Electoral Context and MP Constituency Focus in Australia, Canada, Ireland, New Zealand, and the U.K.*

Transcription

1 Electoral Context and MP Constituency Focus in Australia, Canada, Ireland, New Zealand, and the U.K.* Valerie Heitshusen The Government Affairs Institute at Georgetown University 3333 K St NW, Suite 112 Washington, DC Garry Young Department of Political Science 2201 G St., NW, Ste. 507 The George Washington University Washington, DC David M. Wood Department of Political Science University of Missouri Columbia Columbia, MO Contact YoungG@gwu.edu *Revised version of the paper delivered at the 2002 Meeting of the American Political Science Association, Boston, MA. The project on which this paper is based received generous funding from the National Science Foundation (SBR ) and the University of Missouri Research Board (RB98-055). Our thanks to the 254 MPs who graciously granted us interviews and to the research assistants who have helped with the project: Richard Almeida, Brandon Bartels, Jason Cruse, Matthew Eshbaugh-Soha, Mike Guadagno, Jennifer Karberg, Todd Kennedy, Mikael Pelz, Todd Reed, Moxila Uphadyaya, Amanda Weddle, and Daniel White.

2 Abstract: Legislators in western democracies typically spend large portions of their time engaged in constituency-focused activities such as casework and district visits. Unfortunately we have a limited theoretical and empirical understanding of the factors affecting legislators constituency-oriented activities, in large part because most studies focus on single nations; even those studies that are comparative do not span a variety of electoral systems. In this paper we examine the constituency focus of MPs in six chambers that do provide such variance: the Australian House and Senate, Canadian House, Irish Dáil, New Zealand House, and the British House of Commons.

3 Electoral Context and MP Constituency Focus in Australia, Canada, Ireland, New Zealand, and the U.K. Much of what goes into legislative representation in industrialized democracies occurs outside of the legislature. It involves direct interaction between legislators and their constituents, their partisans, and district interests in a number of different guises. Such interactions require the legislator to expend finite resources such as personal time and energy, as well as staffing. Unfortunately scholars have a limited theoretical and empirical understanding of the factors that affect legislators constituency-oriented activities. Numerous studies exist but most of these examine a single country and virtually all concentrate on single-member district electoral systems. 1 In this paper we examine the stated constituency focus of Members of Parliament from six national chambers in five countries: the Australian House and Senate, Canadian House, Irish Dáil, New Zealand House, and the British House of Commons. Thus we provide a comparison across a spectrum of electoral systems ranging from conventional single member district plurality voting (Canada, New Zealand districts) to single member districts with preferential voting (Australian House) to small-m multi-member districts with a Single Transferable Vote (Ireland) to small-m modified closed-list proportional voting (Australian Senate) to large-m closed-list proportional voting (New Zealand list). Influences on Constituency Effort During we interviewed 254 members of these six chambers. In each nation and chamber we found incumbent legislators who committed an impressive amount of resources towards their constituencies. These efforts include frequent weekend and recess visits to their 1 Work in this area includes Crewe (1975), Irvine (1982), Anagnoson (1983), Komito (1984) Bogdanor (1985), Cain, Ferejohn and Fiorina (1987), Wood (1987), Lancaster and Patterson (1990), Norton and Wood (1993), Searing (1994), Carey and Shugart (1995), Studlar and McAllister (1996), Wood and Young (1997), Gaines (1998), Kreuzer (2000), Ingall and Crisp (2001). 1

4 electorates, 2 the fielding of numerous constituent communications, the production of periodical newsletters and fax or communiqués, and the allocation of staff time towards constituency concerns. Why do legislators engage in so much constituency effort? Any such expenditure necessarily reduces resources potentially directed towards other areas such as producing legislative policy, garnering a party leadership role, or enjoying leisure. While all the incumbents we interviewed engage in some constituency effort, we found large amounts of variation both within and across the chambers. Is the variation solely a matter of preference? Perhaps some legislators simply enjoy interacting with their constituents while others prefer different activities. While difficult to reliably measure, such differences likely explain some variation in constituency effort. However, as we discuss, legislators face important constraints and these constraints significantly influence constituency effort. While most legislators likely seek to achieve multiple goals (e.g., Fenno 1978), and while legislators surely differ in their inclination towards different legislative roles (Searing 1994), virtually all legislators commonly face re-election. Since gaining re-election is a necessary condition to achieving other legislative goals such as attaining a ministership legislator goals are likely ordered in a hierarchical fashion such that lower-ordered goals require the highest priority; after these are realized, extensive resources can flow to the higher-ordered goals (Strøm 1997). Consequently, a natural conjecture holds that the tenor and level of legislator interactions with constituents is affected by the member s particular re-election context. Re-election contexts vary in relation to a variety of different factors. These include influences internal to a system such as a parliamentarian s relative electoral safety, as well as 2 Western democracies use different terms to refer to the geographic constituencies represented in their lower chambers. For example, while the term district is most common in the U.S., Canadians refer to constituencies as 2

