Voting at the Political Fault Line "ss1.40"

STUDY DESIGN

To determine the extent of crossover voting in Assembly races and the underlying motivations for this crossover behavior, we conducted an exit poll of approximately 3000 California voters spread throughout five Assembly districts as they left the polls on June 2, 1998. The five different Assembly districts fit into three distinct classifications—each defined by a unique strategic context that allowed us to answer different questions about the extent of and motivations for crossover voting in Assembly races:

― 109 ―

Type 1 Districts: Competitive Primary in a Dominant Party in an Electorally Safe District. Type 1 districts allow us to examine whether voters in the nondominant party will cross over to vote in the competitive primary of the dominant party. We expect the crossover voting in such districts to be motivated by impact voting. We selected Assembly Districts 9 (Sacramento) and 75 (East San Diego County). Assembly District 9, in the city of Sacramento, was an open seat in the primary election, in a heavily Democratic district, where five Democrats and two Republicans were contesting each party's nomination. On the other hand, Assembly District 75, in East San Diego County, was considered a safe Republican seat, with three strong Republican candidates battling for an open seat. There was one Democrat on the ballot in Assembly District 75.
Type 2 Districts: Competitive Primary in One Party in an Electorally Competitive District Likely to Be Hotly Contested in the Fall. Type 2 districts allow us to see if voters in the party with the uncompetitive primary engage in strategic crossover. These voters should be most likely to engage in hedging behavior. We selected Assembly Districts 53 (South Bay, LAX) and 61 (Pomona and Montclair). District 53 ranges from LAX to the South Bay in Los Angeles County. This was a competitive district, with an open Assembly seat. Seven Democratic candidates and one Republican participated in the primary. In Assembly District 61, which is mainly in San Bernardino County and partly in far eastern Los Angeles County, there was another open Assembly seat in what was likely to be a closely contested district in the fall. There were two Democratic candidates in the race (with one, Nell Soto, a clear frontrunner) and four well-matched Republican candidates.
Type 3 District: Majority-Minority District. We were interested in studying a majority-minority district because we wanted to examine whether racial and ethnic minority candidates and voters were in any way adversely affected by the open primary, or would behave differently than whites. We selected Assembly District 49 (East Los Angeles and Monterey Park). This district is heavily Latino (45 percent) and Asian (11 percent), and is considered a strong Democratic seat. Here also there was an open Assembly seat, with five Democratic candidates in the primary and two Republicans.^[1]

In each of the Assembly Districts, we selected at least eight polling places, and the interviews were all conducted by professional interviewers on election day.^[2] Our exit poll instrument was two pages long, copied on the front and back of one sheet of paper; it was self-administered. The first page of the survey instrument contained brief demographic questions, questions about vote choice in the Assembly race and either the gubernatorial or U.S.

― 110 ―

Senate race, and the partisan identification of the voter. Partisan identification was based on what the respondents "usually thought of (themselves) as"; thus, we were measuring partisan identification, not partisan registration. Since the primary purpose of the survey was to measure the intent of voters, this is the more appropriate partisanship measure (see chapter 5 by Sides, Cohen, and Citrin in this volume). The first page also contained a series of questions vital for our analysis of crossover voting motivations. First, voters were asked to rate candidates on a 1–10 scale, where 1 meant they disliked the candidate, and 10 meant they liked the candidate. Then voters were asked to rate the chances that these same candidates, if nominated, would win the general election in November, again on a 1–10 scale, where 1 meant that the particular candidate was "not likely to win," and 10 meant that the candidate was "very likely to win." Voters without an opinion were asked not to mark anything for a particular candidate. The entire sample of voters in each Assembly race was asked to complete these two sets of rating questions for all Assembly candidates; half of the sample was asked to rate the gubernatorial candidates and the other half was asked to rate the U.S. Senate candidates. The back side of the instrument contained questions asking voters to evaluate the open primary, to state how they had voted on three propositions, their opinions on issues, and additional demographic information.^[3] Respondents were able to request questionnaires in English, Spanish, and Chinese, and bilingual interviewers were sent to precincts with large Latino and Asian populations.

Thus, using our exit poll data, we can operationalize a simple measure of crossover voting. Given a voter's stated partisanship and the vote he cast in the Assembly race, we define crossover voters as those who cast a ballot for an Assembly candidate who was running for the opposite party's nomination. Given that we limit our analysis to Democratic and Republican Assembly races, this means that a crossover voter is either a Democrat who voted for a Republican Assembly candidate or a Republican who voted for a Democratic Assembly candidate.

Among the subset of crossover voters, we want to understand their motivations for crossing over. We refer to sincere voters here as sincere crossers. We refer to hedgers and impact voters as strategic crossers; they are not voting for their first choice, but are trying to ensure that they have the best slate of candidates to choose from in the general election. Raiders are neither voting for their first choice nor trying to have the best set of candidates to choose from in the general election. We used our exit poll data to determine the motivations for crossover voting. Using the information from the two rating scales, we could determine whether crossover voters were sincere crossers, strategic crossers, or raiders. Since not all voters were able to rate all candidates, nor were all voters able to estimate the probability

― 111 ―

that each candidate would win the general election, we are not able to classify all voters into the different categories. Thus the sample we work with here is smaller than the sample we report for overall crossover results. This means we are almost certainly overstating the amount of strategic voting, especially the amount of raiding.^[4] To engage in raiding requires that voters know precisely the information we are asking. If a voter is not able to determine the likelihood of different candidates winning the general election, then they are not able to engage in raiding.

In the next section we present the results from our exit poll on crossover voting in these five Assembly districts. In the subsequent section we examine the motivations underlying the crossover voting. We conclude with a summary of our results and a discussion of whether the crossover vote had an impact in any of these Assembly races.