of these figures for the different primary participation rules (see Appendix for states in each category). While crossover voters are frequently discussed in connection with open primaries, Table 6.1b reveals that the proportions of true crossover voters do not vary much by primary type (4 percent to 7 percent). Most of the voters across all primary types are core supporters composed of partisans and independents leaning toward the party. These core voters constitute between 85 percent (in open primaries) to 92 percent (in closed primaries) of primary voters. What differs across primary types is whether more of these core party voters designate themselves as partisans or whether they view themselves as independents who lean toward that party.

That nearly 90 percent of primary voters are core supporters of the party greatly diminishes the importance of crossover voters in debates over the nature of primary electorates. It also brings up the question of why so few crossover voters exist even under a format, such as open primaries, that allows voters to move across party lines. Conditions for crossover voting are more scarce than often presumed. Even in open primary states, incentives or opportunities for crossover voting may be reduced by the structure of primary ballots. For state-level primary contests, multiple electoral offices appear on the primary ballot. A primary ballot may contain races for governor, senate, House of Representatives, state legislatures and local offices. This combination of races will tend to anchor partisans and independent leaners to their preferred party’s primary. Further, the nationalization of partisan politics in recent decades ties a single partisan identity to races up and down the electoral ballot (Abramowitz and Webster 2016; Jacobson 2015).

Presidential primaries also present few opportunities for crossover voting. Crossover voting needs a circumstance where one’s own party lacks a nomination contest or where the opposition party’s primary contest is more exciting, such as being more closely fought. Such opportunities for crossover voting in presidential primaries occur in two settings. One is the lack of any primary contest in one’s own party, such as when the party’s sole candidate is an unchallenged incumbent president running for renomination (e.g. Republicans in 1984 and 2004, Democrats in 1996 and 2012). The second opportunity is late in the primary season when one’s own party’s nomination has been unofficially won by a leading candidate while the contest remains competitive for the opposing party (e.g. Republicans in 2008). However, circumstances for crossover voting in late presidential primaries are limited because presidential primaries held after mid-March are typically held jointly with the state’s primaries for other offices. Nomination races in these state-level contests will likely keep most partisans from crossing over to the opposite party to participate in the presidential contest.

The Consequences of Primary Participation by Partisan and Crossover Voters

Reformers often argue that the presence of independent voters will ameliorate the ideological extremity of the partisan primary voters. Yet, here too, the ideological nature of partisan and independent primary voters, and especially the nature of the leaning independent primary voters, is not well understood. Independent leaners are typically better educated than weak partisans (Petrocik 1974), and education is one of the determinants of adopting an ideology. Further, in today’s polarized political world, those with higher levels of education are more likely to perceive the positions of political parties and match their partisan identities, or partisan leanings, to these issues. Thus, even independents who lean toward a political party could be “ideologically extreme,” such that adding them to the primary electorate does not greatly move the ideological orientation of primary voters in the moderate direction.

To investigate how partisanship influences the ideological composition of primary electorates, the 2010 CCES is once again used. This survey contains a validated vote for both the general and primary elections. The validated voter from the general election can be used to calculate the mean ideology of voters in each state. Subsequently, each individual primary voter’s ideology can be compared to his or her state’s general election “median voter” to calculate the ideological deviation of the primary voter from the typical voter in the state. This comparison is crucial, because the median voter across the 50 states does not have the same ideological orientation nor do they necessarily register as moderates. The typical red-state voter, on average, would be more conservative than the typical blue-state voter.

The CCES has a five-point ideology scale which was recoded to be centered on 0. Scores above zero reflect more conservative ideologies and scores below zero indicate more liberal identities. With the comparison of primary voters’ ideologies to their states’ general election median voter, positive coefficients indicate the primary voter is more conservative while negative values indicate the primary voter is more liberal. However, for ease of presentation in a bar chart, the absolute value of the ideological deviation from the states’ median voter is used. A higher value indicates more ideological deviation, though this deviation is in the liberal direction for Democratic primary voters and in the conservative direction for Republican primary voters.

