• No results found

3. Preserving Party Unity by Giving Up Control: The Direct Primary in Nineteenth Century

3.5. Case Studies and Theoretical Implications

3.5.3 Focal Candidate

The case studies in the previous two sections suggest that party unity was the foremost concern for county party elites in nineteenth century Pennsylvania. The larger the party vote share, the better the prospects of winning in the general election, and the more fiercely competed for the nominations were. Therefore, the majority party was more likely to suffer from the defection problem than the minority party. But, why would this situation lead to adoption of the direct primary by the leading party? I explain it below using the well-known concept of a focal

point, which was first introduced by Thomas Schelling. According to Schelling (1960), the focal

point is “each person’s expectation of what the other expects him to expect to be expected to do” (p. 57). In other words, it is a way to reach an agreement in the absence of communication, and it works because it seems a natural or relevant choice to the persons in question.

There are reasons to think that the direct primary functioned to prevent defection. Candidates had to publicly announce their candidacy in advance under the primary system, while the full list of nomination seekers was rarely available to voters before or after a county

convention under the delegate system. At county conventions, party leaders and delegates engaged in negotiations. After some deal-making, potential candidates could withdraw before the first balloting, which happened quite often. Candidates did not necessarily have to step forward on their own but could have others nominate them (and pretend that they were not interested). All sorts of things went on without leaving a trace in the recorded official proceedings. Under the delegate system, nomination seekers could pretend that they never had a fair chance at the county convention, and therefore argue that their defection was justified. With the introduction of the direct primary, the full list of candidates was printed in newspapers before the primary day, and the tables of full returns for each candidate were reported in newspapers after the primaries. Party leaders were able to make defection more embarrassing for nomination losers by making the process transparent. Put in another way, the popular vote system enabled party elites to credibly point at who was the focal candidate for each race, and thereby helped line up party voters’ support behind a single contender for each office in the general election. Here is how.

Although, in reality, there were offices to which more than one person was to be elected at the same time, we assume for now that only one person was to be elected per office. We also assume the U.S.-style two-party system, which did apply in most counties in Pennsylvania. When a defector X from the majority party—let us say the Republican Party for now—can freely run in the general election, it becomes a three-way race. What happens in such a race depends

critically on the expectation of the Republican voters. If, for any reason, expectations form among some Republican voters to the effect that X has a better chance of winning against the Democratic candidate D than the official Republican nominee Y, then these voters will abandon Y and support X. The logic follows from Duverger’s law, and this could lead to a disastrous outcome for the Republican Party.

In other words, in the event of defection, the Republican voters face a coordination problem: they must coordinate among themselves and rally behind either X or Y to win in the general election. But how should they decide? In modern politics, public opinion polls may play a critical role in such cases, by allowing party voters to coordinate on the more promising candidate. What about in the nineteenth century, when no opinion polls were available and the level of information was low? The Republican defector X could easily manipulate some party voters into believing that he was more competitive electorally than Y against D. If the process of intra-party coordination was incomplete, so that Republican voters could split their votes between X and Y, then D might actually win the election unless the total Republican vote share was more than two-thirds among the constituency. If there were more than one defector, a coordination failure among the Republican voters might still allow D to win, even if the Republican share of the electorate was greater than two-thirds.

one of the contenders, Y, as the focal candidate, so that all Republican voters would see the person as the obvious “leading contender” to beat D. Running in the general election would then be self-defeating for X, as the Republican voters would coordinate on Y. And how might Republican leaders establish Y as the focal candidate? This could be accomplished through the use of the direct primary. The winner of the popular vote would have proved his popularity among voters. (The results were much more credible than the outcomes of deal-making at a delegate convention). If X decided to defect and run in the general election after losing the primary or without competing in the primary, voters would know it. Party voters have a strong incentive to rally around the primary winner. Winning the primary thus becomes the equivalent of leading in the polls.

Party leaders introduced the direct primary in order to establish someone as the focal candidate in the minds of Republican voters rather than to select the most popular or electable candidate. The introduction of the direct primary made defection very unattractive and had the effect of preventing it to a certain extent. It was a way to induce party unity and hence assure a victory over the opposite party. Therefore, the attractiveness of the direct primary increased as the party’s vote share increased. The same logic applied to the Democratic Party as well as to the Republican Party. Ironically, the stronger the party, the more likely party leaders were to give up control over the nomination.

One thing to note here is that the ideological or issue dimention in terms of calculation by party elites or intra-party schisms is missing in the logic of the focal candidate. In local newspapers, only sectional interests within a county (in terms of urban/rural enmity) and personal feuds were addressed as the background for factionalism and potential defections. It may have been the case that, at the county level, the policy position was not what intra-party conflicts centered around. Or, there might have been some ideological element in the factional strife, but I could collect no data to trace its impact.

3.6. Statistical Analysis

If the focal candidate hypothesis is correct, the party vote share and the probability of adoption of the direct primary should be positively correlated. In this final section of the paper, I test this hypothesis through statistical analysis. Below, I start by discussing dependent and independent variables used in the regressions. Next, I estimate simple logit regressions using raw yearly data. In order to test the robustness of the findings, I will run ordinary least squares (OLS) regressions using the 5-year moving average of all variables, which is expected to ameliorate the problem presented by the missing data.

The unit of analysis is county-party-year, assuming that each county party had a chance to choose CSM each year. The dependent variable is a dichotomous dummy variable,