5 factors that vary across systems most notably the rules of electoral systems themselves. In most cases, obtaining party re-selection is a necessary condition for gaining success in the general election, and success in both re-selection and re-election is necessary for achieving (or maintaining) party positions, legislative offices, or policy influence. Given some uncertainty about achieving any or all of their goals, and given that goals may not always be completely ordered hierarchically, legislators may think of their strategic choices as trade-offs under risk (Strøm 1997: 161). In a real sense the necessity of re-election is simply a constraint. Protecting one s chances of staying in parliament may require the direction of resources away from what an MP wants to do most. After all, few if any legislators are in office simply for the sake of getting reelected, ad nauseam. They run for office because they have policy concerns, leadership ambitions, or a desire to lessen the difficulties faced by constituents. Presumably, if the constraints were lifted that is if an MP were guaranteed re-election then the result is a policyoriented MP who directs all of his resources to policy, a leadership-oriented MP who directs all of her resources to climbing the greasy pole, or a constituency-oriented MP who directs all of his resources to playing the ombudsman. Since few if any MPs ever reach the level of absolute safety, some (or many) resources must necessarily go to re-election even if the MP prefers to distribute resources differently. Thus the approach taken here does not necessarily contradict other perspectives, most notably Searing s (1994). The amount and nature of resources actually distributed to re-selection, re-election, and other goals depends on the particular institutional context of each. Re-selection rules vary by electoral system and often by party within a particular nation s electoral system. Likewise, resource allocation to re-election is directly affected by the incumbent s prospects for re-election. ridings and Australians and New Zealanders use the term electorates. 3

6 In some cases, such as closed-list PR systems, re-election is dictated mainly by successful reselection while in other cases re-election is not possible or relevant, such as with the triggering of term limits or voluntary retirement (Strøm 1997). All legislators have resources available to them for constituency effort, but even two legislators with similar preferences who are in exactly the same electoral circumstances and who have exactly the same overall supply of resources may, nevertheless, have different levels of effort. This is because a variety of different push-pull factors interact with the re-election calculus of re-selection to affect effort. Push factors are those that increase constituency effort while pull factors decrease it. For example, important institutional or party responsibilities, such as a ministership, are pull factors. Expectations The five nations in our study share a common political origin and each remains broadly representative of the Westminster model. Nonetheless these nations elect their legislators using different electoral systems, and Australia uses different systems for its upper and lower houses. 3 (Table 1 provides summary information about the six chambers and Table 2 presents information about our samples. Details regarding our interviews, sampling techniques, etc., are in the Appendix.) Each chamber utilizes a different method to elect its members, with the range extending from classic single member plurality to closed-list proportional. To provide necessary context for our subsequent analysis, below we provide brief descriptions of the election procedures for the six chambers. 3 We include the Australian Senate in the study not just because of its interesting method of election but also because it ranks as one of the world s more powerful upper chambers. 4

7 Elections in the Six Chambers First Past the Post (Canada, New Zealand, U.K.) Canada, the U.K., and New Zealand employ single member district, plurality voting (SMP) or a first-past-the-post system to elect MPs to their lower houses. Prior to 1996, New Zealand used SMP to elect its complete unicameral parliament. Commencing with the 1996 elections, New Zealand began using Mixed Member Proportional (MMP). At the time of our interviews, New Zealand s MMP system featured sixty-five MPs elected from districts, with five of the sixty-five set aside for Maori voters. The remaining fifty-five MPs were elected from closed party lists. The Alternative Vote (Australian House) Australia s lower chamber uses the single member district majority-preferential, or alternative vote (AV) system. Here voters indicate their preferred ranking of candidates. If no candidate garners a majority of first-preference votes cast, the candidate receiving the fewest first preference votes is removed from the contest. The second preferences of the discarded candidate s ballots are then distributed to the other candidates. This process of elimination and redistribution continues until a candidate gains a majority (Bennett 1996). Single Transferable Vote (Australian Senate and Irish Dáil) Members from both the Australian Senate and the Dáil Éireann are elected by the Single Transferable Vote (STV). In Ireland voters ordinally rank candidates running in multi-member districts that have magnitudes ranging from three to five. To be elected each candidate must obtain a particular quota of votes determined by the amount of votes cast and the district s magnitude (Taagepera and Shugart 1989: 27). The excess ballots of the candidates elected on the first count are redistributed among the other candidates, and then the last place-finishing 5

