3. PLAN DE PREVENCIÓN DE RIESGOS LABORALES
3.4. A CTIVIDADES PREVENTIVAS
In order to test the centralized selection hypotheses in this chapter, I begin with a χ2 test to examine the relationship between the dichotomous variables of the analysis and centralized selection. The results of the χ2 tests are reported in Table 2 and show that several of the hypothesized relationships are significant. The variable for left parties is the only variable not significantly related to centralized selection. Therefore there is no evidence in support of H2, which predicted that parties of the left would be less likely to centralize. In order to examine the remaining hypotheses, I constructed a logistic regression model and these results are reported in Table 3.
In addition to the substantive variables of interest, this model includes a variable captur-ing membership in the Eurozone and length of membership in the EU (in years) to account for the depth of a party’s experience in the EU. I also include a few more party level variables to account for each party’s experience with the EU. This includes a measure of the party age (number of years a party has been active) and a measure of the relative size of the EP
23Interview 6.17.2013. European Parliament. Brussels, Belgium
24Interview 5.30.2013. European Parliament. Brussels, Belgium
delegation. Relative size is the proportion of MEPs relative to party politicians at national and EP level (in 2004-2009 term). The average of this term is about 14% but some parties only had members in the EP (UKIP, GERB).
Table 2: χ2 Test of Association for Centralized Selection
Centralized Selection Χ2
Left Party 0.930
National Executive Selection 8.259***
Bicameral Legislature 4.160**
Federalism 3.441*
State Financing 3.441*
*** p<0.01, ** p<0.05, * p<0.1
The results in Table 3show that several of the party and government characteristics are significantly related to centralized selection for the EP.25 Examining the party variables first, the results support H3, executive selection on the national level is a significant, and positive, indicator of centralized selection on the European level. Parties are more likely to remain centralized if they are already structured as such. Party age and the relative size of the EP delegation are also significant and positive indicators of centralized selection. As parties gain more experience with the EU (both in terms of time and personnel in the institutions), they are more likely to centralize their procedures. Table 3 reports the marginal effects of each variable, and although significant, these three effects appear to be relatively small. In order to examine these effects more closely, I calculated predicted probabilities which are available in Table 4.26 These results show these party factors have very little effect on the likelihood of executive selection and the national factors are much more important.
In terms of national characteristics, these results support the relationships between ex-ecutive selection and multipartism and multi-level government as predicted in H4 and H5
25The results reported are the marginal effects with country fixed effects. The results of analysis using country random effects are reported in the Table31in AppendixA. The results remain consistent with the fixed effects model reported in text.
26Predicted probabilities were calculated using Clarify (Tomz, Wittenberg and King 2003). Unless oth-erwise noted, all simulations were conducted with the auxiliary independent variables set at their mean or modal category
Table 3: Logistic Regression for Centralized SelectionMarginal Effects and Executive Selection Relative Size of EP Delegation 0.405***
(0.154) Effective Number of Parties -0.080*
(0.045) Membership (length in years) -0.009**
(0.004)
Observations 45
Wald chi2 297.3
Prob > chi2 0.00
Pseudo R2 0.508
Country clustered standard errors in parentheses Marginal effects reported
*** p<0.01, ** p<0.05, * p<0.1
respectively. Federalism has the largest effect on the likelihood of executive selection but it is negative. Parties in federal systems are almost 17% less likely to centralize their selection procedure than their unitary peers.27 Membership in the EU also decreases this likelihood as the length of membership grows. As membership length moves from its minimum to
27Bicameralism also has a significant relationship with executive selection. While analysis with a χ2 test of association (see Table2) showed that bicameralism is significantly related to centralized selection. I was unable to test the strength or direction of the relationship in the logistic model. Among the parties here, all of the parties using decentralized selection are at least weakly bicameral so there are no decentralized observation in unicameral systems.
maximum the likelihood of executive leadership decreases. This highlights a potentially in-teresting variation between old and new member states with respect to party organization.
The strongest relationship presented by the model is that of Eurozone membership. Coun-tries in the Eurozone are much more likely to centralize selection at the European level. The predicted probabilities show that this difference is close to 80%. The only hypothesis that I did not find support for in the model was that of state financing. This variables is not significant at the 90% level so I cannot conclude that the origin of party financial resources for European elections influences organization in this case.
Overall the model predicts that centralized selection is highly likely under a variety of circumstances. The structure of government institutions has the largest effect on party organization with respect to the selection of their MEPs. This suggest that the experience of parties on the national level is the most important indicator of what parties will do at the European level. National parties appear to be merely replicating the national procedures or centralizing their procedure. The existing consensus in the literature claims that national parties in Europe have been unable or unwilling to seriously consider the role of an MEP in relation to their broader structure. Because of this reluctance to consider how best to deal with MEPs, national parties have not been able to increase the accountability of their politicians who are working at the European level (Ladrech 2007; Poguntke et al. 2007).
While I find no specific adjustment in the majority of parties with respect to their candidate selection procedures, which supports this claim, I find contrasting evidence with respect to accountability. I find that accountability, at least through selection for candidacy, is at least as strong as it is in national elections if not stronger. The tendency to centralize means party leaders have significant control over these MEPs in their placement and removal.