838
European Journal of Political Research 50: 838-866, 2011 doi: 10.1111/j.l475-6765.2010.01985.x
Three roads to institutionalisation: Vote-, office- and policy-seeking explanations of party switching in Poland
IAIN McMENAMIN1 & ANNA GWIAZDA2
1 Centre for International Studies and School of Law and Government, Dublin City University, Ireland;2Department of Political Economy, King's College London, UK
Abstract. Party institutionalisation is a central problem in political science. The literature tends to understand it as a syndrome and therefore has difficulty explaining variations. This article suggests a new approach based on the transaction between a legislative party and its deputies, the failure of which is observable in party switching. Three routes to institutionalisation are identified by appealing to the vote-seeking, office-seeking or policy-seeking motivations of deputies. Poland has had a large volume of party switching, along with wide variation in the incentives facing differently-motivated deputies. Survival analyses of switching in four Polish parliaments find that vote-seeking is the most likely route to institutionalisation for Polish parties. Moreover, in this article a concrete hypothesis is established for comparative testing: legislative parties can survive as long as their popular support exceeds 40 per cent of their share in the previous election.
Keywords: party politics; institutionalisation; legislatures; Poland; democracy
Institutionalising political parties has been one of the great challenges of the third wave of democratisation. There is a clear consensus that institutionalised parties are important for a well-functioning democracy (Mainwaring & Scully 1995:27-28; Reilly 2008:3-4,7-9). Unfortunately, institutionalisation remains a mysterious process. Political scientists do well in separating institutionalised parties from uninstitutionalised parties, but have great problems in explaining why some institutionalise and others do not. We take a new approach that focuses on the basic transaction of the legislative party: the exchange of deputies' autonomy for the benefits of collective action. The failure of this transaction is clearly observable in deputies' decisions to switch legislative party. The legislative party can take three roads to institutionalisation through a sequence of self-enforcement and self-reinforcement. These three routes are associated with three types of parties: vote-seeking parties, office-seeking parties and policy-seeking parties. We suggest a new way of tackling a major issue in political research, but draw our inspiration from well-established literatures.
We have chosen Poland to test our approach as it has experienced a huge variation in party switching over time and across parties. Moreover, there has
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Maiden, MA 02148, USA
THREE ROADS TO INSTITUTIONALISATION
839
been wide variation in the incentives for vote-seeking, office-seeking and policy-seeking motivations to switch party. The core of our empirical analysis consists of survival analyses of party switching in the four most recent Polish parliaments. We find that the vote-seeking perspective is a far stronger explanation for switching than the office- or policy-based explanations. According to our models, parties should be safe from mass switching if they retain over 40 per cent of their support from the previous election. This figure establishes an interesting hypothesis for other country cases. Indeed, the concepts and methods used in this research should travel very easily to other contexts.
The article is organised conventionally. The next section discusses theory, explaining how and why we depart from the conventional 'syndrome' approach to party institutionalisation. Then we explain why the Polish case is an apposite one for our purposes, and introduce the measures and methods with which we test for the three routes to party institutionalisation. After presenting the survival analysis, we discuss the prospects for the institutionalisation of Poland's vote-seeking parties and consider implications for the comparative study of political parties. A conclusion summarises and makes some suggestions for further research.
Theory
Huntington (1968: 12) defines 'institutionalisation' as 'the process by which organizations acquire value and stability'. Randall applies this concept to political parties, defining 'party institutionalisation' as comprising organisational system-ness, embedded decisional autonomy, value infusion, a definite public image and presence, and a relatively stable basis of support (Randall 2006; Randall & Svasand 2002). This definition has a lot in common with understandings of party system institutionalisation (Bertoa 2009; Gwiazda 2009:352-355; Mainwaring 1999:26-27; Mainwaring & Scully 1995; Sikk 2005; Krupavicius 2004; Markowski 2001).These multidimensional conceptions have been the subject of a huge amount of research, both implicit and explicit, and they undoubtedly point to some profound differences among political systems. However, the multidimensional definitions constitute syndromes, rather than theoretical models. As dependent variables they can separate institutionalised from uninstitutionalised systems. They also identify some powerful vicious and virtuous circles. The problem with syndromes is that the relationships between the variables are not sufficiently explicated. Syndromes are not good at generating predictions or explaining variations in institutionalisation. The literature has not provided concrete suggestions on how parties break in and out of vicious and virtuous circles. One reason for these weaknesses may be the
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
840
IAIN McMENAMIN & ANNA GWIAZDA
ambitious generality of the literature, which tends to ignore the differences between types of political parties by holding them all to an abstract standard of institutionalisation.
We take a more concrete approach that studies party institutionalisation in terms of clear causal mechanisms rather than syndromes or circular processes. Our focus is on the institution of the legislative party. We measure institutionalisation by party switching or 'any recorded change in party affiliation on the part of a politician ... holding elective office' (Mershon & Shvetsova 2008: 104). Obviously, a legislative party that cannot retain its deputies has not acquired 'value and stability'. Party switching has been rare in most established democracies (Heller & Mershon 2009a: 4). However, it is prevalent in many newer and non-Western democracies, such as Brazil, India and the Philippines. Law- and constitution-makers in many of these countries seem to agree with political scientists who emphasise stable party systems, and have banned or sanctioned party switching (Janda 2009: 3-4; Reilly 2008:15-16).
Greif and Laitin (2004) follow game theorists in presenting institutions as transactions. All legislative parties are sustained by the same basic transaction: deputies trade their autonomy for the benefits of collective action. They also share the common institutional form of the legislative party. Nonetheless, these institutions vary in their basic aims. Str0m (1990) distinguishes among vote-seeking, office-seeking and policy-seeking parties. Logically, these different types of parties have different ways of enforcing the institution of the legislative party. Defectors from a vote-seeking party cannot seek re-election on the party's ticket. Defectors from an office-seeking party will not be nominated to leadership positions in the executive or the legislature. Defectors from a policy-seeking party will not be able to influence its policy position. These enforcement mechanisms depend on parameters that are outside the direct control of the party (Greif & Laitin 2004: 634). The importance of a party's nomination for the legislative election depends on its level of popular support. The benefits of office are not available to all parties and usually depend on a given party's legislative strength and place in the policy space. Policy influence also depends on legislative strength and place in the policy space, as well as the overall structure of public policy. All other things being equal, a party seeking policies that are consistent with inherited policy structures is more likely to be successful than one seeking radical changes. While these are given parameters at a point in time, over time parties can influence them. Thus, they are quasi-parameters (Greif & Laitin 2004: 639).