8 candidate (whose ballots are also redistributed) is eliminated. This process continues until the necessary number of candidates reaches the quota (Taagepera and Shugart 1989: 27). In Australia, each state has twelve senators. Because of staggered terms, district magnitude typically is six. 4 Voters may opt to ordinally rank all candidates as with regular STV but they have the alternative of simply checking the box of their preferred party. By doing the latter, voters vote for the party s preferred ordinal ranking of candidates (sometimes called group ticket voting [GTV]). GTV is the equivalent to casting a vote for a closed party list. Though there is some variation by state (Reilly and Maley 2000: 52), this simpler option of voting above-the-line (because this option is listed on the top half of the ballot) is chosen by the overwhelming majority of voters (Bennett 1996: 56). Closed Party List As discussed above, New Zealand now uses an MMP system. In the 1996 election, fiftyfive MPs were elected from national closed party lists. Voters thus cast a ballot for a district candidate and then another for their favored party. A party s proportion of the House is determined by the party list voting, with the stipulation that the party must receive at least five percent of the vote or win at least one district. Victories in the district races are included in the count towards the party s allowable proportion (calculated from the list results) of the overall composition (Johnston and Vowels 1997). Expectations for Single-Member District (SMD) MPs We first examine factors that influence an MP s focus on his constituency. By beginning with the SMD MPs, we address only limited variation in electoral system; rather, the emphasis is more on the factors that vary within each chamber in the study factors including re-election 4 In the case of double dissolution, the complete set of Senate seats are up for election. Two Australian territories the Australian Capital Territory and the Northern Territory have two senators, each with three-year terms. 6

9 pressures, as well as other push and pull factors that might influence an MP s focus. Thus far our discussion of what constitutes constituency effort has been mainly at the conceptual level. There are many ways to measure constituency effort for example, by counting MP trips to the district, the amount of staff allocated to the constituency, or perhaps the proportion of financial resources devoted to constituency mailings. These and other measures, alone and in combination, can give us insights into constituency effort. Yet, devising comparable measures across all our chambers is challenging because of differences in, for example, sitting schedules. 5 Our approach in this paper is to rely on the stated focus of the MPs. In each country we asked the MP to rank her professional priorities be they serving the constituency, serving party interests, policy interests, or attaining a leadership position. We coded an MP as having a constituency focus if she placed serving the constituency as the sole highest priority. Such a variable gets at our concern with constituency effort in an indirect way, since it reflects a stated attitude rather than an observed behavior, yet it enjoys broad comparability across the countries. Recalling Strøm s emphasis on a hierarchy of goals, we can start to consider the factors that might explain constituency focus. If a legislator s ultimate goals are only realized after ensuring re-election, then an MP s electoral context should have a strong influence on his level of constituency focus. Since SMD MPs are dependent on their local constituents for electoral support and since local constituents expect MP attentiveness MPs with marginal seats must pay more attention to the local district while safe MPs will have more flexibility to pursue other goals. 6 However, there may be other factors that mitigate these pressures. For example, MPs who 5 Some parliaments have schedules that are conducive to trips each weekend (e.g. New Zealand s Tuesday-Thursday schedule), while others (e.g. Canada s) are not quite as amenable to such frequent visits for all MPs. In addition, the length of recess periods varies quite a bit across the chambers. This difficulties are not insurmountable and future papers will feature measures of these aspects of activity. 6 All of the countries in our study feature a high level of party based voting when compared, for example, with the U.S. As a consequence, opportunities for developing a personal vote are necessarily more limited in these countries, even for SMD MPs (e.g., Cain, Ferejohn, Fiorina 1987). While we will address the personal vote more directly in 7

10 have served for many years may build up a base of constituency goodwill after having established a reputation for district attentiveness. Therefore, an MP who is a relative newcomer may have to work harder to convince his constituents of his attentiveness than does a much more senior MP, even if both are in similar electoral circumstances. We should thus expect that both electoral safety and years served in parliament should be negatively associated with MP focus on the constituency. While electoral context and newcomer status may effectively push the MP to be more attentive to the local district, other factors may pull the MP away from doing so. First, other responsibilities especially any that require personal time of the MP away from her district should diminish constituency attentiveness. The most obvious possibility would be policy or party responsibilities that keep the MP in the capital or traveling around other parts of the country. Of course, many MPs have such responsibilities from having policy portfolios for their party to chairing parliamentary committees, or perhaps fulfilling other party responsibilities (e.g. whip) that might require additional personal time away from the constituency. Therefore, we expect that MPs with extra party or parliamentary demands will be less attentive to their constituencies. In addition, other obstacles may exist to pursuing a constituency focus. In each of the countries in this study, many MPs have serious practical constraints on the time they devote to their constituency. In particular, many districts in Australia and Canada are geographically very large, very far from the capital, or both. Some MPs face additional travel constraints if their districts (or some portions of them) are not easily accessed via major modes of transportation. future work, virtually all of our interviewed SMD MPs claimed that some amount of personal vote, however modest, was possible and thus important for an MP facing a close election. In addition, many of our MPs stressed the specter of punishment for failing to do enough constituency work rather than the possibility of gaining votes by doing constituency work. Obviously the punishment effect should more directly concern marginal MPs. 8