Figure 6.1 demonstrates the influence of different groups of partisan voters to the ideological nature of the Democratic Party’s primary electorate. The first line in each set is the ideological orientation of the primary electorate if only strong Democrats participated. The second line adds in weak partisans, so this line is the ideology of the Democratic primary electorate if only Democrats (weak or strong) voted. The third line reflects the cumulative ideology of the primary electorate when independents leaning toward the Democratic Party are added. The pattern continues until the last line adds in the ideology of strong Republicans who crossed over and voted in the Democratic primary. With this addition, the final line also reflects the actual ideology of all voters in the 2010 primary. The influence of the addition of each group of voters is affected by both their ideology and the size of the group. Thus, while Republicans voting in Democratic primaries have more conservative orientations, their small numbers diminish how much influence they can have on the overall ideology of voters in Democratic primaries.

Ideology of Voters in Democratic Primaries, 2010

Figure 6.1   Ideology of Voters in Democratic Primaries, 2010

Source: 2010 Cooperative Congressional Election Survey.Key: SD = strong Democrat, WD = weak Democrat, ID = independent leaning toward Democratic Democratic Party, I = pure independent, IR = independent leaning toward Republican Party, WR = weak Republican, SR = strong Republican.

As Figure 6.1 reveals, the biggest change in the ideological orientation of Democratic primary electorates occurs with the change from only strong partisans to the inclusion of weak partisans. The subsequent addition of independents leaning toward the Democratic Party has little effect on the overall ideological orientation of the primary electorate. These three groups are the core voters in Democratic primaries, constituting almost 90 percent of voters. Both weak partisans and independent leaners ameliorate the more extreme ideology of strong partisans. Subsequent additions of other independents and crossover voters incrementally move primary electorates in the moderate direction, but none of these changes is as dramatic as the shift in ideology between strong and the weak partisans (and independent leaners).

Results in Figure 6.1 also illustrate patterns across primary types. Contrary to conventional wisdom, the most extreme electorate (represented by the position of the seventh line in each set) is found in open primary states. The extreme ideology of strong partisans in these states helps to move the overall ideological orientation of voters in open Democratic primaries to be more liberal than their states’ general election voter. In contrast, strong partisans in closed primary states are the most moderate. One explanation for this pattern is that in closed primary states, voters’ perceptions are influenced by party registration. The legal attachment shapes the attitudinal attachment and leads to more partisan adherents representing a greater diversity of issue positions and ideological identities. In open primary states, without a legal attachment, partisanship is influenced more by issues, and only those who strongly agree with a party’s issue positions are likely to view themselves as strong partisans.

Figure 6.2 presents results for Republican primaries. Once again, the biggest change in ideological composition of primary electorates, in most cases, comes from the switch from an electorate composed of only strong partisans to one that includes weak partisans, as well. This is true for all primaries combined and for the closed and semi-closed primaries. However, for the semi-open and open primaries the addition of leaning independents also has a significant influence on the overall composition of the primary electorate. A partial explanation for the pattern for semi-open primary states is the predominance of southern states in this group. In these southern states, the dominance of Republican voters in the general election moves the state’s median voter toward the preferences of the Republican Party. Thus, even strong partisans voting in southern Republican primaries are closer to their state’s general election median voter.

Ideology of Voters in Republican Primaries, 2010

Figure 6.2   Ideology of Voters in Republican Primaries, 2010

Source: 2010 Cooperative Congressional Election Survey.Key: SD = strong Democrat, WD = weak Democrat, ID = independent leaning toward Democratic Party, I = pure independent, IR = independent leaning toward Republican Party, WR = weak Republican, SR = strong Republican.

Partisanship and Vote Choice in Presidential Primaries

A final consequence of any differences between crossover voters and partisan voters would occur if the two groups varied in their choice of candidates. Prior research on crossover voters in presidential primaries comes from the late twentieth century and generally finds little influence for crossover voters on primary outcomes (e.g., Southwell 1988, 1991). However, an update to contests at the beginning of the twenty-first century could change that view. Some recent candidates had greater appeal among independent voters. This would include John McCain and Ron Paul on the Republican side and Bernie Sanders in the Democratic contest. However, if presidential primaries are generally won by large margins, any differences in candidate preferences between partisan and crossover voters would have diminished influence on the outcome of primaries. This section of the chapter will first review the competitive levels of recent primary outcomes and second, measure the influence of crossover voters by using the media exit polls.