Parties are expected to compete for votes. Parties manoeuvre for control of government and legislative positions. Similarly, they can and do change the policy space and the structure of public policy. In this sense, the institution of the legislative party has three potential mechanisms of self-reinforcement. This
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
841
means the basic institutional transaction is compatible with a wider range of parameter values. In other words, it has gained value to its deputies beyond the immediate exchange of autonomy for the benefits of collective action. Vote-seeking parties with a record of delivering votes should suffer fewer defections. Office-seeking parties that have in the past distributed the benefits of office to their deputies should have a lower switching rate. Finally, parties that have achieved policy successes in the past will be surer of the loyalty of their deputies. Thus, we identify three clear paths to legislative party institutionali-sation. By emphasising electoral, office and policy benefits to deputies in exchange for their loyalty parties ensure their own value and stability. Clearly, these types are not mutually exclusive. Indeed, Str0m (1990:570-573) presents them as trade-offs and locates a given party in a three-dimensional space.
Str0m's behavioural theory of political parties and Greif and Laitin's theory of institutional change are both parsimonious and synthetic. Str0m (1990:568-570) seeks to combine the deductive work on political competition in the Downsian tradition with the more inductive comparative literature on political parties. Similarly, Greif and Laitin (2004: 634-636) bring together the deductive work of game theorists and the more inductive approach of historical institutionalists. Our work does not add to the conceptual and theoretical profusion that hampers social science. Instead, it seeks to address a major problem in the literature by building on some of the major strands in research developed over the last number of decades.
There is a small but growing literature on party switching. Although it rarely does so explicitly as we do, it operates very much within Str0m's understanding of the three types of political parties (Heller & Mershon 2009b: 33, 2009c: 289; Kato & Yamamoto 2009:260). Most explicitly test for vote-seeking behaviour. Desposato (2006:71) argues that some Brazilian switches are motivated by short-term electoral gain, while Mershon (2009: 394) notes higher levels of switching in the lead-up to general elections, or around the time of second-order elections, which provide important information on the re-election chances of deputies. Zielinski et al. (2005) show how Polish deputies switched parties in order to escape electoral accountability for weak government performance in their constituency. Reed and Schemer's (2003: 473) work on Japan also emphasises 'electoral incentives'. Similarly, the office-seeking party is often found in studies of legislative switching. Desposato's (2006: 70) Brazilian 'pork' is delivered through executive offices. Mershon (2009:394) identifies a benefits stage, during which government and legislative offices are allocated. Finally, policy-seeking is also identified as a motivation for legislative switching. Mershon (2009: 34) manages to isolate a control-of-policy stage, while Heller and Mershon (2005: 543) argue that some Italian switching has been caused by the fuzzy party labels that provide little
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
842
IAIN McMENAMIN & ANNA GWIAZDA
information on policy goals. Desposato's (2006:71) model of Brazilian switching assumes parties wish to minimise ideological heterogeneity, and Reed and Scheiner (2003: 473) explain the puzzle of the collapse of Japan's Liberal Democratic Party by looking at the policy preferences of its deputies.
The Polish case
We study party switching in the lower, and much more powerful, house of the Polish parliament, the Sejm (Sanford 2002: 103-104). The Polish case is an apposite test of routes to party institutionalisation because of the large volume of switching and the substantial variation in the independent variables. There were major changes in support for parties in every term. Therefore, there was intra-party and cross-party variation in potential for vote-seeking institutionalisation during every term. There have been no grand coalitions so there has always been the basic contrast between government and opposition. Moreover, in all parliaments, except 1993-1997, the party composition of the government changed during the term. Therefore, there is scope to study variation in the incentives of parties seeking the most important offices. Variations in the size, position in the ideological space and organisation of parties (Szczerbiak 2001; Str0m 1990: 593) are likely to have resulted in different levels of office-and policy-seeking.
Another advantage of the Polish case is that the changes in the overall institutional configuration are unlikely to have introduced bias into our conclusions. We do not argue that post-communist Poland has been a paragon of constitutional stability. Instead, we suggest that the rules directly impacting on legislative switching have not changed substantially. Most relevantly, the key standing orders of the Sejm have not changed: they distinguish between caucuses (3-14 members) and groups (15 or more members) (Sejm Rzeczypos-politej Polskiej 1992: Article 8). Heller and Mershon (2005) capitalise on the two electoral systems used simultaneously in Italy. There was no such profound internal variation in Poland. The essence of the electoral system has remained a quasi-list system (Shugart 2005), which gives voters some scope to modify the order of deputies on parties' lists. However, there is a big contrast between the proportionality of the electoral system in the first (1991) election and later elections, which had thresholds of 5 per cent for parties and 8 per cent for electoral coalitions.
Our analysis begins in 1993. A less momentous reform in 2001 increased the proportionality of the system, but was not associated with changes in fragmentation (Gwiazda 2009: 368). The executive has been semi-presidential, with a dominant prime minister. The 'Little Constitution' of 1992 introduced
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
843
important changes in executive-legislative relations, again before our study begins. The 1997 Constitution introduced some reductions in the president's powers (McMenamin 2008:124), but these did not affect incentives for party switching. At any rate, such incentives would operate through membership of a governing party, which we measure.
Polish deputies have always been free to switch parties as they wish. The leader of the populist Serf-Defence Party tried to eliminate switching by requiring all candidates to sign an agreement that would make them liable for a large amount of money on defection. However, this was ruled illegal on several grounds, including unconstitutionality (Gazeta Wyborcza 2005). It is doubtful whether there would be any public support for constraints on switching, as Poles hold very negative opinions on political parties. Indeed, their opinions on, and participation in, parties are negative and low even when compared to other East-Central European countries and other Polish institutions (Szczerbiak 2001:195-199). It is not clear to what extent public opinion disapproves of switching. However, there is some evidence that defectors can evade accountability for the failures of governing parties (Zielinski et al. 2005).