11 Even in the smallest country New Zealand some electorates in the South Island are very sparsely populated or include natural characteristics that pose additional travel challenges. Therefore, we try to account for these difficulties by including a measure for the amount of time it takes an MP to travel back to his district. Intuitively, we might expect higher travel times to decrease the attention an MP provides the district; however, other studies have found the opposite relationship. For example, Wood and Young (1997: 226) find this result in Ireland and the U.K. They suggest that more outlying areas (e.g. Wales, Scotland, and the West of Ireland) have higher expectations for an MP to be here, among us than constituents in the more accessible areas. In addition, these areas often have additional geographic obstacles to travel (e.g. islands and scattered villages) requiring additional personal time. Alternatively, perhaps once an MP has expended much of her time traveling to the constituency, he is more likely to focus exclusively on those activities, having already expended sunk costs to get there. Expectations for All MPs If, in the hierarchy of goals, an MP must first secure her electoral fortunes, then differences in electoral mechanisms should strongly affect the focus of an MP. This is especially true to the extent that different systems rely on different decision-makers for electoral success. Of course, while multi-member district (MMD) systems vary somewhat in the exact criteria and methods for list selection and placement, they all share common features. they all share common features. First, all MMD systems, by definition, involve geographic overlap among legislators. No MP in such a system has a constituency for which she is uniquely responsible. This generally presents fewer clear electoral incentives for cultivating a personal relationship with a set of constituents in any one geographic area. In addition, and perhaps more importantly, list MPs must please some subset of party activists to ensure ballot access and high list placement. For 9

12 closed-list incumbents in our cases, the New Zealand list MPs and Australian Senators high list placement is paramount for election and thus general constituency work contributes little to reelection. SMD-elected MPs must rely on a relatively diverse set of general election voters in their districts to ensure re-election. This helps promote a constituency focus. It is not a foregone conclusion, however, that MMD MPs will necessarily ignore constituency work. First, some do have geographic constituencies. In our study, Australian Senators represent their state as a whole, making them no less representatives of a geographic area than are U.S. Senators. For the New Zealand list MPs, one could argue that the whole country is their constituency. However, the lists MPs do have geographic constituency responsibilities that are narrower. Most parties have assigned each of their list MPs one or more duty electorates. That is, the list MP is the designated elected representative of the party in that geographic constituency; almost all of these MPs have office space and staff in these constituencies (although sometimes shared with another MP), and they are expected to deal with casework requests. They are also often called upon to represent the party at events in that electorate. Note that, for the most part, these assignments are made only to electorates whose district-elected MP is from another party. Thus, many of the list MPs are indeed asked to carry out geographically-concentrated constituency work especially by constituents who, for whatever reason, do not want to approach their district MP. Second, MPs in these list systems may have some electoral incentives to engage in geographically concentrated constituency work. For Australian Senators, the below-the-line (open-list) aspect of the ballot provides incentives to cultivate a personal vote, something that might be done via casework. A Senator who is not highly placed by his or her party may hope that a reputation for constituency work will yield below-the-line votes, resulting in higher 10

13 rankings. However, openly engaging in such a strategy risks alienating the party leadership and, as a consequence, risks exclusion from the next ballot. In the New Zealand system, incentives are a bit different. In many of our interviews with list MPs, it became clear that many would prefer to be electorate MPs. (Indeed, many had been in the previous parliament, but the new MMP system resulted in the elimination of many electorates; therefore, many sitting MPs some quite high-profile politicians had to settle for getting elected via the list.) While some of these MPs were resigned to (if also unsettled by) their new list role, a few were attempting to regain an electorate seat. Clearly, engaging in high levels of constituency work in a given electorate is a reasonable strategy for achieving such a goal. Third, in the Australian Senate and Irish Dáil cases, candidates for parliament compete not just against members of opposing parties, but also against members of their own party. While the closed-list option on the Australian ballot gives senators limited recourse, in the Irish case this intra-party competition sets up a very real incentive for cultivating a personal vote (Carey and Shugart 1995). All the Australian senators we interviewed denied overt attempts to encourage below-the-line voting. The Irish deputies, in contrast, readily recognized the need for a personal following as a fact of Irish political life. While closed-list MPs in each system have the opportunity to engage in constituency work as well as some incentives to do so the basic mechanisms of accountability in the MMD versus SMD systems differ significantly. For the closed-list MPs, their best hope for election still depends on pleasing party elites, not a geographic constituency. Since the demands of these two groups are often at odds, we expect that closed-list MPs will have to focus most of their attention on party work and much less on constituency work. We thus expect to see strong electoral system effects, with SMD MPs pursuing more constituency activities than the closed-list MPs. 11

