Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

Parliamentary Behavior with Two Principals: Preferences, Parties, and Voting in the European Parliament Author(s): Simon Hix Source: American Journal of Political Science, Vol. 46, No. 3 (Jul., 2002), pp. 688-698 Published by: Midwest Political Science Association Stable URL: http://www.jstor.org/stable/3088408 Accessed: 04/09/2008 05:57 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showpublisher?publishercode=mpsa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. http://www.jstor.org

Parliamentary Behavior with Two Principals: Preferences, Parties, and Voting in the European Parliament Simon Hix London School of Economics and Political Science The European Parliament has become one of the most powerful institutions in the European Union. Members of the European Parliament (MEPs) can now enact legislation, amend the European Union budget, veto the nominee for the European Union Commission President, and censure the Commission. But, we know little about what determines MEPs' voting behavior. Do they vote according to their personal policy preferences? Do the EP parties force MEPs to toe the party line? And, when national party and EP party preferences conflict, which way do MEPs respond-to the principals who control their election (the national parties) or the principals who control their influence in the EP (the EP parties)? The results reported here show that national party policies are the strongest predictors of voting behavior in the EP. 688 TH Tow Members of the European Parliament vote is increasingly important. As a result of several reforms of the European Union Treaties,' MEPs can now enact legislation in a wide range of policy areas, amend most lines in the budget, veto the governments' nominee for the Commission President, and censure the Commission. But, we know little about what determines MEPs' voting behavior. Do they vote according to their personal policy preferences? Do the political groups in the EP force them to vote along European party lines? Or are MEPs beholden to their national parties? Legislative behavior in the EP is also interesting from a more general perspective. The EP is a relatively new chamber, with a high number of political parties, a variety of decision rules, and multiple influences operating upon its members such as diverse policy preferences, national interests, national party policies, and European party affiliations. For example, when deciding how to vote, MEPs must respond to two different "principals": national parties, who control the selection of candidates in EP elections, and the political groups in the EP, who control a variety of private goods in the EP, such as leadership positions, committee assignments, speaking time, and the legislative agenda. These pressures are similar to those faced by U.S. Representatives, between constituency and legislative caucus interests, and by legislators in parliamentary systems, between local parties and Simon Hix is Reader in European Union Politics and Policy, London School of Economics and Political Science, Houghton Street, London WC2A 2AE, United Kingdom (s.hix@lse.ac.uk). Earlier versions of this article were presented at a workshop at the University of Washington, in Seattle, and at the 2001 European Community Studies Association conference in Madison, Wisconsin. I would like to thank Alison Alter, James Caporaso, Fabio Franchino, Simon Hug, Simon Jackman, Joe Jupille, Keith Krehbiel, Roger Scully, Kaare Strom, Gerald Schneider, Diego Varela and the three anonymous referees for their comments on the article. The research for this article was funded under the "One Europe or Several?" program of the Economic and Social Research Council of the United Kingdom (Grant#: L213 25 2019), and by the Nuffield Foundation (Grant# SGS/00387/G), and the Leverhulme Trust (Grant# RG/8/2000/0042). 'In 1987, the Single European Act gave the EP a right to two readings of legislation relating to the EU single market. In 1993, the Maastricht Treaty established the so-called "codecision procedure," which gave the EP a veto on most socio-economic legislation. Finally, in 1999, the Amsterdam Treaty reformed the codecision procedure, giving the EP equal legislative power with the Council, and gave the EP a veto in the selection of the Commission President. See, inter alia, Hix 1999a (56-98). American Journal of Political Science, Vol. 46, No. 3, July 2002, Pp. 688-698?2002 by the Midwest Political Science Association ISSN 0092-5853