Table 6.2 investigates the competitive level of presidential primaries over the last three election cycles. Official vote percentages for winning and second-place candidates are compared. To maintain continuity with the next analysis of crossover voters, Table 6.2 only includes primaries and caucuses (Iowa and Nevada) where exit polls were conducted. However, the media cease to conduct exit polls after a race becomes uncompetitive, so primaries included in Table 6.2 are the most competitive ones. The typical margin of victory in presidential primaries is not narrow. The median margin of victory in presidential primaries ranges between 10 and 15 percentage points. Among the recent nomination races, the 2008 Republican contest was the most competitive (signified by the lower median margin of victory), and therefore presented more opportunities for switches in outcomes due to crossover voters. Of course, narrow victories do occur in any election cycle, and figures in Table 6.2 show the smallest margins between first- and second-place are 2 percentage points or less. On the other hand, the largest margins of victory are quite substantial. These largest margins of victory are often found in a candidate’s home state, such as Bernie Sanders winning Vermont’s primary by 72 percentage points, and Mitt Romney winning the 2008 Utah primary by 84 points and the 2012 Massachusetts primary by 60 points. Candidates also win by large margins when a state has a hefty number of their core supporters. For example, Clinton in 2016 won South Carolina’s primary by 58.7 percentage points and Mississippi’s primary by 66.2 percentage points due to the strong presence of African-American voters in southern Democratic primaries and their overwhelming support for Clinton.

Table 6.2   Winning Margins in Presidential Primaries and Caucuses

Median Margin

Smallest Margin

Largest Margin

Number of Primaries with Exit Polls

Democrats 2008

15.3

.8

43.8

37

Democrats 2016

15.6

.2

72.4

27

Republicans 2008

9.8

1.3

84.1

26

Republicans 2012

14.0

2.0

60.0

17

Republicans 2016

14.2

.2

35.1

26

Note: Table includes only primaries and caucuses for which exit polls are available. Entries based on official vote sources as compiled by the author (Norrander 2009, 2013, 2017).

To gauge the influence of crossover voters in presidential primaries, media exit polls are used to compare preferences of partisan voters to the actual outcome of the election. Media exit polls only ask a three-part party identification question: whether a voter is a Democrat, independent or Republican. Thus, while independents leaning toward the party holding the primary should be considered as core supporters of the party, media exit polls do not allow for this refinement. The media exit polls are used to measure which candidate won the highest support among partisan voters. The difference between the support for a candidate solely among partisans and the actual primary outcome is one indicator of the influence of crossover voters. When the winning candidate among partisan voters is different from the winning candidate in the actual primary vote, crossover voters changed the outcome of that primary.

Table 6.3 lists those presidential primaries in which a change in winner would occur if only partisans had voted. The table lists the actual winner of the primary and the margin of that win, followed by the “winner” if only partisans had voted and the margin of that victory. The 2012 Republican contest is missing from this table because none of the primary winners would have differed if only Republicans voted. This might be unexpected, since without any Democratic contest in that year, voters would have had more incentives to crossover to the Republican primaries. Perhaps the extent of partisan polarization today hinders the desires of voters to cross over and vote in the primary of their lesser preferred party.

Table 6.3   Changes in Presidential Primary and Caucus Outcomes if Only Partisans Voted

Event

Actual Winner

Winning Margin

Winner if Only Partisans Voted

Partisan Vote Margin

Democrats 2008

Connecticut

Obama

4.0

Clinton

2

Missouri

Obama

1.4

Clinton

3

Democrats 2016

Oklahoma

Sanders

10.4

Clinton

9

Michigan

Sanders

1.4

Clinton

18

Indiana

Sanders

4.9

Clinton

6

West Virginia

Sanders

15.6

Clinton

4

Wisconsin

Sanders

13.5

Clinton & Sanders tie

0

Republicans 2008

New Hampshire

McCain

5.4

Romney

1

South Carolina

McCain

3.3

Huckabee

1

Florida

McCain

5

Romney & McCain tie

0

Alabama

Huckabee

4.2

McCain

2

Georgia

Huckabee

2.3

3-way tie: Huckabee, McCain & Romney

0

Missouri

McCain

1.5

Huckabee

5

Oklahoma

McCain

3.2

Huckabee & McCain tie

0

Republican 2016

Missouri

Trump

.2

Cruz

3

Note: Winner if only partisans voted based on the candidate support by party identification in the exit polls as posted on CNN website. There was no change in winning candidates for the 2012 Republican contest.