14 As for the other type of list MPs the Irish deputies we expect high levels of constituency effort, perhaps as high or higher than the SMD MPs, on average. Given the many demands made on an MP s time and resources, he must put his resources and attention into the activities that are best suited to secure his proximate goals that of retaining his seat in the legislature. Therefore, by running our analysis for all MPs while differentiating among them based on the electoral system in which they were elected we should find strong electoral system results. Constituency Focus Results Analysis for SMD MPs We first examine factors that influence an MP s focus on the constituency by looking at the district-elected MPs separately. By beginning with the SMD MPs only, we are not addressing extensive variations in electoral system; rather, the focus is more on factors that vary within each chamber in the study factors that still address the above-mentioned push and pull factors that might influence an MP s focus on constituency. As stated earlier, the dependent variable measures the MPs stated focus on constituency, party, or policy activities. This is a dummy variable coded 1 for MPs who ranked constituency as their primary focus and 0 for all other MPs. (See Table 3 for a breakdown of this variable by election system type and by chamber.) The independent variables include: Electoral Safety (exact coding explained in the Appendix); Travel Time (the number of hours needed to travel from the parliamentary office to the constituency); Portfolio Responsibilities (ranging from 0 to 4 see Appendix for detail); and Years Served. Table 4 presents logit estimates of constituency focus for SMD MPs. Electoral Safety is negative and statistically significant, thus suggesting that marginal MPs have a stronger constituency focus. As parliamentary obligations (Portfolio Responsibilities) increase, 12

15 constituency focus declines. The same negative relationship holds for length of time an MP spends in office (Years Served), though the effect is not statistically significant at conventional levels. We find that the amount of time it takes for an MP to travel to her/his constituency is positively associated with constituency focus at a statistically significant level. This is consistent with Wood and Young s (1997) results for junior legislators in Ireland and the U.K. Finally, after accounting for the variables of interest, are there still country specific effects? We scored dummy variables for Canada, New Zealand, and the U.K., leaving Australia as the baseline. Only New Zealand is statistically significant. Table 5 presents data on substantive effects. Here we calculated the probability of a constituency focus in light of a set independent variable levels. (We also include confidence intervals around the probability estimates.) Setting all the independent variables to their mean yields a.53 chance that an MP has a constituency focus. We report additional estimates for only the independent variables that were statistically significant. Holding all other independent variables at their means, we varied the Electoral Safety variable across its minimum and maximum values. Doing this yielded a.50 shift in the probability of a constituency focus. As Portfolio Responsibilities track upward, constituency focus probability marches steadily downward. Indeed, moving from no portfolio responsibilities to ministerial duties, the constituency focus probability plummets from.72 to.13. Shifting minimum to maximum Travel Time yields a range of.36 to.92. Thus, as discussed above, MPs who have the hardest time reaching their electorates are very likely to place their constituencies first, an indication, perhaps, of a center-periphery cultural effect. Table 5 also suggests a substantial New Zealand effect. Given the model s controls, New Zealand electorate MPs are very likely (.93) to prioritize constituency the highest. Ad hoc 13

16 explanations for the national differences are readily available, but theoretically informed explanations are more difficult to come by. For example, there are at least three reasons for the New Zealand effect: 1) the transition to MMP increased electorate sizes dramatically, thus forcing incumbent electorate MPs to heighten their constituency presence; 2) the presence of competing list MPs in their electorates, some of whom operate competing electorate offices, places pressure on the New Zealand electorate MPs to focus more on their constituency (see also Ward 1998); and 3) New Zealand electorates are smaller and generally more rural than Australian, Canadian, and British electorates, thus encouraging a closer MP-constituency relationship. Analysis for All MPs By including all MPs, we can examine a variable that accounts for the differences between MPs elected in different systems. We use the same dependent variable as in the SMD analysis. We then add a dummy variable for the electoral system under which the MP was elected (Election Type: 1=MMD, 0=SMD). We exclude the variable for travel time because New Zealand list MPs do not have a single identifiable geographic constituency (short of the nation). In addition, with one exception we have excluded the country dummies here since their inclusion would conflate electoral system and country effects. The exception is Ireland, which is a necessary inclusion since the expectations from Ireland differ dramatically from the expectations for the other MMDs. Table 6 presents the results for all MPs. Not surprisingly, the coefficient for Election Type is negative and statistically significant. MPs elected from MMDs are less likely to focus on constituency than SMD MPs. Though measured somewhat differently (see Appendix) Electoral Safety remains negative and statistically significant. Portfolio Responsibilities remains negative 14

17 and statistically significant as well. In the SMD-only model (Table 4), Years Served was negative but not significant. Adding the MMD chambers keeps the variable negative and raises it to statistical significance. Finally, as expected, the control for Ireland is positive and statistically significant. Relative to other MMDs, Irish TDs are more likely to focus on constituency. As for substantive effects, Table 7 shows the all-at-mean prediction for this model is.38, much lower than in Table 5 because of the presence of the list MPs. Indeed the table shows a large effect for MMD versus SMD, with MMD MPs presenting only a.06 chance for constituency focus compared to the.64 chance for SMD MPs. The effects of the Portfolio variable are not quite as pronounced in this model as with Table 5, though they still show a notable range from.46 to.23 from no portfolio responsibilities to ministership, respectively. Given the controls, Irish TDs are.88 likely to prioritize constituency. Conclusion Through an examination of MPs from six national chambers, we have demonstrated that an MP s focus on constituency activity can be understood as shaped by the electoral considerations that must, by necessity, undergird many legislators decision calculus. MPs who are more dependent on local constituent support relative to others i.e. relative to other more safe or more senior SMD MPs, or relative to MPs who are elected in a list-based system with a different set of electoral incentives are more oriented towards a constituency focus. In addition, other factors also help us explain variation in constituency focus e.g. other parliamentary responsibilities and the constituency s proximity to the capital. Several of these findings deserve elaboration. First, our statistical results, and indeed most of our interviews, do suggest an electoral connection. Part of an MP s motivation for constituency service derives from electoral concerns. As one Canadian MP put it, If you want 15