PARLIAMENTARY BEHAVIOR WITH TWO PRINCIPALS parliamentary factions. Hence, understanding how the battle between the two principals is resolved in the EP sheds light on the general issue of legislative behavior with multiple principals. To tackle these issues, I outline three explanations of MEP voting behavior: personal ideological beliefs, European party discipline, or national party discipline. These explanations are then tested by comparing MEPs' policy preferences with their legislative behavior. I use a survey of MEPs to measure their "posited" policy preferences and then apply NOMINATE to over 1,000 roll-call votes to measure their "revealed" policy locations. The Question: "Where's the Party?" in the European Parliament A central claim in research on the EP is that the political groups have a significant impact on MEP voting behav- ior. These legislative parties enable MEPs to overcome a variety of collective action problems. By organizing a 689 division of labor with like-minded legislators, MEPs can secure access to the legislative agenda, resources, and committee assignments. In return, MEPs follow their EP party leaders' instructions on how to vote. The main evidence for these claims is the growing level of intra-party voting likeness, or "party cohesion" (e.g., Attina 1990; Brzinski 1995; Raunio 1997, 79-124), as Table 1 shows. But, it may be premature to conclude from these crude measures that the EP parties are the main determining forces of MEP voting. In the U.S. context, for example, several scholars claim that legislative parties are the main forces behind voting behavior in the House and Senate (e.g., Rohde 1991; Aldrich 1995; McCarty, Poole, and Rosenthal 2001). Krehbiel has recently questioned the logic behind these claims: "A common form of inference pertaining to the significance of parties in politics goes as follows. A phenomenon is important. Party is correlated with it. Therefore, party is important-moreover, by implication, in a causal way" (1993, 237). Consider roll-call voting. If a high percentage of legislators in the same party vote the same way in a series of votes, then the party organization is assumed to have produced TABLE I 'Cohesion' of the Party Groups in the European Parliament Percent of MEPs Voting Cohesion ('Index of Agreement') Party Groups, from Left to Right EP2 EP3 EP4 EP5 EP2 EP3 EP4 EP5 (group abbreviations in brackets) (1984) (1989) (1994) (1999) (1984-89) (1989-94) (1994-99) (1999-04) Communists (LU) - 2.7 - - -93.8 Radical Left (EUL, EUL/NGL) 9.4 5.4 4.9 6.7 71.2 92.3 83.8 61.4 Greens (G) -5.8 4.1 - -87.5 85.2 Regionalists / Greens & Regionalists 4.6 2.4 3.4 7.7 67.8 69.5 98.0 75.4 (ERA, G/EFA) Socialists & Social Democrats (PES) 30.0 34.7 34.9 28.8 62.2 78.6 89.0 84.8 Liberals (ELDR) 7.1 9.5 7.6 8.1 69.5 85.7 80.1 87.8 Christian Democrats & Conservatives 25.3 23.4 27.7 37.2 84.1 88.2 90.2 81.1 (EPP) Nationalist-Conservatives 6.7 3.9 9.4 4.8 75.7 64.5 93.2 58.3 (EDA, FE, UPE, UEN) British Conservatives (EDG) 11.5 6.6-82.9 92.2 Anti-Europeans (EN, EDD) 3.4 2.6 70.7 40.5 European Right (ER) 3.7 3.3 96.1 99.9 Non-attached MEPs 1.6 2.3 4.8 4.2 Source: Calculated from data in Attina (1990), Raunio (1996), Hix and Lord (1997) and Hix (2000). The scores are from a sample of roll-call votes in each parliament. Collated roll-call data is not currently available from the first directly-elected EP-the 1979-1984 parliament. Note: An Index of Agreement is a measure of the relationship between the three modalities of votes cast by the members of a party group ('yes', 'no', and 'abstain'), in relation to the total number of votes cast by the members of the group. Formally: highest modality - sum of the other two modalities IA = x 100 total number of votes cast by the party group In other words, the index is equal to 100 if all the members of the group vote the same way (in every vote), and is equal to 0 if exactly half the members of a group vote one way and the rest vote another way (in every vote).

