Allocating Campaign Effort in Spain: Evidence from Four General Elections

Abstract

This paper analyses which districts are targeted by Spanish political parties in their electoral campaigns. We find that the major Spanish parties - PP and PSOE - mobilise districts where they are more likely to win a new seat or are in danger of losing one they already hold. The predicted closeness of the district race is more relevant in the smallest districts. We also find that Spanish parties mobilise their strongholds. We suggest that, apart from the pure office-seeking strategies, political finance motivations might also play a role in the mobilisation choices made by Spanish party elites.

Keywords:

• Election campaigns
• proportional representation systems
• swing and core districts
• PP
• PSOE

Acknowledgments

We would like to thank the two reviewers of this article, Joaquín Artés, Alfonso Echazarra, Sandra León y Rubén Ruiz-Rufino for their extremely detailed and useful comments. We would also like to acknowledge the Spanish Ministry of Economy and Competitiveness for its financial support through the project ‘Comportamiento electoral y políticas públicas’ [Reference: CSO2013-40870-R].

Notes

1. We drop these two districts from the analysis because parties pay little attention to them in their campaigns, as they are frequently safe districts.

2. To be more specific, IU and Union, Progress and Democracy (Unión, Progreso y Democracia – UPyD) are excluded from the analyses because the former was almost irrelevant in the 2004 and 2008 elections (less than six seats in each of them) and the latter only ran in the 2008 and 2011 elections. Moreover, since our dependent variable uses newspaper information, as explained below, including these two small parties could bias the analysis because it is likely that the newspapers do not cover all the campaign activities conducted by these parties.

3. Empirical evidence from post-election surveys supports this point because the proportion of voters who declared having been contacted by parties does not significantly differ across districts.

4. We have also used different weights, such as E_w´ = 0.67 × R + 0.33 × other events. Results are similar and are available upon request.

5. In the four examined elections, the correlation between predicted closeness and actual closeness by district was + 0.58. This shows that contemporary information from the CIS (Centro de Investigaciones Sociológicas – Centre for Sociological Research) pre-electoral polls predicts well enough how close the election in each district will be.

6. Results are robust to alternative specifications of the party strength variable which draw on further previous elections (such as the average vote shares in the last two or three general elections). The use of these alternative measures allows us to believe we do not have an endogeneity problem in our estimations. These results are available upon request. We are confident that our party strength variable is a good proxy of the influence of the local party branch in the national organisation in the case of the PP because the number of delegates that the local branches send to the National Convention depends on their previous general election results. In the case of the PSOE, the number of delegates that the local branches send to the National Convention is not a direct function of their previous general election results. However, the distribution of delegates between provinces still correlates (+ 0.33 in the 2012 Party Convention) with the vote share in the previous election.

7. District magnitude is relevant not only because the district allocates more seats, but also because in larger districts strategic voting will be less likely (Lago 2008; García-Viñuela, Artés & Jurado 2015).

8. Incumbent takes value 1 for the PP in 2000 and 2004 elections (0 otherwise), and value 1 for the PSOE in the 2008 and 2011 elections (0 otherwise).

9. Although formally dependent on the Ministerio de la Presidencia (Ministry of Presidency) of the Spanish government, the CIS is an independent agency, with its own legal status and funding. Its purpose is to conduct scientific studies of Spanish society.

10. This is the reason why the sample size of the pre-electoral surveys is four times larger than the size of the post-election surveys. Obviously, the CIS survey is one of several surveys conducted these days. We are not claiming that parties consider only it when allocating their mobilisation effort. What we claim is that the CIS survey captures important information that determines parties’ strategies.

11. All the mentioned robustness tests are available upon request. Results are also robust to bootstrapping.

12. We have re-run the analyses using the percentage of abstainers in the past election as an independent variable and have found the same null results.

13. The party models are also robust in two additional ways. As the decisions on where to organise an event or a rally can be interdependent between parties, we have first of all included the number of rallies or events of one party as an independent variable in the other party’s model. Secondly, we have estimated the number of events and rallies for each party simultaneously with seemingly unrelated regressions. The results of these exercises are available upon request.