18 to be in this job for a long time then you make sure you do your riding right (Personal Interview, 1998). This is not to claim that the circumstances of gaining a personal vote in these countries resembles that of the United States. Opportunities for developing a sizable personal vote are constrained. Yet, the ability to gain a few personal votes, or to avoid getting a reputation for shirking that could provoke punishment votes, is a motivation for the more marginal MPs. Second, our finding that SMD MPs have more of a constituency focus than MMD MPs, is not a surprise. Yet, we do demonstrate that many MMD MPs, though primarily those from Ireland, do focus on a general constituency. 7 Finally, our findings about marginality and SMD versus MMD say something potentially important about overall representation. Insomuch as marginal MPs feel obligated to pour a relatively high amount of resources into constituency work, they are left with a relatively small amount of resources to allocate in other ways, such as pursuing advancement to the leadership. Similarly, closed-list MPs, relatively free of constituent demand, have more freedom to allocate resources elsewhere. For pure SMD systems, this implies an advantage to safe MPs for leadership advancement and developing policy expertise. For mixed systems, in our case New Zealand, it suggests the possibility that list MPs will have a similar advantage over the SMD MPs, especially over those SMD MPs from marginal seats. 8 Overall, such differential constraints can affect the overall composition of party and policy leadership in the parliament, and thus affect the nature of policy outputs produced by the parliament. 7 As Table 3 indicates, virtually no MMD MPs outside of Ireland place constituency as their top priority. When we look at more direct measures of behavior such as casework, as we will do in future work, MMD MPs in New Zealand and the Australian Senate do engage in a surprisingly high amount of constituency work. 8 This point has not been lost on New Zealand electorate MPs, as a number noted it in our interviews. 16

19 Appendix Sample Selection The selection of our legislator samples in each country was basically random, with some adjustments for replacement done in the field. First, sample legislators were randomly selected from a list of sitting backbenchers. We defined backbenchers somewhat loosely. In general, we excluded Speakers, top party leaders, government cabinet-level ministers, and chief whips from the population list from which we drew. However, we included MPs from opposition parties with portfolio responsibilities chiefly because such responsibilities are so widely shared in most opposition parties that excluding them would have seriously winnowed our population (and significantly biased it along a number of dimensions). This means that those we interviewed from small parties and from the official opposition parties are more likely to include MPs with some portfolio responsibilities than those from the government parties. Once each of the sample MPs were contacted, we relied on a randomly generated backup list for replacement. The replacement process itself was not random, however. In other words, if the MPs who refused to be interviewed numbered heavily from a certain party, region, or gender, we selected the first MP on the back-up sample list who fit the underrepresented criterion. Since even our replacement MPs did not uniformly grant interviews, the final interview pool does not perfectly reflect the population. In some cases, replacement interviews included MPs that had initially been excluded from the backbencher list. Therefore, the final pool does include a small number of sitting ministers (7) and chief whips (2). One additional complication occurred in Australia, which somewhat biased our state-by-state representation. In late August of 1998, Australia called a federal election a move we expected several weeks in advance. In a endeavor to speak to as many MPs as possible before the election was called especially based on our (apparently justified) fear that marginal legislators in the sample might lose their seats we added marginal MPs to our sample who had offices nearby other interviewees (mostly in Tasmania) with whom we had scheduled August interviews. Since some of these MPs were indeed defeated in the election, this tactic allowed us to interview a reasonable number of marginal MPs, at the expense of perfect regional representation. Also in relation to the Australian election, note that the sample versus population figures in Table 2 re flect the pre-1998 election partisan breakdown in the Australian Parliament, not the post-election figures, reflecting the fact that our sample was drawn from the pre-election Parliament. 17