690 SIMON HIX this cohesion. But, what if these legislators would have voted the same way anyway because they share the same preferences on the subjects of these votes? The effect of party organization on legislative behavior, independent of legislators' preferences, cannot be determined. This issue is further complicated in EP because of the existence of two principals: EP parties and national parties. So, when an EP party shows a high level of cohesion, we do not know whether this is because the EP party leaders have forced their MEPs to vote together, whether the national member parties in the EP party have policy positions that correlate with those of the EP party (and so have instructed their MEPs to vote with the EP party), or whether the individual MEPs in the EP group share the same policy positions, independently of their national party affiliations. Three Explanations of MEP Voting Behavior Consider three explanations of voting in the EP. Personal Policy Preferences If MEPs vote purely on the basis of their personal ideo- logical beliefs, the distance between an MEP's ideal policy position and the average policy position of the MEP's EP party should be a stronger predictor of the number of times the MEP votes for/against the EP party. If an MEP has an ideal policy position close to the mean position of the EP party, the MEP is unlikely to be torn between vot- ing with his/her personal beliefs and following the voting instructions of the EP party leadership. Conversely, if an MEP has an ideal policy position far from the mean position of the EP party, he/she is more likely to be on the opposite side of a cutting-point (the point that divides the "yes" camp from the "no" camp in a vote) to the ma- jority of the EP party. Whether an MEP is to the left or the right of the mean member of the party should not make a difference. Intuitively, one might expect MEPs on the right of the main center-left party (the Party of European Socialists) and on the left of the main center-right party (the European People's Party) to be more likely to defect from their parties than MEPs on the opposite wings of both parties. If most status quo points are centrally located, in comparisons between the status quo and an alternative proposed by one of the two main parties, the MEPs located between the two parties are more likely to be caught on the "wrong side" in a vote. This is not the case in the EP. First, there are several other parties in the EP, in addition to the two main party groups. There are two parties to the left of the Party of European Socialists (the Greens/European Free Alliance and the European United Left/Nordic Green Left), one party between the Party of European Socialists and European People's Party (the European Liberal, Democrat and Reform Party), and two parties to the right of the European People's Party (the Union for a Europe of Nations and the Group for a Europe of Democracies and Diversities). Hence, unless an MEP is located on the far right of the most right-wing party in the EP or the far left of the most left-wing party in the EP, the MEP will be situated between the ideal points of two parties. Second, formal models of European Union bargaining often assume that the status quo is at one extreme of the decision space (e.g., Garrett 1992; Tsebelis 1994). However, since "no European Union legislation" means maintaining a patchwork of national regulations, the status-quo could be located anywhere in the European Union decision-space. Because of this randomness of the location of the status quo, votes in the EP are rarely simple left-versus-right splits. As Figure 1 shows, the vote-splits between July 1999 and June 2000 were evenly distributed between 95 percent on the left versus 5 percent on the right and 5 percent on the left versus 95 percent on the right (apart from a large number of near unanimous votes). In other words, regardless of their position on the left-right spectrum, MEPs are equally likely to find themselves on the other side of a cut-point from their EP parties. European Party Discipline If the EP parties are powerful legislative organizations, they should be able to impose sanctions on MEPs who defect from the "party line." Thus, regardless of an individual's ideological location, MEPs should follow the voting instructions of the leadership of that EP party. Hence, the "revealed" policy location of an MEP-where they are located in their voting behavior-should be close to the ideal policy position of their EP party. Furthermore, some EP parties are more likely to be able to enforce their wishes on their member than others. The two main parties in the EP-the European People's Party and the Party of European Socialists-effectively run the EP. With more than 65 percent of the MEPs between them, when working together these two parties control who wins the key offices in the EP, who gets what committee chair, and how the legislative agenda in committees and on the floor of the EP is organized (especially Raunio 1997 and Kreppel 2001). Hence, these two parties