Two other nomination races had minimal influence from crossover voters. In the 2016 Republican contest, only one primary outcome would have changed if only Republicans had voted. This was the essentially tied Missouri primary where it took five weeks for Donald Trump to be declared the official winner (Associated Press 2016). If only Republicans had voted, Ted Cruz would have won over Trump by 3 percentage points. The 2008 Democratic contest between Barack Obama and Hillary Clinton also would be essentially unchanged due to switches in first-place finishes. In only two primaries did crossover voters favor Obama to the extent that Clinton would have won if only Democrats voted. A switch in a second-place finish, however, may have changed the timbre of the media coverage of the Iowa caucuses. While Obama clearly won the Iowa caucuses, John Edwards garnered less than a half of a percentage point more support than did Clinton. Thus, Edwards was declared the second-place finisher while Clinton’s third-place finish became fodder for media questions on the validity of Clinton’s frontrunner status (Reuters 2008). If only Democrats had participated in the Iowa caucuses, Clinton would have garnered second-place by 8 percentage points over Edwards.

In 2016, Clinton also was disadvantaged by the presence of crossover voters in five states. If only partisans voted, Clinton would have won in four of these states and tied with Sanders in Wisconsin. Thus, crossover voting in the 2016 Democratic primaries made it a bit more difficult for Clinton to win the nomination. Nevertheless, across all primaries, Clinton won the most votes and the most pledged delegates. Even if all the presidential primaries had been open primaries, Sanders still would not have won the Democratic nomination (Enten and Silver, May 26, 2016). This is true in part because across all primary formats, partisans outnumber independent voters.

The presidential nomination contest most affected by crossover voters was the 2008 Republican contest. In five primaries, John McCain was the actual winner while if only Republicans voted another candidate would have won. However, which candidate would have won varied by the primary and, in some cases, the difference would have been McCain sharing first place with another candidate. Additionally, in two primaries, it was Huckabee who benefited from the support of crossover voters, and McCain was among the disadvantaged candidates. Perhaps Romney in 2008 was the most disadvantaged by the presence of crossover voters. Without them, Romney would have won in New Hampshire and tied with McCain in Florida. Romney dropped out of the contest after the Super Tuesday primaries, having won only three primaries: his “home states” of Michigan, Massachusetts and Utah. Without crossover voters, Romney would have added two more early victories to his column. Romney also was the rare candidate disadvantaged by the delegate distribution rules used in Republican primaries. With the actual rules, which varied across the primaries, McCain led Romney 680 to 270 delegates at the close of Super Tuesday. If all of the primaries used statewide proportional representation rules, Romney would have led McCain by 425 to 422 delegates (Norrander 2010, 85). Romney in 2008, more than any other recent candidate, had the right to complain that the rules were stacked against him.

Conclusions

Much ado is made of the role of crossover voters in primary elections. Some view crossover voters as an asset, bringing a more moderate voice to the primary electorate and helping to choose a more electable candidate. Others worry that crossover voters distort the preferences of the party’s core supporters and create opportunities for the mischief created when opposing partisans raid a primary. Most social science research, however, tends to downplay the significance of crossover voters.

This chapter contributes to that theme. While the nuances of the partisanship of presidential primary voters are difficult to uncover due to a lack of survey data, data from off-year primary elections demonstrated that nearly 90 percent of primary voters are partisans or independents who lean toward the party holding the primary. Thus, the primary electorate is composed of those with a loyalty to the party, especially today in a world of polarized parties and the nationalization of elections at all levels of government. Examining different primary formats (open, closed and in between) reveals few differences in the general partisan composition of primary electorates. Open primaries have more independents than closed primaries, but again these are overwhelmingly independents leaning toward the party. Moreover, under all formats, partisans comprise the majority of primary voters.