20 Interview Format All interviews were conducted in person by one or two of the authors. All but seven interviews were taperecorded with the legislator s consent; the exceptions were six Canadian MPs and one British MP. Each legislator was promised anonymity. Interviews typically lasted between thirty minutes to an hour, with many lasting a bit longer. Nine of the interviews with Quebec MPs were conducted in French. Legislators were asked a variety of questions approximately thirty-one that differed slightly across electoral systems. Many questions were openended in nature, while others requested answers that fell into categories (basically, multiple choice). The Canadian interviews were done in Ottawa, in two waves forty were conducted in late May and early June of 1998, while the remaining twenty-seven occurred in September The Australian interviews were also conducted in two waves. In August of 1998, twelve House interviews and six Senate interviews were conducted in the electorate/state office of the legislator. An election was subsequently called, making it necessary to postpone remaining interviews. The second wave of interviews was conducted in Canberra after the election, in November and December of The New Zealand interviews were conducted in Wellington in September and October of We intervie wed the Irish TDs and UK MPs in February and March of Electoral Safety Although we have interview data on subjective electoral safety (legislator perceptions), the electoral safety data used in the analysis is coded objectively, based on the most recent election prior to our interviews (Canada: 1997; Australian House: 1996; Australian Senate: 1993 and 1996; New Zealand: 1996; Ireland: 1997; UK: 1997). We coded this variable in two different ways. For the estimates based on SMD MPs only, we use the two-party margin for the incumbent. For Canada and the UK, the margin is the winner s share of the vote minus the second place candidate s share; the New Zealand electorate measure is similarly calculated. In the Australian House case this is the difference in the two-party preferred vote share. Since we cannot use similar measures for estimates including the MMD MPs, we used ordinal ranking of safety for the analysis that includes them. For the New Zealand electorate, Canadian, and Australian MPs, seats are classified as safe if the margin was greater than 20%, competitive for margins greater than 10% but below 20%, and marginal for MPs with margins of 10% or less. As for the list-elected legislators in New Zealand and Australia, we use relative list placement to measure seat safety. (Some of the Australian Senators have been appointed to fill vacancies; they are excluded from the analyses.) We rank each sitting legislator among all list-elected legislators 18

21 from that party, according to their placement on the list. Legislators who fall in the top third of their party s elected list MPs are coded as having a high place on the list (and are therefore safe). Those in the second third are in competitive slots, and those in the bottom third are coded as marginal. For Irish TDs, we coded the difference between the MP s first-preference votes and the district s quota. We then sorted these margins and divided them into thirds, coding them as safe, competitive, and marginal (as with the Australian Senators and New Zealand list MPs). Portfolio Responsibilities MPs with no responsibilities are coded 0. A value of 1 is coded for the following positions: deputy critics/spokespersons, past major officials (i.e. minister or party leader). MPs with the following positions received a 2 : minor party critics/spokespersons; parliamentary secretaries, deputy party or chamber officials (i.e. deputy whips, deputy speakers). A 3 is coded for critics/spokespersons/shadow ministers from the official opposition party, committee chairs, and full party or chamber leadership positions (i.e. party leader, chief party whip, Speaker). Only sitting ministers receive a 4. 19

22 References Anagnoson, J. Theodore Home Style in New Zealand, Legislative Studies Quarterly 8: Bean, Clive The Personal Vote in Australian Federal Elections, Political Studies 38: Bennett, Scott Winning and Losing: Australian National Elections. Carlton South, Australia: Melbourne University Press. Bogdanor, Vernon, ed Representatives of the People? Parliamentarians and Constituents in Western Democracies. Aldershot, U.K.: Gower. Cain, Bruce, John Ferejohn, and Morris Fiorina The Personal Vote: Constituency Service and Electoral Independence. Cambridge, MA. and London: Harvard University Press. Carey, John M., and Matthew Soberg Shugart Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas, Electoral Studies 14: Crewe, Ivor Electoral Reform and the Local M.P. In Adversary Politics and Electoral Reform, ed. S.E. Finer. London: Anthony Wigram. Docherty, David C Mr. Smith Goes to Ottawa Vancouver: UBC Press. Erickson, Lynda Canada, in Passages to Power: Legislative Recruitment in Advanced Democracies Pippa Norris, ed. Cambridge: Cambridge University Press. Erickson, Lynda, and R.K. Carty Parties and Candidate Selection in the 1988 Canadian General Election, Canadian Journal of Political Science 24: Fenno, Richard Home Style: House Members in Their Districts. Boston: Little, Brown. Gaines, Brian Incumbency Advantage and the Personal Vote in the Anglo-Atlantic Democracies. Ph.D. Dissertation, Stanford University. 20

23 Gaines, Brian The Impersonal Vote? Constituency Service and Incumbency Advantage in British Elections, Legislative Studies Quarterly 23: Ingall, Rachel and Brian Crisp Determinants of Home Style: The Many Incentives for Going Home in Colombia. Legislative Studies Quarterly 26: Irvine, William P Does the Candidate Make a Difference? The Macro-Politics and Micro-Politics of Getting Elected, Canadian Journal of Political Science 15: Johnston, Richard and Jack Vowles The New Rules and the New Game in New Zealand Elections: Implications for the Campaign, Paper presented at the 1997 American Political Science Association meeting, Washington, D.C. King, Gary, Michael Tomz, and Jason Wittenberg Making the Most of Statistical Analyses: Improving Interpretation and Presentation, American Journal of Political Science 44: Komito, Lee Irish Clientelism: A Reappraisal, The Economic and Social Review 15: Kreuzer, Marcus Electoral Mechanisms and Electioneering Incentives: Vote-Getting Strategies of Japanese, French, British, German, and Austrian Conservatives. Party Politics 6: Norris, Pippa, R.K. Carty, Lynda Erickson, Joni Lovenduski, and Marian Simms Party Selectorates in Australia, Britain, and Canada: Prolegomena for Research in the 1990s, Journal of Commonwealth and Comparative Politics 28: Norris, Pippa The Puzzle of Constituency Service, Paper presented at the annual meeting of the American Political Science Association, Chicago, IL. 21