PARLIAMENTARY BEHAVIOR WITH TWO PRINCIPALS 691 FIGURE 1 Distribution of Vote-Splits in the European Parliament 140 120 - o o C (- c CO 0) 100-80- a) (D) 0 0 0 O, 60-40- z 20-0 0-9.9 15-19.9 25-29.9 35-39.9 45-49.9 55-59.9 65-69.9 75-79.9 85-89.9 95-100 10-14.9 20-24.9 30-34.9 40-44.9 50-54.9 60-64.9 70-74.9 80-84.9 90-94.9 Percent of MEPs voting with the left have a lot to offer MEPs who behave, and by controlling the levers of power in the EP, these two parties have more sanctions than the other EP parties. National Party Discipline Because of the way EP elections are organized, national parties have a strong influence on the reelection prospects of MEPs. In all European Union states, national party leaderships have some control over candidate selection in EP elections-either via the national party executive determining the list of candidates (as in France, the Netherlands, Belgium, Luxembourg, Greece, Spain, Portugal, Sweden, Denmark, Finland or Austria), or via the national party executive approving candidates selected by regional organs (as in Britain, Germany, Italy, or Ireland). Moreover, once elected to the EP, most national parties hold meetings of the "national party delegations" of their MEPs. These meetings elect the officers of the national party delegation and determine the national party's positions in the EP. The leader of a national delegation often sits on the party's executive committee at home, and so influences national party policy on European Union issues. National party positions, as articulated by these national delegations, often conflict with the position of the parent EP party. Some national delegations, particularly the larger ones, issue voting instructions to their MEPs when the national party position is different from the EP group position. Because the national party, rather than the EP party, controls whether an MEP can stand in the next election, when there is such a conflict, the MEPs can be expected to vote with their national party against their EP party. So, the further the ideal policy position of a national party is from the ideal position of the EP party, the more likely the MEPs from that national party will find themselves facing a choice between their national party and their EP party. And, as discussed before, whether a national party is to the right or the left of the EP party should not have a systematic effect. Hence, the distance between a national party and the EP party should be a strong predictor of MEP defection from the EP parties. However, the policy positions of the EP parties and the national parties are not completely independent. The leadership of each EP party is composed of the leaders of

692 SIMON HIX the larger national delegations in that EP party. Thus, some national parties are more influential than others when setting the policy positions of the EP parties. Hence, there is likely to be less policy conflict for the MEPs from these national parties. For example, in the European People's Party in the 1999-2004 parliament, the German Christian Democratic Union (CDU) is the largest national delegation in the European People's Party and provides the Presidency and Secretary-General of the Euro- pean People's Party. Similarly, in the Party of European Socialists in the 1999-2004 parliament, the German Social Democratic Party (SPD) is the largest national delegation and the Spanish Socialist Workers Party (PSOE) provides the President. FIGURE 2 2a. Left-Right a> c) Q) C.3 - Kernel Density Plot of Ideological Self-Placement of PES and EPP Survey Respondents PES Measurement o - 1 Left-Right self-placement (1 = left, 10 = right) 10 To test these explanations we need measures of the posited and revealed positions of the MEPs. 2b. European Integration.3- Left-Right and Pro/Anti-Europe Self-Placement There are two main ideological dimensions in European Union politics-the traditional left-right dimension and the pro/anti-europe dimension (concerning the speed of European integration and how the European Union institutions should be designed; Hix 1999b). Where an MEP is located on these two dimensions should give a good indication of their underlying ideological preferences on issues on the EP agenda. To measure these basic policy locations of the MEPs, I use the results of a survey of MEPs, which was conducted by the European Parliament Research Group in September 1999 via a written questionnaire. The questionnaire was completed by 195 MEPs (31 percent of the 626 members), and these returns constitute a good sample of the total population of MEPs (see the appendix). The questionnaires were numbered so that we know the identity of the individual MEPs who responded to the survey. The questionnaire included a set of questions asking the MEPs to locate themselves on several policy scales. One such question was the standard "left-right self- placement" question: "Where would you place yourself on the Left-Right spectrum?," and the respondents were given a scale from 1 to 10, where 1 was labeled "Left" and 10 was labeled "Right." Another question asked about attitudes towards European integration: "Where would you place yourself on the question of European integration?," and the respondents were given a scale from 1 to 10, where 1 was labeled "European integration has gone to a a) 0 - EPP PES 1 10 European Integration self-placement (1 = anti, 10 = pro) far" and 10 was labeled "The EU should become a federal state immediately." Figure 2 shows the kernel density plots of the responses of the sixty-one Party of European Socialist and seventy-two European People's Party MEPs who answered these questions.2 There is considerable overlap in the ideal policy preferences of the MEPs in the two main EP parties. This suggests that if MEP's vote in the EP purely on the basis of their underlying preferences, many members of these parties will find themselves closer to the ideal position of the other EP party than their own. 'These kernel density plots were generated using the "kdensity" command in Stata 7.0. A kernel density plot is in effect a "smoothed" histogram, where instead of plotting a series of bars, the figures plot probability distributions of being located at a particular point on a policy scale-by calculating the likelihood that an MEP will fall at each point on the scale by averaging the likelihood that a member will fall in the intervals surrounding the scale.