The ideological orientation of primary voters is influenced by the presence of crossover voters, but the biggest change comes between strong partisans (who are the most ideologically extreme) and weak partisans and leaning independents (who are somewhat less extreme). The influence of crossover voters’ ideology (other independents and opposite party partisans) is diminished by their small numbers. Patterns are similar across the different participation formats (e.g., open versus closed) and mostly the same across the two parties. However, the overall ideology of primary electorates does not conform to the conventional wisdom that open primary electorates are more moderate than those found in closed primary states. On the Democratic side, open primaries had the most ideologically extreme electorate, because they have the most ideologically extreme strong partisans. For Republican primaries, closed and semi-closed primaries do have more ideologically extreme electorates. The semi-open Republican primaries have the most moderate electorates, but this is explained by the predominance of this format in the southern states where Republican voters have moved the general election median voter further to the conservative side.

In some, but not all, presidential nomination contests, candidate choice is slightly influenced by the presence of independents and opposite party partisans participating in a party’s primaries. The existence of crossover voters had no effect on the 2012 Republican contest and altered the outcome of only one or two primaries in the 2008 Democratic and 2016 Republican nomination races. Clinton in 2016 was disadvantaged in five primaries that Sanders won with the support of crossover voters, and Romney lost two crucial early contests (New Hampshire and Florida) in 2008 that he would have won if only Republicans had voted. Still, these cases of a switch in the winning candidate comprise only 11 percent of the primaries in which this change could be measured.

Several reasons explain why crossover voters have minor effects on the outcome of presidential primaries. One reason is that the average primary is won by a margin of more than 10 percentage points. Narrow victories are generally required for a change in a portion of the vote to alter the outcome. A second explanation is that across participation formats (open, closed and in between) the majority of primary voters are partisans and not independents or other crossover voters. Third, the candidate advantaged by the presence of crossover voters is not always consistent across nomination contests, such as the 2008 Republican contest where in some cases Huckabee rather than McCain was the advantaged candidate. Finally, even if a number of primary outcomes may have changed, there is little reason to suspect that this would have altered the overall outcome of the nomination. In the end, crossover voters are neither the bugaboo feared by their critics nor the savior of the primaries as proposed by their supporters.

Appendix

A.1   Classification of Primary Type by States and Parties, 2010

Closed

Semi-Closed

Semi-Open

Open

Connecticut

Delaware

Florida

Kentucky

Maryland

Nevada

New Mexico

New York

Oklahoma

Oregon

Pennsylvania

South Dakota – Rep.

Utah – Rep.

Alaska – Rep.

Arizona

California

Colorado

Iowa

Kansas

Maine

Massachusetts

Nebraska

New Hampshire

New Jersey

North Carolina

Rhode Island

South Dakota – Dem.

West Virginia

Wyoming

Alabama

Arkansas

Georgia

Illinois

Indiana

Mississippi

Missouri

Ohio

South Carolina

Tennessee

Texas

Alaska – Dem.

Hawaii

Idaho

Michigan

Minnesota

Montana

North Dakota

Utah – Dem.

Vermont

Wisconsin

Source: Holbrook and La Raja (2008).

Note: For the 2010 analyses, Louisiana and Washington State are excluded because of their nonpartisan primaries. Virginia is excluded because of no validated votes in CCES. The Alaska Democratic Party shared a blanket primary with the Alaskan Independence Party and Libertarian Party. Since Alaska Democratic Party’s candidates received the bulk of the primary votes, we have classified it as an open primary. In 2010, the South Dakota Democratic Party allowed independents to vote in its primary (Springer 2012). Utah law allows the political parties to choose the format for their primaries (FairVote 2012). Some states use a different primary format for presidential primaries. For example, Arizona holds semi-closed congressional primaries and closed presidential primaries.

Note

The U.S. Supreme Court in Democratic Party of the United States v. Wisconsin ex rel. LaFollette, 450 US 107 (1981), supported the national party’s efforts to ban open primaries.