24 Norton, Philip, and David M. Wood Back from Westminster: British Members of Parliament and Their Constituents. Lexington, KY: The University Press of Kentucky. Rilley, Ben and Michael Maley The Single Transferable Vote and the Alternative Vote Compared, in Elections in Australia, Ireland, and Malta under the Single Transferable Vote: Reflections on an Embedded Institution, Shaun Bowler and Bernard Grofman, eds. Ann Arbor: University of Michigan Press. Searing, Donald D Westminster's World: Understanding Political Roles. Cambridge, MA, and London: Harvard University Press. Strøm, Kaare Rules, Reasons, and Routines: Legislative Roles in Parliamentary Democracies. Journal of Legislative Studies 3: Studlar, Donley, and Ian McAllister Constituency Activity and Representational Roles among Australian Legislators, Journal of Politics 58: Taagepera, Rein and Matthew Shugart Seats and Votes. New Haven: Yale Univ. Press. Tomz, Michael, Jason Wittenberg, and Gary King CLARIFY: Software for Interpreting and Presenting Statistical Results. Version Cambridge, MA: Harvard University, June 1, Ward, Leigh Second-Class MPs? New Zealand s Adaptation to MMP, Political Science 49: Wood, David M The Conservative Member of Parliament as Lobbyist for Constituency Economic Interests, Political Studies 35 (September): Wood, David M. and Garry Young Constituency Activity by Junior Legislators in Britain and Ireland: A Comparison, Legislative Studies Quarterly 22:

25 Table 1 Chambers in Study Electoral System Country Chamber Name Size Type Magnitude Australia House of Representatives 148 Alternative Vote (Majority- M=1 Preferential) Australia Senate 76 Single Transferable Vote Proportional with GTV (closed list option) Canada House of Commons 301 Single Member, Plurality Ireland Dáil Éireann 166 Single Transferable Vote New House of 120 Mixed Member Zealand Representatives Proportional United House of 651 Single Member, Kingdom Commons Plurality M=6 (States) M=2 (Territories) M=1 M=3 to 5 M=1 (65 seats) M=55 (remaining 55 seats) M=1

26 Table 2 Sample Statistics Australian House Pop. % of Pop. Sample % of Sample Overall Parties Labor (ALP) Liberal National Independents/Other States New South Wales Victoria Queensland W. Australia S. Australia Tasmania ACT / N. Territory Australian Senate Pop. % of Sample % of Sample Pop. Overall Parties Labor (ALP) Liberal National Independents/Other States New South Wales Victoria Queensland W. Australia S. Australia Tasmania ACT / N. Territory Canadian House Pop. % of Pop. Sample % of Sample New Zealand House Pop. % of Pop. Sample % of Sample Overall Parties Bloc Québécois Liberal New Democratic Prog. Conservatives Reform Independent Regions West (Alb.,BC,Man.,Sas.) Ontario Quebec Maritimes (NB,Nfd.,NS,PEI) Yukon, NWT, Nun Overall Parties ACT Alliance Labour National NZ 1 St /Indep United NZ Type of MP Electorate (General) Electorate (Maori) List Irish Dáil Pop. % of Pop. Sample % of Sample UK House of Commons Pop. % of Pop. Sample % of Sample Overall Parties Fianna Fáil Fine Gael Labour Democratic Left Progressive Dems Others Overall Parties Labour Conservative Liberal Democrats Other

27 Table 3 MPs Stated Focus Constituency Other MMD MPs 17 (21%) 64 (79%) SMD MPs 93 (54%) 80 (46%) Australian House 19 (43%) 25 (57%) Australian Senate 0 (0%) 20 (100%) Canadian House 38 (57%) 29 (43%) Irish Dáil 16 (39%) 25 (61%) New Zealand Electorate 12 (71%) 5 (29%) New Zealand List 1 (5%) 19 (95%) UK House of Commons 24 (53%) 21 (47%)

28 Table 4 Constituency Focus of SMD MPs (Logit Estimates) Variable Estimate Std. Error Electoral Safety *** Portfolio Responsibilities *** Years Served Travel Time ** Canada New Zealand *** U.K Constant n 173 LR χ *** McKelvey-Zavoina R *** p <.01 ** p <.05 * p<.10

29 Table 5 Estimated Probability of Constituency Focus for SMD MPs Variable Pr(Y=1) 95% Conf. Int. All at mean Electoral Safety at minimum at maximum Portfolio Level 0 (None) (Minister) Travel Time at minimum at maximum New Zealand = = Note: All other variables are held at their mean while the statistically significant variable is set to the given value. Probability estimates produced using CLARIFY & Stata 6.0. (Tomz, Wittenberg, King 1999; King, Tomz, and Wittenberg 2000).