PARLIAMENTARY BEHAVIOR WITH TWO PRINCIPALS NOMINATE Scores To establish the revealed behavior of the MEPs, I collected all of the roll-call votes in the first year of the fifth directly elected EP (N=1,031). These votes covered a range of issues on the European Union agenda: regulatory policies, such as environmental, consumer, and social standards (N=384); inter-institutional relations, such the reform of the European Union Treaties (N=232); ex- penditure policies, such as the agricultural and regional funds (N=139); external relations, including security and trade issues (N=1 17); justice and interior affairs issues (N=106); and general economic issues, such as labor market reform, Economic and Monetary Union and tax harmonization (N=53).3 One must be cautious when using roll-call votes to generalize about voting behavior in the EP. Under the EP's rules of procedure, only certain votes must be taken by roll-call. All other votes are either by a show of hands or by "electronic vote," and in either case, how each individual MEP votes is not recorded. Rule 134 of the EP's rules of procedure specifies that any EP party or thirtytwo MEPs can request a roll-call. In practice, roll-call votes tend to be called by EP parties or a large national delegation, either to show their position to the public on an issue, to embarrass another party, or to keep a check on their own members' behavior (Corbett, Jacobs, and Shackleton 2000, 149-150). In the period under investigation, roll-call votes constituted approximately onethird of all votes. Consequently, roll-call votes do not tell the complete story of EP voting, and the outcome of these votes may be systematically biased (Carrubba and Gabel 1999). Nevertheless, roll-call votes are the only votes that we can study in detail. Roll-call votes, at least in this period, covered a broad range of issues on the European Union agenda. And, roll-call votes do not appear to be called disproportionately by one EP party or another. Hence, without empirical evidence to prove that roll-call voting is systematically biased towards a particular EP party or set of issues, it is reasonable to assume that these votes should produce a fairly accurate picture of voting behavior in the EP. I analyzed the roll calls using the NOMINATE scaling method (Poole and Rosenthal 1997, 233-251). Despite some of its limitations (see Clinton et al. 2001), the NOMINATE algorithm has rapidly become the standard method for generating ideal point estimates from rollcall voting behavior. The underlying assumption of NOMINATE is that legislative behavior can be simplified 3Previous analysis of EP roll-call voting reveals little systematic difference in how MEPs vote in these different policy areas (Hix 2001; Noury 2002). into a small number of dimensions and that each legislator can be represented by an ideal point in this space. The output of NOMINATE is a set of Cartesian coordinates in the unit space for each legislator. Applying NOMI- NATE to the 1,031 votes cast between July 1999 and June 2000 produces a predominantly one-dimensional space: a single dimension effectively explains about 85 percent of all vote decisions (Hix 2001; cf. Noury 2002).4 There is a large, significant, and positive correlation between the positions of the MEPs on the left-right selfplacement scale in the survey and the location of these same MEPs on the dimension produced by NOMINATE (.749). There is also a significant relationship, although smaller and negative, between the positions of the MEPs on the European integration self-placement scale in the survey and the dimension produced by NOMINATE (-.211). In other words, the dimension produced by NOMINATE from the voting records is most similar to the traditional left-right dimension, but also captures some positions on the other main policy dimension in European Union politics, relating to the speed of European integration. The Variables and Hypotheses 693 The dependent variable, the propensity of an MEP to vote against their EP party, is the absolute distance between an MEP's NOMINATE score and the mean NOMINATE score of all the MEPs in their EP party. This is a continuous variable, ranging from 0 to 2. The independent variables are specified as follows. First, the ideological distance between an MEP and the MEP's EP party is defined in two ways: (1) the distance between an MEP's "posited" left-right location and the mean "posited" left-right location of all the members of the MEP's EP party (using the left-right self-placement question in the survey), which is a continuous variable with ranging from 0 to 9; and (2) the distance between an MEP's "posited" European integration location and the mean "posited" European integration location of all the members of the MEP's EP party (using the European integration self-placement question in the survey), which is a continuous variable with ranging from 0 to 9. 4NOMINATE requires votes to be set-up in a binomial fashionwith only "yes" and "no" positions. In the EP, however, there are three voting alternatives-"yes," "no," and "abstain." Where voting is by a simple majority, abstaining is the equivalent of not participating in the vote. But, where voting is by an absolute majority, abstaining is the same as voting "no," as any vote that is not a "yes" undermines the ability of securing the required 314 "yes" votes for the act to be passed. Hence, were a simple majority was required (886 votes in the analysis) abstentions were excluded, and where an absolute majority was required (145 votes) abstentions were coded as "no" votes.