References

Abramowitz, Alan I. 2010. The Disappearing Center . New Haven, CT: Yale University Press.
Abramowitz, Alan I. , and Steven Webster . 2016. “The Rise of Negative Partisanship and the Nationalization of U.S. Elections in the 21st Century.” Electoral Studies 41 (1): 12–22.
Abramowitz, Alan I. , John McGlennon , and Ronald Rapoport . 1981. “A Note on Strategic Voting in a Primary Election.” Journal of Politics 43 (4): 899–904.
Adamany, David . 1976. “Communications: Cross-over Voting and the Democratic Party’s Reform Rules.” American Political Science Review 70 (3): 536–541.
Alvarez, R. Michael , and Jonathan Nagler . 1997. “Analysis of Crossover and Strategic Voting.” Pasadena: California Instititute of Technology Social Science Working Paper Number 1019.
Ansolabehere, Stephen. 2012. Cooperative Congressional Election Study, 2010: Common Content. [Computer File] Release 2: August 10, 2012. Cambridge, MA: Harvard University, http://cces.gov.harvard.edu.
Associated Press. April 19, 2016. “Cruz Won’t Seek Recount of Missouri Primary Loss to Trump.” Fox News, http://www.foxnews.com/politics/2016/04/19/cruz-wont-seek-recount-missouri-primary-loss-to-trump.html
Berdahl, Clarence A. 1942. “Party Membership in the United States, I.” American Political Science Review 36 (1): 16–50.
Burden, Barry C. , and Steven Greene . 2000. “Party Attachments and State Election Laws.” Political Research Quarterly 53 (1): 63–76.
Campbell, Angus , Philip E. Converse , Warren E. Miller , and Donald E. Stokes . 1960. The American Voter. New York: John Wiley & Sons.
Cohn, Nate . 2015. “Donald Trump’s Strongest Supporters: A Certain Kind of Democrat.” New York Times, December 31, http://www.nytimes.com/2015/12/31/upshot/donald-trumps-strongest-supporters-a-certain-kind-of-democrat.html (accessed January 4, 2016).
Enten, Harry , and Nate Silver . May 26, 2016. “The System Isn’t ‘Rigged’ Against Sanders.” FiveThirtyEight http://fivethirtyeight.com/features/the-system-isnt-rigged-against-sanders/
FairVote. 2012. “Congressional and Presidential Primaries: Open, Closed, Semi-Closed, and ‘Top Two’.” May, 2012. http://www.fairvote.org/research-and-analysis/presidential-elections/congressional-and-presidential-primaries-open-closed-semi-closed-and-top-two/, accessed May 5, 2014.
Finkel, Steven E. , and Howard A. Scarrow . 1985. “Party Identification and Party Enrollment: The Difference and the Consequence.” Journal of Politics 47 (3): 620–642.
Fowler, Linda L. , Constantine J. Spiliotes , and Lynn Vavreck . 2003. “Sheep in Wolves’ Clothing: Undeclared Voters in New Hampshire’s Open Primary.” PS: Political Science & Politics 36 (1): 159–163.
Geer, John G. 1986. “Rules Governing Presidential Primaries.” Journal of Politics 48 (4): 1006–1025.
Gerber, Alan S. , Gregory A. Huber , and Ebonya Washington . 2010. “Party Affiliation, Partisanship, and Political Beliefs.” American Political Science Review 104 (3): 720–744.
Hedlund, Ronald D. 1978. “Crossover Voting in a 1976 Open Primary.” Public Opinion Quarterly 41 (3): 498–514.
Hedlund, Ronald D. , Meredith W. Watts , and David M. Hedge . 1982. “Voting in an Open Primary.” American Politics Quarterly 14 (1): 55–73.
Holbrook, Thomas M. and Raymond J. La Raja . 2008. “Parties and Elections.” In Politics in the American States: A Comparative Analysis, 9th edition, ed., Virginia Gray and Russell L. Hanson . Washington, D.C.: CQ Press, pp. 61–97.
Jacobson, Gary C. 2015. “Obama and Nationalized Electoral Politics in the 2014 Midterm.” Political Science Quarterly 130 (1): 1–25.