Research design
Studies of party switching vary in the precision of their observations. Des-posato (2006: 72) examines individual deputies' decisions to switch across a large number of short time-periods. Zielinski et al. (2005: 380) examine individual politicians also, but only from election to election, and Heller and Mershon (2005: 544-548) look at deputies at the beginning and end of a term. Reed and Scheiner (2003) examine a small number of splitting episodes at the individual level. Mershon (2009: 401, 404) and Mershon and Shvetsova (2008: 110) report aggregate switching rates but in precisely defined time periods. Finally, Shabad and Slomczynski (2004: 153-154) use aggregate data on switching rates - again observed only at election time. These differences are summarised in Table 1.
We analyse Polish switching at a high level of precision, as do Desposato and Heller and Mershon. We study four terms intensively (with some data from a fifth), in contrast to Desposato's two and Heller and Mershon's one. Our variables are intended to test the vote-, office- and policy-seeking routes to party institutionalisation, and most of our measures are similar to those used in the literature discussed above. The data and codebook are available online at: http://webpages.dcu.ie/~mcmenami
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
844 IAIN McMENAMIN & ANNA GWIAZDA
Table 1. Classification of quantitative studies of switching
Time-period observations
Many
Few
Observation of deputies
Individual
Desposato (2006), Heller & Mershon
Zielinski et al. (2005), Reed & Scheiner
(2005)
(2003)
Shabad & Slomczynski (2004)
Aggregate
Mershon & Shvetsova
(2008), Mershon (2009)
The dependent variable
Our data include measurements on all deputies who served during the four terms since 1993. Their affiliations have been taken from the official record of the Sejm. The data note the exact day of any switch. Stable affiliations have been divided into 15-day periods to facilitate the inclusion of opinion poll data. The large number of observations essentially results from the multiplication of the number of 15-day periods by the number of deputies. Some deputies were replaced in each term, particularly after elections to the European Parliament in 2004. The 1991-1993 parliament ended early due to a presidential dissolution after a no-confidence vote in the government. In 2007, the Sejm dissolved itself.
While switching is clearly a discrete phenomenon with important theoretical and substantive implications, it is also a heterogeneous category. In a parliamentary regime, the direction of switching between government and opposition seems to be the most politically important (Mershon 2009). In Italy, switching was concentrated within the opposition and within governing parties. This has clearly not been the case in Poland. The most common direction has been within the opposition, but the politically explosive category of government to opposition has been the second most frequent in Poland (Table 2). A direction, which does not feature in the literature, but is important to our concern with institutionalisation, is from old to new parties. In addition to overall switching rates, we will analyse switching in the two most frequent directions of within the opposition and from government to opposition, as well as from old to new parties.
Independent variables
We employ a range of measures of vote-, office- and policy-seeking motivations. Poll is the party's opinion poll score as a proportion of its popular vote
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION 845
a
-a a
«3
O
a
+^ o o
S-h
-a o
f2
«3 I
M ON
ON
o o
<n
I
m o o
m o o
<n
I
o o
o o
<n
I
r~-
ON ON
ON ON
cn ON ON
cn
ON ON
ON ON
O cn r~-
O   <n   O O
o  o cn dodo
o
o
<n O
d d
o cn
cn <n
m o
d d
o o o r--
m (n
cn r- o
d d d
o m  o o o  cn o cn  o  cn o
dodo
o o
o m o o  o  r- o
dodo
o o o o
ON   \D O
O O
v-5     O     T-H O
a o
o
O
o
a
a ■—
> O
0
a
> o
o
o O
a o
a
o O
> o
0
a a o o
S S
o -a
o o
T-H S-h
o
o
ON
o o
<n
a o
B o
p £
«3 03 &
2h ^
£ a a «3
<D ^ 03
,—i -t-j
O
o o
S-h
o
c/3
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
846
IAIN McMENAMIN & ANNA GWIAZDA
at the last election. All polls are from CBOS; the vast majority of these polls are monthly or more frequent. There were two separate monthly interpolations in the 1997-2001 parliament. For the 1993-1997 parliament, there were three separate monthly interpolations and the first seven months are an interpolation between the election result and the May 1995 poll. For all data points, except May-August 1996, the survey question asked only for which party respondents intended to vote. For four months, the question also named the party leader. For these months, the party-only question has been estimated from the relationship between party-only and party-and-leader formats in March and April 1996. These data limitations probably contribute to the comparatively weak effect of the Poll variable in this term. Polls did not measure support for the large number of micro-parties, never mind independents. Such missing values have been coded zero, which reflects the minimal chances of success for deputies who contest an election for a minor list. We reran all our analyses of overall switching in the four terms, excluding all poll values of zero. This made little difference to the performance of the Poll variable. Also, we ran our models with lags of 15 and 30 days. The lagged versions did not perform nearly as well as the straightforward version, so we do not report them.
Figures 1-4 report voting preferences for the largest parties in each term. During the 1993-1997, 1997-2001 and 2001-2005 terms there were massive swings in popular support for the parties during the parliamentary term. In the 1993-1997 term, the left-wing SLD and centrist UD/UW went through troughs and over peaks, but their deputies never had to endure a long period in which many of them would have had to worry about re-election. In contrast, the Polish Peasant Party ended the term with much lower support than they had at the election. Moreover, the right-wing KPN and BBWR lost much of their electorate before the end of their separate existence half way through the term. The period 1997-2001 saw the spectacular obliteration of support for the election-winning AWS right-wing bloc, with a comparably catastrophic fate for the centrist Freedom Union; 2001-2005 exhibits a similar pattern, but this time it is the left-wing SLD that suffers a disastrous collapse in support. The right-wing Law and Justice and centre-right Civic Platform both benefitted from a tripling of support. In the 2005-2007 parliament, preferences for the major parties represented a stalemate, but two smaller governing parties (Self-Defence and the League of Polish Families) found themselves below the electoral threshold of 5 per cent for much of the second year of the term.
Our second variable is List. Vote-seeking politicians who can expect a high list position in the next election should be less likely to switch. List is a dummy variable that distinguishes between the first and lower places on a party's electoral list. In the 2001-2005 parliament, 41 per cent of deputies had been
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALIZATION
847
Figure 1. Voting preferences, 1993-1997.