694 SIMON HIX Second, EP party membership (for the European People's Party, Party of European Socialists, Greens/European Free Alliance and Union for a Europe of Nations) is a dummy variable, coded 1 if the MEP is a member of the EP party in question and 0 otherwise. Third, the distance between national party preferences and EP party preferences is defined in two ways: (1) the distance between the mean posited left-right location of the members of an MEP's national party and the mean posited left-right location of all members of the MEP's EP party (on the left-right self-placement question in the survey), which is a continuous variable ranging from 0 to 9; and (2) the distance between the mean posited European integration location of the members of an MEP's national party and the mean posited European integration location of the members of the MEP's EP party (on the European integration self-placement question in the survey), which is a continuous variable ranging from 0 to 9. Fourth, national party membership, for the nine largest national party delegations in the EP, are dummy variables, coded 1 if the MEP is a member of the national party in question and 0 otherwise. The correlations between the independent variables are shown in Table 2. The three explanations of EP voting have different predictions about the size, significance, and sign of the effects (beta-coefficients) of these independent variables on the dependent variable: Hypothesis 1 (MEP ideology): the coefficients relating to the measures of individual MEP preferences should be significant and positive (in other words, the MEP should be more likely to defect from the EP party than the baseline group of MEPs), and the coefficients relating to national party preferences should not be significant. Hypothesis 2 (EP party organization): coefficients relating to the measures of individual MEP preferences should be significant but not large; the coefficients relating to national party preferences should not be significant; the coefficients relating to the two main EP parties should be significant and negative (in other words, the MEP should be less likely to defect from the EP party than the baseline group of MEPs); and the coefficients relating to the two smaller EP parties should be significant and positive. Hypothesis 3 (National party policy and power): the coefficients relating to the measures of individual MEP preferences should not be significant; the coefficients relating to national party preferences should be significant and positive; the coefficients relating to the CDU, SPD, and PSOE should be significant and negative; and the coefficients relating to the other large national parties should be significant, but could be either negative or positive (in other words, the national party either in- structs their MEPs to vote more with the EP group than the baseline group of MEPs, or less with the EP group than the baseline group of MEPs). Results Table 3 shows the results of the regression model. With the thirteen dummy variables, the baseline are the fortyfour MEPs (who responded to the survey) who are not members of either the four EP parties or the nine national parties analyzed.5 The results confirm the view that voting in the EP is driven by national party preferences and membership. First, national parties' policy locations are more influential than individual MEP locations for explaining MEP defection from the EP parties. Both measures of the ideo- logical position of national parties vis-a-vis the EP parties are significant-although ideological variance from the EP party on left-right issues is a stronger predictor of MEP defection than ideological variance on European integration issues. The combined effect of the two measures of national party policy preferences is also more significant than the combined effects of individual MEP preferences or EP group membership. Second, which national party an MEP belongs to also makes a difference. MEPs from the three national parties who control the European People's Party and Party of European Socialists (the CDU, SPD, and PSOE) are less likely to defect than the baseline group of MEPs (although this effect is only significant for SPD MEPs). And, where the other large national parties are concerned, membership of the British Conservative Party dramatically increases the likelihood that an MEP will vote against the EP party, even when the policy variance of the British Conservative members from the European People's Party is taken into account. Some British Conservatives have personal preferences close to the European People's Party mean. But, the average British Conservative is considerably less pro-european and more to the right of the average European People's Party member. As a result, regardless of a British Conservative MEP's personal preferences, the fact that their national party is a preference-outlier in the European People's Party means that all Conservative MEPs have a high propensity to vote against the European People's Party. The results clearly refute the argument that individual MEP preferences drives voting behavior in the EP. The individual ideological position of MEPs compared to 5The members of the nine national parties in the model are all members of either the EPP or PES party groups.

