previous sub-section
Should I Stay or Should I Go?
next sub-section

CONCLUSION: DID CROSSOVER VOTING MATTER?

In only one of the five Assembly races we studied—Assembly District 61—was there enough crossover voting to have had an impact on the outcome of either party's primary nomination. In the Republican race, DeMallie won with 19.2 percent of the actual election's total vote; his closest rival was Skropos (15.5 percent of the total vote). In this race,15.1 percent of the Republicans crossed over to vote in the Democratic races. If these Republican crossover voters had to vote for one of the four Republican candidates, our predictive model (presented in table 6.4) states that 15.4 percent would have supported DeMallie and 84.6 percent would have voted for Skropos.[8] Adding these crossover voters to the DeMallie and Skropos vote columns from our exit poll sample, though, would give DeMallie a 33.9 percent to 32.1 percent victory over Skropos in this race among Republican voters. Given the sample size, this is essentially a statistical dead-heat. Thus Skropos's chances were compromised by the blanket primary.[9]

The second race in Assembly District 61 in which crossover voting might have mattered was the Democratic campaign between the two frontrunners McLeod and Soto. Here,12.0 percent of the Democrats crossed over to vote in the Republican primary. Our predictive model states that 36.8 percent of these Democratic crossover voters would have supported McLeod in a closed primary and that 63.2 percent would have voted for Soto in a closed primary.[10] By adding these votes to each candidate's vote total from our exit poll sample, we end up with Soto receiving 46.1 percent of the Democratic vote to McLeod's 41.1 percent (with the remainder going for "other"). Soto still wins this primary election.

Not only do we find little evidence that crossover voting could have changed any of the five Assembly races we studied, we have found generally low levels of crossover voting in the Assembly races. Furthermore, very little of the crossover voting we observe in any of these different races is motivated by strategic reasons. Instead, crossover voting in these races is overwhelmingly sincere—Californians in these Assembly Districts were crossing over to support candidates they sincerely liked more than the candidates who were seeking nomination in their own party.

These results are generally consistent with the conclusions we reached in earlier analyses of crossover voting in Washington State primaries and in presidential primaries (Alvarez and Nagler 1997) and in other analyses of the June 1998 primary in California (Alvarez and Nagler 2000a). In all of these studies, we find that most crossover voting is motivated by sincere considerations, and that very little of such voting is motivated by a desire


120
TABLE 6.4 Multinominal Logit Estimates for Assembly District 61 Voting
  Probability of Supporting:
Independent Variables DeMallie McLeod Skropos Thalman Wickman Other Candidates
NOTE: Soto is the baseline or comparison category in this model; thus the coefficients are interpreted as giving the relative impact of each predictive variable on the probability that a voter would choose any of the Assembly candidates relative to the probability that they would choose Soto.
Females 0.41 −0.13 0.02 −0.96 0.00 −0.63
  (0.44) (0.35) (0.40) (0.58) (0.65) (0.40)
Age 0.07 0.03 −0.50 −0.16 −0.80 −0.27
  (0.22) (0.17) (0.20) (0.26) (0.35) (0.20)
Latinos −1.20 −1.30 −1.80 −2.10 −0.77 −0.40
  (0.68) (0.55) (0.71) (1.10) (0.91) (0.52)
Hispanics −1.90 −0.34 −1.10 −2.20 −0.62 −0.29
  (0.59) (0.37) (0.46) (0.84) (0.69) (0.42)
Prop 226 0.54 0.57 1.10 −0.13 1.20 0.28
  (0.41) (0.34) (0.38) (0.52) (0.61) (0.40)
Prop 227 0.46 0.30 0.12 0.84 0.54 −0.24
  (0.45) (0.34) (0.40) (0.62) (0.65) (0.40)
Ideology 1.60 −0.21 0.78 1.50 1.20 0.69
  (0.27) (0.19) (0.23) (0.33) (0.37) (0.22)
California economy −0.10 0.27 −0.09 −0.09 0.16 0.33
  (0.25) (0.19) (0.23) (0.32) (0.37) (0.21)
Jobs and economy 0.05 0.39 −0.12 0.23 0.07 −0.18
  (0.47) (0.35) (0.42) (0.58) (0.68) (0.41)
Crime −0.79 −0.10 −0.45 −1.90 −0.27 −0.54
  (0.44) (0.35) (0.40) (0.59) (0.65) (0.40)
Experience −0.12 0.29 0.45 0.10 0.79 0.59
  (0.57) (0.43) (0.50) (0.71) (0.80) (0.48)
Abortion −0.69 −0.19 −0.67 0.16 −1.60 −1.10
  (0.76) (0.60) (0.75) (0.83) (1.40) (0.88)
Bilingual education −0.26 −0.39 −0.45 −0.04 0.09 −0.32
  (0.49) (0.41) (0.46) (0.57) (0.69) (0.44)
Special interests 1.00 1.00 1.20 1.20 −0.30 0.20
  (0.51) (0.42) (0.48) (0.59) (0.91) (0.52)
Health care −0.76 0.45 −0.26 −0.14 −0.19 −0.31
  (0.48) (0.35) (0.42) (0.58) (0.70) (0.42)
Environment −2.30 −0.33 −1.50 −0.64 0.52 0.40
  (1.10) (0.48) (0.74) (0.83) (0.84) (0.53)
Cutting taxes 0.92 −0.002 0.17 0.27 −0.22 −0.41
  (0.44) (0.36) (0.41) (0.55) (0.70) (0.44)
New leaders −0.28 0.28 0.25 −0.61 0.98 0.09
  (0.53) (0.41) (0.47) (0.71) (0.69) (0.47)
Constant −5.50 −0.76 −1.30 −4.60 −4.90 −1.80
  (1.40) (0.94) (1.10) (1.80) (1.90) (1.10)

121
to raid the other party's primary. The main motivation for crossover voting, in fact, stems from incumbency; when an incumbent is in a primary election, there is usually high crossover for the incumbent candidate.

Outside of crossover voting for incumbents, though, why is there so little crossover voting? More to the point, why is there so little strategic crossover voting? We believe that there is a simple explanation: strategic crossover voting requires a great deal of voter coordination (Cox 1997). That is, to be a strategic crossover voter requires a fair amount of information about both the dynamics within your own party and the other party primaries and about what is likely to happen in the general election. To be successful, strategic crossover needs to be coordinated on certain candidates—and it is difficult for voters on their own to engage in such coordination. Only when political elites, the mass media, or candidate campaigns provide these coordination cues will there be high levels of strategic voting and will this strategic voting sway the election outcome. Thus the important question really should be, Under what circumstances will elites or candidates be willing and able to undertake the costs of voter coordination in blanket primary elections?


previous sub-section
Should I Stay or Should I Go?
next sub-section