Notes: SLD = Alliance for the Democratic Left; PSL = Polish Peasant Party; UD/UW = Democratic Union/Freedom Union; BBWR = Non-party Bloc for the Support of the Reforms; KPN = Confederation for an Independent Poland.
placed first on an electoral list. In the 2001 election, in constituency number one, the first placed candidates on the nine lists had a 0.66 probability of election, with those in second place scoring a probability of less than 0.03. We have chosen the 2001-2005 parliament because the European elections of 2004 occasioned the exit of many first-placed candidates and the entry of many lower-placed candidates, thereby maximising variation in list position.
Next are the office-seeking variables. Government indicates membership of a governing party. Office-seeking deputies are predicted to have a lower switching hazard when their party is in government. Members records the number of members of the parliamentary party on the first day of the parliamentary term or on the day the party first appeared in official records. This is a proxy for legislative power and the rewards it may bring to ambitious deputies. The larger the caucus, the less likely office-seeking deputies are to switch. An alternative version of this variable was to include dummies for the official categories of caucus and group from the Sejm's standing orders. The raw count of members performed better in the following empirical models than these indicators. Another measure of legislative office-seeking would have been membership of legislative committees. Unfortunately, the Sejm only
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
Figure 2. Voting preferences, 1997-2001.
Notes: AWS = Solidarity Electoral Action; SLD = Alliance for the Democratic Left; PSL = Polish Peasant Party; UW = Freedom Union.
Figure 3. Voting preferences, 2001-2005.
Notes: SLD = Alliance for the Democratic Left; PSL = Polish Peasant Party; PO = Civic Platform; PiS = Law and Justice; Samoobrona = Self-Defence.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
849
Figure 4. Voting preferences, 2005-2007.
Notes: PiS = Law and Justice; PO = Civic Platform; Samoobrona = Self-Defence; LPR = League of Polish Families.
records these memberships at the beginning of each term, and since they change quite frequently, this would have been misleading.
Finally, we employ several measures of policy-seeking. Old party is a proxy for the clarity of a party's policy position. In newer parties, policy positions will be fuzzier and policy-seeking politicians are more likely to switch. In 2005, no new party surpassed the electoral threshold, but new parties were formed within the parliament itself. A similar logic underpins Experience. Deputies who have already served in the Sejm will be less likely to join a party that is incompatible with their policy preferences and should be less likely to switch. Right is a dummy variable, separating right-wing parties from others, as well as independents. The ideological vocabulary of Polish politics is unusual. The left-right dimension is much more strongly defined by socio-cultural than by socio-economic issues (Markowski 2006: 817-818; Szawiel 1999). The Polish right is relatively nationalist, religious and, especially, anti-(post) communist (Szczerbiak 2002). It is not necessarily committed to market economics. Indeed, Poland's most famous trade union, Solidarity, is 'right-wing'. Similarly, the left, is more coherent on socio-cultural than on socio-economic matters. However, its dominant representative, the post-communist Alliance of the Democratic Left, has increasingly insisted that it is a typical European social-democratic party. Centrist parties are genealogically anti-communist, but have
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
850
IAIN McMENAMIN & ANNA GWIAZDA
downplayed the relevance of the historic cleavage to contemporary politics. They have also tended towards economic liberalism. The Right variable can also be interpreted as a proxy for ideological or policy heterogeneity. The right's ideology has not given its adherents very helpful cues in relation to the pressing matters of economic and social policy that faced Poland during all four terms studied here. Moreover, the indivisible issues that define the right most closely pose difficulties for compromise. Such issues repeatedly caused conflict with rightist parties that became notorious for the bitter personalisation of politics (Kitschelt et al. 1999: 374). Policy-seeking deputies should be more likely to switch from these diverse parties than from the generally more coherent left and centre.
For the 1997-2001 and 2001-2005 parliaments, we are able to use a sharper variable to capture the incentives facing policy-seeking deputies. Heterogeneity is the mean standard deviation of expert responses to a series of eight questions on the policy positions of major parties contesting the 2001 parliamentary elections (Benoit & Laver 2006: 123-152; Depauw & Martin 2009). The higher the score, the more likely should policy-seeking deputies be to switch. This variable is missing for 6.9 per cent of observations in 1997-2001 and a much more serious 23.3 per cent in 2001-2005.
Mershon's approach to studying policy-seeking switches is to identify periods of heightened intensity of policy making in the legislature. Comparing switching rates in this period to other periods during a parliament provides a test of the importance of policy-motivated switching. We replicated Mershon's method in all five parliaments since 1991. In only one of the five parliaments did this approach provide any evidence of policy-seeking switches. The Appendix presents data by stage for all five parliaments, as well as contrasting data from Mershon and Shvetsova's work on Italy and Russia. Table 3 summarises the link between indicators and concepts.
Data analysis
We conducted parametric survival analyses on switching for the four parliamentary terms since 1993. We tested exponential, Gompertz, Weibull, log normal, log-logistic and generalised gamma for goodness of fit according to the Akaike Information Criterion across the four types of switching and the four parliaments. The Weibull model performed best by far. Thus, we present our analyses in the form of Weibull regressions in the relative hazard metric. Coefficients above one increase the hazard of switching, while coefficients below one reduce the hazard. Our results for overall switching are shown in
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
Table 3. Summary of indicators
Shared
Differing
o
o
Institution	institutional	institutional		Observable			Observable
(transaction)	elements	elements	Self-enforcement	Implication	Quasi-parameter	Self-reinforcement	implication
Party	Legislative	Aim of party	Defectors cannot	Less likely to	Popular	Parties that gain	Decreasing
(autonomy	party	is to maximise	run on the	switch from	support	re-election are more	volatility
for benefits		votes	party's ticket	popular parties. More		likely to retain	
of collective				likely to switch		popular support	
action)				if in lower list position		than new parties	
		Aim of party	Defectors	Less likely to	Bargaining	Governments	Return of
		is to control	cannot be	switch from governing	power	gain popular	governing
		maximum	ministers or	parties, large parties. More		support and	parties
		number of	committee chairs	likely to switch during		change the	
		ministries,		benefits stage		policy space	
		committee				to their	
		chairs, etc.				own advantage	
		Aim of party is	Defectors	More likely to	Public	Past policy	Parties with
		to influence policy	reduce the party's ability to influence policy	switch from new parties. Experienced deputies less likely to switch. More likely to switch from heterogeneous party. More likely to switch during policy stage	policy	influence changes the policy space and structure of public policy to make future policy influence more likely	past policy successes are more influential
a
W W
Vi O
>
Ö
m
H O
HH
2 m H
HH
H C| H
HH
o
2
> r
HH
CT)
%
HH
O 2
Source: Table adapted from Greif and Laitin (2004: 641).