Jennings, M. Kent , and Richard G. Niemi . 1966. “Party Identification at Multiple Levels of Government.” American Journal of Sociology 72 (1): 86–101.
Keith, Bruce E. , David B. Magleby , Candice J. Nelson , Elizabeth Orr , Mark C. Westlye , and Raymond E. Wolfinger . 1992. The Myth of the Independent Voter. Berkeley, CA: University of California Press.
Lengle, James I. 1981. Representation and Presidential Primaries: The Democratic Party in the Post Reform Era. Westport, CT: Greenwood Press.
Norrander, Barbara . 1989. “Explaining Cross-State Variation in Independent Identification.” American Journal of Political Science 33 (3): 516–536.
Norrander, Barbara . 1992. Super Tuesday: Regional Politics and Presidential Primaries. Lexington: University Press of Kentucky.
Norrander, Barbara . 2009. “Democratic Marathon, Republican Sprint: The 2008 Presidential Nominations.” In The American Elections of 2008, ed., Janet M. Box-Steffensmeier and Steven E. Schier . Lanham, MD: Rowman and Littlefield, pp. 33–54.
Norrander, Barbara . 2010. The Imperfect Primary: Oddities, Biases, and Strengths of U.S. Presidential Nomination Politics. New York: Routledge.
Norrander, Barbara . 2013. “Fighting Off Challengers: The 2012 Nomination of Mitt Romney.” In The American Elections of 2012, ed. Janet M. Box-Steffensmeier and Steven E. Schier . New York: Routledge, pp. 48–72.
Norrander, Barbara . 2017. “The Conventional Versus the Unconventional: Presidential Nominations in 2016.” In Winning the Presidency 2016, ed. William J. Crotty . New York: Routledge.
Norrander, Barbara and Jay Wendland . 2016. “Open Versus Closed Primaries and the Ideological Composition of Presidential Primary Electorates.” Electoral Studies 42 (2): 229–236.
Petrocik, John R. 1974. “An Analysis of the Intransitives in the Index of Party Identification.” Political Methodology 1 (1): 31–47.
Petrocik, John R. 2009. “Measuring Party Support: Leaners Are Not Independents.” Electoral Studies 28 (3): 562–572.
Ranney, Austin . 1972. “Turnout and Representation in Presidential Primary Elections.” American Political Science Review 66 (1): 21–37.
Reuters. January 3, 2008. “Hillary Clinton Places Third in Iowa Setback: CNN.” January 3, http://www.reuters.com/article/us-usa-politics-clinton-idUSN0214716420080104
Southwell, Priscilla . 1988. “Open Versus Closed Primaries and Candidate Fortunes, 1972–1984.” American Politics Quarterly 16 (2): 280–295.
Southwell, Priscilla . 1991. “Open Versus Closed Primaries: The Effect on Strategic Voting and Candidate Fortunes.” Social Science Quarterly 44 (3): 789–796.
Springer, Jennifer . 2012. “South Dakota Voter Registration Down.” Ballotpedia, August 14, 2012. http://ballotpedia.org/South_Dakota_voter_registration_down, accessed May 5, 2014.
Stephenson, E. Frank . 2011. “Strategic Voting in Open Primaries: Evidence from Rush Limbaugh’s ‘Operation Chaos’.” Public Choice 148 (3): 445–573.
Thornburg, Matthew P. 2014. “Party Registration and Party Self-Identification: Exploring the Role of Electoral Institutions in Attitudes and Behaviors.” Electoral Studies 36(1): 137–148
Wekkin, Gary D. 1988. “The Conceptualization and Measurement of Crossover Voting.” Western Political Quarterly 41 (1) : 105–114.
Wekkin, Gary D. 1991. “Why Crossover Voters are Not ‘Mischievous Voters’: The Segmented Partisanship Hypothesis.” American Politics Quarterly 19 (1): 229–247.
Wolf, Michael R. , and Andrew Downs . 2007. “The Missing 3,847 Voters: Strategic Voting in a Congressional Primary.” Indiana Journal of Political Science 10: 10–22.
Search for more...
Back to top

Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.