TABLE 2 Correlations Between the Independent Variables 0. 8. c co8 c :o w GesE.S cw X X X O, ED. X - -?e Aln L.." o -. o.... Dsac 1 bewe ME &u E.o o ol r -a Distance. =o c.9 e - Pt betwen.4 a, ;=. e pary0e-inegrtiu. :- -- l Lua.a L LU L C =. = m co cl 'cl mlc 0 C = o. 0._ O LLU LU a. LU )J i LU " 0 Distance between MEP & EP Party of European Socialists -.179* -.230** -.516** - Greens/European Free Alliance -.077 -.136 -.204** -.180* - Union for a Europe of Nations.062.125 -.124 -.109 -.043 - Distance between nat. party.553**.070.217**-.350** -.160*.152* & EP party left-right location Distance between nat. party & EP party EU integr. location.027.607**.193**-.345** -.071.151*.169* Christian Democratic Union (CDU)- Germany -.009.131.335**-.173* -.068 -.042 -.134.009 Social Democratic Party -.046 -.037 -.158*.307** -.055 -.034 -.126 -.242** -.053 (SPD)- Germany Socialist Workers Party (PSOE) - Spain.002 -.058 -.136.264** -.048 -.029 -.099 -.090 -.046 -.037 Popular Party (PP) - Spain Socialist Party (PS)- France Go Italy (Fl)- Italy Democrats of the Left (DS)- Italy Conservative Party - United Kingdom Labour Party - United Kingdom Notes: * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). -.041 -.054.252** -.130 -.052 -.031 -.111 -.112 -.049 -.040 -.034 -.103 -.136 -.187**.362** -.065 -.040 -.086 -.164* -.063 -.051 -.044 -.047 - -.050.045.212** -.109 -.043 -.026 -.076 -.016 -.042 -.034 -.029 -.031 -.040 -.030 -.097 -.158*.307** -.055 -.034 -.168* -.164* -.053 -.043 -.037 -.040 -.051 -.0.097.348**.366** -.209** -.083 -.050.180*.443** -.079 -.064 -.055 -.060 -.076 -.0.023 -.042 -.136.264** -.048 -.029 -.099 -.107 -.046 -.037 -.032 -.034 -.044 -.0

696 SIMON HIX TABLE 3 Regression Models of MEP Voting Behavior Dependent variable = absolute distance between an MEP's NOMINATE score and the mean NOMINATE score of the MEP's EP group Unstandardized Coefficients Standardized Independent Variables Beta Std. error beta coefficient Constant.012.016 Individual MEP Policy Preferences Distance between MEP & EP party left-right location.008.008.063 Distance between MEP & EP party EU-integration location -.008.005 -.103 EP Party Organization European People's Party.024.015.116 Party of European Socialists.041*.017.192 Greens/European Free Alliance -.006.021 -.014 Union for a Europe of Nations -.019.031 -.030 National Party Policy Preferences Distance between national party & EP party left-right location.037'*.011.218 Distance between national party & EP party EU-integration location.014*.007.139 National Party Organization Christian Democratic Union (CDU)- Germany -.040.023 -.098 Social Democratic Party (SPD)- Germany -.056*.028 -.112 Socialist Workers Party (PSOE) - Spain -.056.030 -.097 Popular Party (PP)- Spain -.003.028 -.006 Socialist Party (PS) - France -.047*.024 -.110 Go Italy (Fl)- Italy.009.032.015 Democrats of the Left (DS)- Italy -.046.027 -.091 Conservative Party - United Kingdom.209**.021.592 Labour Party - United Kingdom -.027.030 -.046 Adjusted R-squared.586 F-statistic 17.13 N 195 Joint significance tests for the clusters of independent variables: F p > F Individual MEP policy preferences (F(2,177)) 1.62.2013 EP party organization (F(4,177)) 2.29.0615 National party policy preferences (F(2,177)) 7.79.0006 National party organization (F(9,177)) 14.18.0000 Notes: OLS estimates, **p <.01 (two-tailed test), *p <.05(two-tailed test). their party is not a strong predictor of MEPs' voting behavior, either for policy preferences on left-right or European integration issues. The results also refute the view that voting in the EP is driven by the EP party groups. Of the two larger party groups, only membership of the Party of European Socialists is significant. But, this relationship is in the opposite direction than predicted: members of the Party of European Socialists are more likely to defect from their EP party than the baseline group of MEPs. Conclusions These results suggest that the main factors behind voting in the EP are the policy positions of national parties. Despite the fact that the parliamentary principals in the EP control important benefits-such as committee assignments and speaking time-it is the principals that control candidate selection (the national parties) who ultimately determine how MEPs behave. When the national parties in the same parliamentary group decide to vote together, the EP parties look highly cohesive. But when these parties take opposing policy positions, the cohesion of the EP parties break down. This result consequently has implications for understanding the constraints on a genuinely "transnational" party system in the European Union. The EP parties will only be able to organize voting behavior in the EP if national parties no longer control "their" MEPs. To break national party control, EP elections would have to be held under a system which limited the power of national parties to control candidate selection; for example, with candidates chosen by regional rather than central party