5 t
oo
852
IAIN McMENAMIN & ANNA GWIAZDA
Table 4. This table employs only the variables that we have been able to measure for each of the post-1993 terms.
The Poll variable is the only consistent explanation for switching. It is statistically significant in the predicted direction during all four terms. It is strongest in 2005-2007 and next strongest in 1997-2001, with the remaining two parliaments at similar levels. The Government variable reduced switching during the two left-wing governments of 1993-1997 and 2001-2005, but actually increased switching during the right-wing governments from 2005-2007. Membership of an old party reduced switching in 2001-2005 and 2005-2007. In no term, were Experienced Deputies less likely to switch. Members of right-wing parties were much more likely to switch in 1993-1997, but not in any of the later terms. Indeed, they were less likely to switch in 2001-2005 and 2005-2007. Members (the size of parliamentary parties) never reduced the likelihood of switching.
This strong contrast between Poll and the other variables extends to different types of switching. For government to opposition switching, Poll was significant and in the right direction in 1997-2001 and 2001-2005. The number of members in the parliamentary party helped to explain defections from government to opposition in 2005-2007. In these equations, the Old party variable suffered from collinearity problems in all terms but 2005-2007 and the Experienced Deputy variable had to be dropped in 2005-2007. Neither was ever significantly in the right direction. It is unsurprising that no variable explained this type of switching from 1993-1997 as only nine switches from government to opposition occurred. Switching within the opposition was much more frequent. The Poll variable explains significant variation in these switches across all four terms, as does Right in 1993-1997 and 1997-2001. The Old party and Experienced Deputy variables were significantly in the right direction in 2005-2007 and 2001-2005, respectively. For switching from old to new parties, Poll is again consistently significant with the predicted sign. Government and Experienced Deputy reduced this type of switching in 1993-1997, while Right increased it during the same term. Members reduced switching from old to new parties in 2001-2005. These results are summarised in Table 5.
For the 2001-2005 parliament, we were able to add a dummy for List position number one. This is a powerful predictor for overall switching, without seriously attenuating the impact of Poll. This variable is not significantly associated with the other types of switching. For the 1997-2001 and 2001-2005 parliaments, we can use expert surveys to proxy the heterogeneity of some parties. For the former parliament, the variable has no effect. However, for the latter, in which many observations of this variable are missing, it is a powerful predictor. It does not substantially reduce the impact of the Poll variable.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
Table 4. Explanations of switching: Weibull regression in proportional hazard form
	1993-1997	1997-2001	2001-2005	2005-2007
Poll	0.13012 (0.04307)***	0.01185 (0.00617)***	0.1613 (0.02328)***	0.00239 (0.00247)***
Government	0.13194 (0.08936)**	1.1784 (0.85693)	0.0707 (0.0236)***	15.053 (6.5781)***
Old party	0.68696 (0.18418)	2.0004 (1.1078)	0.2928 (0.0442)***	0.10846 (0.04794)***
Experienced Deputy	0.77065 (0.1324)	1.2677 (0.24645)	0.9772 (0.0956)	1.0307 (0.2129)
Right	3.3709 (0.5995)***	1.26 (0.34387)	0.1858 (0.0405)***	0.0916 (0.04638)***
Members	1.0017 (0.00578)	0.99835 (0.00515)	1.008 (0.0023)***	1.0287 (0.00725)***
Shape parameter (p)	1.083 (0.12601)	0.6783 (0.10172)	1.408 (0.07602)	0.7481 (0.09727)
Log likelihood	-149.257	-285.412	-306.838	-160.243
LR Chi2(6)	550.06***	239.92***	785.44***	399.38***
Observations	45,280	45,627	46,420	23,060
a &
w w
o
> d
m
H O
HH
2 m H
HH
H C| H
HH
o
2
>
HH HH
O
2!
Note: * p < 0.05; ** p < 0.01; *** p < 0.001.
oo
854 IAIN McMENAMIN & ANNA GWIAZDA
Table 5. Summary of models explaining different types of switching across four parliamentary terms
	Government to Opposition	Within Opposition	From old to new
Poll	2/4	4/4	4/4
Government	NA	NA	1/4
Old party	0/1	1/4	NA
Experienced Deputy	0/3	1/4	1/4
Right	-	2/4	1/4
Members	1/4	0/4	1/4
Notes: The denominator is the number of terms for which the variable is included. The numerator is the number of terms for which the variable was statistically significant and in the predicted direction. The denominator for Government to Opposition varies since some variables had to be dropped because of collinearity. Right was collinear in all four terms.
Vote-seeking drove the likelihood of a switch across all four terms and across different types of switching. Office- and policy-seeking motivations were, at best, only intermittently important. Therefore, the Poll variable merits further exploration. The first way of doing this is to contrast hazard rates for different levels of the variable. Figures 5-8 show hazard rates for switching at the minimum and maximum values of Poll for each term, as well as 1 (no change in support since the election) and 0.5 (halving of support since the election). If the hazard rate declines over time, the Weibull function tends to produce an extreme prediction for the very earliest period, as seen in the 1997-2001 and 2005-2007 parliaments. This is especially true for low values of the Poll variable, and, of course, no Polish party lost large amounts of popular support in the few weeks between the election and the opening of parliament. Indeed, that the Poll variable is important in parliaments with quite different hazard functions is another indicator of the robustness of our conclusion regarding the importance of vote-seeking motivations for party switching.
In the 1993-1997 parliament, the hazard rate was almost flat over time for popular parties. For less popular parties, with a higher likelihood of switching, there was a modest increase in the hazard over time. The hazard rate of parties that suffered a halving of popular support is over twice that of parties that maintained support relative to the general election. There is an even greater contrast between parties that halved their support and the hazard rate for the minimum Poll score of only 0.17. The 1997-2001 parliament exhibits a different relationship between Poll values and switching. The overall hazard rate is
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
855
o c
ra X
O O
O
O O O
m
o o o
o
o -o
o O o o
Second Parliament
Time
Poll=0.17 ---
Poll=1 —
•— Poll=0.5 — Poll=1.7
Figure 5. Hazard function for switching, 1993-1997.
