• No results found

Change in Stability and Imitation not Driven by Pre vious Electoral Results

Until now, the dependent variables have been the level of stability and imitation in plat- form symbolism. However, there are important reasons to go a step deeper and examine

changes in stability and imitation, instead of their absolute levels. Theoretically, the

behavior of a party in the current election should be compared to its behavior in the previous election if we want to know whether previous outcomes changed how they be- haved. A party that had been extremely stable may deviate from its previous language but still appear to be more stable than in periods when its language is changing regularly from election to election. Therefore, it makes sense to root the measures of stability and imitation in what the party did in the previous election, standardizing current behavior against recent history.

The empirical reality of these data compounds the need to examine changes in, rather than just levels of, stability and imitation in platform symbolism. As shown in the first section of these results, elite campaign language has become generally more consistent since the Civil War, particularly since the 1950’s. Platform language from the 19th century was not very consistent across elections while modern platforms are much more stable. Because a great deal of the variance in stability and imitation is connected to this long-term trend toward more consistent elite language, it is important to ensure that short-term effects predicted by accommodation theory are not simply being numerically overwhelmed in the level data.

Table 6.9: Change in Symbolic Stability as a Function of Previous Electoral

Outcomes

Predictor ∆ Policy Stability ∆ Value Stability

coefficients (s.e.)

Presidential Votet−1 .20 .14

(.18) (.22)

Off-Year Congressional Vote .26 −.16

(.28) (.41) Constant −.08 −.11 −.06 .09 (.10) (.14) (.11) (.20) Adjusted r2 .00 −.02 −.01 −.02 N 72 54 72 54 *=p<.05

Note: Unit of analysis is the change in symbolic stability and each party’s performance in recent elections. The dependent variable in the first two columns of results in the change in policy-specific symbolic stability for each party, or the similarity of a party’s current platform to their own previous platform minus the similarity of their previous platform to their platform of eight years earlier. The dependent variable in the last two columns is the same change measure within value symbolism. The independent variable in the first and third columns of results is the party’s two-party vote share in the previous presidential election. The independent variable in the second and fourth columns of results is the party’s vote share in all House races from the off-year congressional election.

The dependent variable in table 6.9 is the change in symbolic stability from one election to the next. In other words, this variable is the similarity of the current platform to the previous platform minus the similarity of the previous platform to the platform of eight years earlier. Just as with levels of stability, accommodation theory predicts that strong electoral showings should cause a party to be more stable and poor results should cause a party to deviate from its previous symbolic appeals. The independent variable in the first

and third columns is the party’s share of the two-party vote in the previous presidential election. The independent variable in the second and fourth columns is the party’s share of the congressional votes cast in the off-year election between presidential races. Once again, there is little support for accommodation theory to be found. Neither of the electoral variables emerge as significant predictors of change in symbolic stability, which holds both for policy and value symbols. In addition, these models explain virtually none of the variance in the dependent variable.

Table 6.10: Change in Symbolic Imitation as a Function of Previous Electoral

Outcomes

Predictor ∆ Policy Imitation ∆ Value Imitation

coefficients (s.e.)

Presidential Votet−1 .12 −.22

(.26) (.19)

Off-Year Congressional Vote .39 .34

(.32) (.33) Constant −.05 −.17 .12 −.15 (.13) (.15) (.09) (.16) Adjusted r2 −.01 .01 .00 .00 N 72 54 72 54 *=p<.05

Note: Unit of analysis is the change in symbolic imitation and each party’s performance in recent elections. The dependent variable in the first two columns of results in the change in policy-specific symbolic imitation for each party, or the similarity of their current platform to their opponent’s previous platform minus the similarity of their previous platform to their opponent’s platform of eight years earlier. The dependent variable in the last two columns is the same change measure within value symbolism. The independent variable in the first and third columns of results is the party’s two-party vote share in the previous presidential election. The independent variable in the second and fourth columns of results is the party’s vote share in all House races from the off-year congressional election.

Table 6.10 reports the same regressions, this time treating change in imitation as the dependent variable. Just as with change in stability, recent electoral results do not predict whether a party will be more or less inclined to imitate its opponents than it was in the previous election.

The results of this section bolster the conclusion that fluctuations in symbolism are not primarily driven by recent electoral results. Parties that do well are no more likely to be symbolically stable than they were in the previous election and parties that do poorly are no more inclined to imitate their opponents. Even after removing the long-term trend

toward consistency in campaign discourse there is no evidence that parties respond to election results in the short-term ways predicted by accommodation theory.

Symbolic Stability and Imitation Driven by the Same