PARLIAMENTARY BEHAVIOR WITH TWO PRINCIPALS organs, and a percentage of MEPs chosen by the EP party leaderships on European-wide "top-up" lists. This would dramatically increase the authority of the party leaders in the EP, who would become key actors in the European Union policy process. Whether this is a good thing for European Union democracy depends on one's view of the EP party leaders. Finally, the results have relevance to the general issue of how legislative agents behave when they have more than one principal. Without centralized candidate selection, parliamentary parties have no way of punishing legislative defection or rewarding allegiance. In most parliaments in Europe, party cohesion is ensured through the centralization of candidate selection in the hands of the parliamentary leadership. But, where candidate selection is decentralized, legislative cohesion breaks down. In the U.S. House of Representatives, for example, where candidates are chosen in district primaries, Members of Congress are more beholden to the median voter in their district than the leader of their party caucus in the House. Similarly, in Italy, where broad electoral alliances are formed between multiple parties, each party within an alliance chooses their candidates in the constituencies they are assigned. As a result, once the elections are over, these alliances fragment into multiple parliamentary parties. 697 Appendix Representativeness of the MEP Survey European Party Group European People's Party Party of European Socialists European Liberal, Democrat and Reform Party Greens/European Free Alliance European United Left/Nordic Green Left Union for a Europe of Nations Europe of Democracies and Diversities Non-attached European Union Member State Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom Gender Men Women Seniority First elected in the 1999 elections First elected before the 1999 elections Total Number Percent Actual Survey Actual Survey MEPs Respondents MEPs Respondents 232 72 180 61 52 20 48 13 42 14 30 5 16 5 26 5 Correlation =.994, t = 2.798 21 2 25 6 16 7 16 7 87 22 99 27 25 8 15 4 87 23 6 5 31 15 25 11 64 17 22 10 87 31 Correlation =.940, t = 4.555 438 144 188 51 Correlation = 1.000, t = 2.745 340 110 286 85 Correlation = 1.000, t = 14.86 626 195 37.1 36.9 28.8 31.3 8.3 10.3 7.7 6.7 6.7 7.2 4.8 2.6 2.6 2.6 4.2 2.6 Correlation =.994, t =.000 3.4 1.0 4.0 3.1 2.6 3.6 2.6 3.6 13.9 11.3 15.8 13.8 4.0 4.1 2.4 2.1 13.9 11.8 1.0 2.6 5.0 7.7 4.0 5.6 10.2 8.7 3.5 5.1 13.9 15.9 Correlation =.940, t =.000 70.0 73.8 30.0 26.2 Correlation = 1.000, t =.000 54.3 56.4 45.7 43.6 Correlation = 1.000, t =.000 Note: The survey was conducted in September-December 1999, by written questionnaire, translated into all the officialanguages of the European Union.

698 References Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Party Politics in America, Chicago: University of Chicago Press. Attina, Fulvio. 1990. "The Voting Behaviour of the European Parliament Members and the Problem of Europarties." European Journal of Political Research 18:557-579. Binder, Sarah A., Eric D. Lawrence, and Forrest Maltzman. 1999. "Uncovering the Hidden Effect of Party." Journal of Politics 61:815-831. Brzinski, Joanne Bay. 1995. "Political Group Cohesion in the European Parliament, 1989-1994" In The State of the European Union, Vol. 3, ed. Carole Rhodes and Sonia Mazey. London: Longman. Carrubba, Clifford, and Matthew Gabel. 1999. "National Politics by Other Means? Voting Behaviour in the European Parliament in the Post-Reform Era." Presented at the annual meeting of the American Political Science Association, September 1999, Atlanta. Corbett, Richard, Michael Shackleton, and Francis Jacobs. 2000. The European Parliament. 4th ed. London: Catermill. Garrett, Geoffrey. 1992. "International Cooperation and Institutional Choice: The European Community's Internal Market." International Organization 46:533-560. Hix, Simon. 1999a. The Political System of the European Union. Basingstoke: Macmillan. Hix, Simon. 1999b. "Dimensions and Alignments in European Union Politics: Cognitive Constraints and Partisan Responses." European Journal of Political Research 35:69-109. Hix, Simon. 2000. "How MEPs Vote." ESRC One Europe or Several? Programme, Briefing Note 1/00, April 2000. Falmer: University of Sussex. SIMON HIX Hix, Simon. 2001. "Legislative Behaviour and Party Competition in the European Parliament: An Application of Nominate to the EU." Journal of Common Market Studies 39:663-688. Hix, Simon, and Christopher Lord. 1997. Political Parties in the European Union, London: Macmillan. Krehbiel, Keith. 1993. "Where's the Party?" British Journal of Political Science 23:235-266. Kreppel, Amie. 2001. The European Parliament and the Supranational Party System, Cambridge: Cambridge University Press. Kreppel, Amie, and George Tsebelis. 1999. "Coalition Formation in the European Parliament." Comparative Political Studies 32: 933-966. McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2001. "The Hunt for Party Discipline in Congress." American Political Science Review 95: 673-687. Noury, Abdul. 2002. "Ideology, Nationality and Euro-Parliamentarians." European Union Politics 3: 33-58. Poole, Keith T., and Howard Rosenthal. 1997. Congress: A Political-Economic History of Roll Call Voting, Oxford: Oxford University Press. Raunio, Tapio. 1997. The European Perspective: Transnational Party Groups in the 1989-1994 European Parliament, London: Ashgate. Rohde, David W. 1991. Parties and Leaders in the Postreform House, Chicago: University of Chicago Press. Tsebelis, George. 1994. "The Power of the European Parliament as a Conditional Agenda-Setter." American Political Science Review 88:128-142.