Note: Hazard function at different levels of the poll variable derived from model in Table 4.
Third Parliament
o
Time
Poll=0.06 -■ Poll=1 —
----Poll=0.5
----Poll=2
Figure 6. Hazard function for switching, 1997-2001.
Note: Hazard function at different levels of the poll variable derived from model in Table 4.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
856
IAIN McMENAMIN & ANNA GWIAZDA
Fourth Parliament
O -
o
in O O o
0
500
analysis time
1000
1500
Poll=0.11 Poll=1 —
----Poll=0.5
----Poll=3.63
Figure 7. Hazard function for switching, 2001-2005.
Note: Hazard function at different levels of the poll variable derived from model in Table 4.
Fifth Parliament
O 1
O
O
o
Time
Poll=0.09 -• Poll=1 —
■— Poll=0.5 - - Po!l=1.95
Figure 8. Hazard function for switching, 2005-2007.
Note: Hazard function at different levels of the poll variable derived from model in Table 4.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
857
o
,17 (Actual value)       .4 .6 .8 1
Figure 9. Simulations of the effects of minimum poll values on switching, 1993-1997. Note: Figures derived from separate simulations based on Model in Table 4. Each bar represents the number of deputies predicted to have switched parties by the end of the term according to different minimum values of the poll variable.
much higher and has a different shape, with the hazard initially falling quickly after the beginning of the term and falling much more slowly thereafter.
Except for the very early days, there is little difference between parties that have maintained or increased support and those that have seen their support halved. The real contrast is between the minimum value - a paltry 0.06 - and the others. In 2001-2005, the hazard rate was between that of the 1993-1997 and 1997-2001 parliaments. The hazard is relatively flat for parties with stable popularity, but increases markedly over time for parties that are losing popular support. Unlike the previous two terms, there are big differences between the hazard rates right across the range of poll values. The hazard function of the 2005-2007 parliament is reminiscent of 1997-2001. The hazard rate declines quickly and is only substantively affected by the Poll variable at very low values, which suggest the virtual obliteration of the party.
The models of the 1997-2001 and 2005-2007 terms have dramatic implications (Figures 9-12). As the simulations in Figures 10 and 12 suggest, had there not been really drastic collapses in the popularity of some parties, there would have been virtually no switching in either parliament. In the 1993-1997 and 2001-2005 parliaments, as illustrated in Figures 9 and 11, the effect of the Poll value is more gradual and extends further across the range. However, the
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
858
IAIN McMENAMIN & ANNA GWIAZDA
.06 (Actual value)       .4 .6 .8 1
Figure 10. Simulations of the effects of minimum poll values on switching, 1997-2001. Note: Figures derived from separate simulations based on Model in Table 4. The bar represents the number of deputies predicted to have switched parties by the end of the term according to different minimum values of the poll variable, o
o -
cm
.11 (Actual value)       .4 .6 .8 1
Figure 11. Simulations of the effects of minimum poll values on switching, 2001-2005. Note: Figures derived from separate simulations based on Model in Table 4. Each bar represents the number of deputies predicted to have switched parties by the end of the term according to different minimum values of the poll variable.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
859
o
o
« CM
o d
.09 (Actual value)       .4 .6 ,8 1
Figure 12. Simulations of the effects of minimum poll values on switching, 2005-2007. Note: Figures derived from separate simulations based on Model in Table 4. Each bar represents the number of deputies predicted to have switched parties by the end of the term according to different minimum values of the poll variable.
substantive implications are quite different because of the contrast between the huge numbers of switchers in 2001-2005 and the small number switching in 1993-1997. For 2001-2005, the model predicts a reduction of 67 in the number of deputies switching by the end of term, as the poll value moves from its actual minimum of 0.11 to a relatively stable 0.8. In 1993-1997, a move from the real minimum of 0.17 to 0.8 is predicted to result in 23 fewer switchers.
The institutionalisation parameter
In this section, we consider the progress of, and prospects for, legislative party institutionalisation in Poland, and its implications for comparative politics more generally. In doing so, we begin with Greif and Laitin's emphasis on parameters, or the conditions under which an institution can achieve self-enforcement. Then, we go on to analyse quasi-parameters. A quasi-parameter, like a parameter, sets the bounds of institutional self-enforcement. However, its values can be influenced by the process of institutional self-enforcement. If an institution can widen the parameter values under which it is able to enforce its basic transaction, then it can be described as a self-reinforcing institution.
Our results show that a legislative party in Poland is most likely to institutionalise if it convinces its members that it can deliver votes. Our simulations can
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
860
IAIN McMENAMIN & ANNA GWIAZDA
provide figures for the parameter that governs the ability to the legislative party to enforce itself as an institution for its deputies: 10 per cent of deputies switching during a term is a reasonable maximum for an institutionalised set of legislative parties. Indeed, Heller and Mershon (2009a: 11-13) report figures of over 10 per cent in systems that either have a reputation for under-institutionalised parties, or were undergoing systemic change. Using this figure, our models estimate a popular support parameter of 0.4 or less for all parliaments except for 2001-2005, for which the parameter is over one. In three out of four parliaments, Polish deputies did not switch until their parties had endured massive losses in popular support. Polish politicians are electorally sensitive, not hyper-sensitive.
Across the four parliaments since 1993, the mean minimum for our poll variable has been 0.1 - literally decimation! By comparison, take Canada, which in the 1990s exhibited unusual party system instability for an established democracy. The Progressive Conservatives who were, for one parliament, reduced to two deputies and took a number of parliaments to recover, scored a minimum of 0.26 on the poll measure in the 1988-1993 parliament and over 0.4 in the next two parliaments. Ireland is currently undergoing one of Europe's most dramatic economic crises. In 2009 it reported the largest budget deficit in the European Union, greater even than that of stricken Greece. The main governing party, Fianna Fail, has not scored less than 0.41 of its last general election result. These examples suggest that the rarity of party switching in most established democracies is not because they have a lower parameter of public support than Poland's younger democracy, but rather that levels of public support never fall outside the parameter values that underpin the basic transaction of the institution of the legislative party. However, the model of the 2001-2005 parliament implies a completely different situation, in which deputies panic and switch as soon as their party begins to lose popular support. If the parameter value of public support across parliaments were greater than one, then institutionalisation would be highly unlikely as all parties cannot continue in popularity.
If a party can reduce the parameter value under which it can enforce itself, it can be described as self-reinforcing. Parties that can prevent legislative switching for one term should be less vulnerable to the popular support parameter thereafter. Under this self-reinforcement mechanism, parties that achieve re-election to the parliament should be more likely to retain support than new parties. The most obvious indicator of this mechanism is volatility. There has been no clear decline, or even trend, in the Pedersen (1979) volatility index, or in Bartolini and Mair's (2007) bloc volatility, over the five elections (Gwiazda 2009: 358). Powell and Tucker (2008) have recently decomposed volatility into two types, which are particularly relevant for our purposes. Type A is volatility resulting from party entry and exit, while Type B is volatility among stable parties. Again, there is no trend, even in Type B, which reduces
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
861
confusion from phenomena such as party switching (Powell & Tucker 2008: 33). Of course, neither has there been a widening in the public support parameter, which would indicate that it might be a quasi-parameter. Therefore, we detect a robust enforcement mechanism in vote-seeking legislative parties, but these parties have not managed to adapt this mechanism to self-reinforcement. The vote-seeking legislative party appears to be a potential route of institutionalisation, but one that has not yet been successful.
The institutionalisation of Polish parties has been the subject of debate amongst country specialists. A few years ago, Szczerbiak (2007) and Markowski (2006) both expressed scepticism about the prospects for institutionalisation in the near future, but more recently, Gwiazda (2009) has argued that the Polish party system is undergoing institutionalisation. The theory and data presented in this article suggests that all analysts should be careful not to over-emphasise office-based variables such as the possible first re-election of a Polish government, or policy-based variables such as the relatively narrow terms of debate in the 2010 presidential election. Instead, we advise a focus on measures that matter to vote-seeking parties. The popular support of the principal parties in the 2007-2011 parliament has so far been stable (Centrum Badania Opinii Spolecznej 2010) and switching has been minimal. Threats to the ongoing self-enforcement of the legislative parties could come from a change in the parameter threshold for switching or from a sharp decline in popular support. Given that in only one parliament has the threshold crossed 0.4, this first scenario seems unlikely. However, catastrophic falls in popular support have been routine in Polish politics, so this is much more probable. One obvious cause of a sudden massive loss of popularity is a corruption scandal (Manys 2009), but the largest party in the current parliament seems to have ridden out one such crisis relatively unscathed (Manys 2010). If, as seems likely, the same parties are elected in similar strengths, there would be a fall in volatility and an indication that Polish parties have achieved a measure of self-reinforcement. It might be that Polish parties will finally move along the vote-seeking route from self-enforcement to self-reinforcement, and will thereby institutionalise.
Conclusions
Party institutionalisation is a central problem in political science. The existing literature tends to present institutionalisation as a syndrome rather than a causal model. Such syndromes offer us the opportunity to document variation but not to explain it. Moreover, the extant research on institutionalisation seems to underplay, or even ignore, differences between types of parties. We take an approach to party institutionalisation that is both concrete and flexible.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
862
IAIN McMENAMIN & ANNA GWIAZDA
It is concrete because we identify specific and measurable basic transactions of the legislative party, and its self-enforcement and self-reinforcement. It is flexible because we identify three routes to institutionalisation: vote-seeking, office-seeking and policy-seeking. In doing so, we are inspired by major literatures that have already proven their worth to political scientists.
We think our study of Poland demonstrates the potential of this approach to party institutionalisation. The volume of party switching and the variation in vote-, office- and policy-seeking incentives make Poland an apposite test case. We show that legislative party switching, which undermines the institutionalisation of legislative parties, is powerfully and consistently driven by vote-seeking. In contrast, office- and policy-seeking perspectives on legislative party switching in Poland are, at most, intermittently plausible. Polish deputies switch in order to secure their political survival, not out of short-term ambition for higher office or because of a chaotic and polarised policy space. Our empirical technique allows us to provide a quantitative estimate of a vital parameter of party institutionalisation: public support. In Poland, legislative parties can enforce basic discipline and survive if their public support exceeds 40 per cent of their vote at the previous parliamentary election. This, rather than the office- or policy-seeking channels, seems the most likely route for the institutionalisation of Polish parties.
This result has an important implication for comparative politics. All the measures used here should be easily replicable in other parliaments, with the exception of the ideological coding of right-wing parties. Moreover, the parameter value establishes an interesting hypothesis. Does party switching in other contexts have the same popular support threshold of 40 per cent of the previous election result? If so, it is the instability of popular opinion, rather than manoeuvring over governments and battles over policy, that threatens party institutionalisation. Conversely, parties that want to institutionalise need to reassure their deputies that voters will be retained, rather than delivering the spoils of office or public policy goods.
Acknowledgments
Earlier versions of this article were presented at the Sussex European Institute and the Centre for International Studies, Dublin City University. We are grateful to all participants, especially Tim Bale and Aleks Szczerbiak. We would also like to thank the editors and referees of EJPR for suggesting important improvements to our work.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
863
o o
<n i
in O
o
O
o
i
o o
o o
i
o o\ -* <n r-> o -* O o o
v~i o o co    (n '—I
o   o o
o r->
O <n
o o
r-   <n r->
O    <n <n
o   o o
rJ vi
o   o o
.2 ^
.a a
o Q
O
o
ON VO -*
-* <n -* o\ o   o   o o
O 00 CI
O
o   o   o o
o o
a 2
00
o o
■d
2 2
°-> rH
^ -3
o
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
864
IAIN McMENAMIN & ANNA GWIAZDA
References
Bartolini, S. & Mair, P. (2007). Identity, competition and electoral availability: The stabilisation of European electorates, 1885-1985. Colchester: European Consortium for Political Research.
Benoit, K. & Laver, M. (2006). Party policy in modern democracies. London: Routledge. Bertoa, F. (2009). Party system institutionalisation and the quality of democracy in Eastern
Europe. DISC Working Papers 2009/7. Budapest: Central European University, Centre
for the Study of Imperfections in Democracy. Centrum Badania Opinii Spolecznej. (2010). Prefencje Partyjne w Maju. Komunikat z Badah
BS/62/2010. Warsaw: Centrum Badania Opinii Spolecznej. Depauw, S. & Martin, S. (2009). Legislative party discipline and cohesion in comparative
perspective. In D. Giannetti & K. Benoit (eds), Intra-party politics and coalition government. London: Routledge. Desposato, S. (2006). Parties for rent? Ambition, ideology and party switching in Brazil's
Chamber of Deputies. American Journal of Political Science 50(1): 62-80. Gazeta Wyborcza. (2005). Polmilionowe weksle-straszaki Leppera niewazne. Gazeta
Wyborcza, 31 August.
Greif, A. & Laitin, D. (2004). A theory of endogenous institutional change. American Political Science Review 98(4): 633-652.
Gwiazda, A. (2009). Poland's quasi-institutionalized party system: The importance of elites and institutions. Perspectives on European Politics and Society 10(3): 350-375.
Heller, W. & Mershon, C. (2005). Party switching in the Italian Chamber of Deputies. Journal of Politics 67(2): 536-559.
Heller, W. & Mershon, C. (2009a). Introduction. In Political parties and legislative party switching. Basingtoke: Palgrave Macmillan.
Heller, W. & Mershon, C. (2009b). Integrating theoretical and empirical models of party switching. In Political parties and legislative party switching. Basingtoke: Palgrave Macmillan.
Heller, W. & Mershon, C. (2009c). Conclusions. In Political parties and legislative party
switching. Basingtoke: Palgrave Macmillan. Huntington, S. (1968). Political order in changing societies. New Haven, CT: Yale University
Press.
Janda, K. (2009). Laws against Party Switching, Defecting or Floor-crossing in National
Parliaments. Paper delivered at the World Congress of the International Political Science
Association, Santiago, 12-16 July. Kato, J. & Yamamoto, K. (2009). Competition for power: Party switching and party system
change in Japan. In W. Heller & C. Mershon (eds), Political parties and legislative party
switching. Basingtoke: Palgrave Macmillan. Kitschelt, H. et al. (1999). Post-communist party systems: Competition, representation and
inter-party competition. Cambridge: Cambridge University Press. Krupavicius, A. (2004). Party systems in Central East Europe. In, Lithuanian political science
yearbook. Vilnius: Vilniaus universiteto leidykla/VU Tarptautiniy santykiu ir politikos
moksly institutas.
Mainwaring, S. (1999). Rethinking party systems in the third wave of democratization: The
case of Brazil. Stanford, CA: Stanford University Press. Mainwaring, S. & Scully, T. (eds) (1995). Building democratic institutions. Party systems in
Latin America. Stanford, CA: Stanford University Press.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
THREE ROADS TO INSTITUTIONALISATION
865
Manys, K. (2009). Opozycja: komisji nie starczy czasu. Rzeczpospolita, 5 November.
Manys, K. (2010). Szybki final czy dalsze sledztwo. Rzeczpospolita, 7 July.
Markowski, R. (2001). Party system institutionalisation in new democracies: Poland - A
trend-setter with no followers. In PG. Lewis (ed.), Party development and democratic
change in post-communist Europe. London: Frank Cass. Markowski, R. (2006). The Polish elections of 2005: Pure chaos or a restructuring of the party
system? West European Politics 29(4): 814-832. McMenamin, I. (2008). Semi-presidentialism and democratisation in Poland. In R. Elgie & S.
Moestrup (eds), Semi-presidentialism in Central and Eastern Europe. Manchester:
Manchester University Press. Mershon, C. (2009). Legislative party switching and executive coalitions. Japanese Journal of
Political Science 9(3): 391-414. Mershon, C. & Shvetsova, O. (2008). Parliamentary cycles and party switching in legislatures.
Comparative Political Studies 41(1): 99-127. Pedersen, M. (1979). The dynamics of European party systems: Changing patterns of electoral volatility. European Journal of Political Research 7: 1-26. Powell, E. & Tucker, J. (2008). New Approaches to Electoral Volatility: Evidence from
Postcommunist Countries. Paper presented to the Annual Meeting of the American
Political Science Association, Chicago, IL, 3-6 April. Randall, V. (2006). Party Institutionalization and Its Implications for Democracy. Paper
presented to the World Congress of the International Political Science Association,
Fukuoka, 9-13 July.
Randall, V. & Svasand, L. (2002). Party institutionalization in new democracies. Party Politics 8(1): 5-29.
Reed, S. & Scheiner, E. (2003). Electoral incentives and policy preferences: Mixed motives behind party defections in Japan. British Journal of Political Science 33(3): 469-490.
Reilly, B. (2008). Introduction. In B. Reilly & P. Nordlund (eds), Political parties in conflict-prone societies: Regulation, engineering and democratic development. Tokyo: United
Nations University Press. Sanford, G. (2002). Democratic government in Poland. Basingstoke: Palgrave Macmillan. Sejm Rzeczypospolitej Polskiej. (1992). Regulamin Sejmu Rzeczypospolitej Polskiej.
Warsaw: Sejm Rzeczpospolitej Polskiej. Shabad, G. & Slomczynski, K. (2004). Inter-party mobility among parliamentary candidates
in post-communist East Central Europe. Party Politics 10(2): 151-176. Shugart, M.S. (2005). Comparative electoral systems research: The maturation of a field and
new challenges ahead. In M. Gallagher & P. Mitchell (eds), The politics of electoral
systems. Oxford: Oxford University Press. Sikk, A. (2005). How unstable? Volatility and the genuinely new parties in Eastern Europe.
European Journal of Political Research AA: 391-412. Str0m, K. (1990). A behavioral theory of competitive political parties. American Journal of
Political Science 34(2): 565-598. Szawiel, T. (1999). Zroznicowanie w Lewicowo-Prawicowe i Jego Korelaty. In R. Markowski
(ed.), System Partyjny, Postawy Polityczne, Zachowanie Wyborcze. Warsaw: Institute of
Political Studies.
Szczerbiak, A. (2001). Poles together? The emergence and development of political parties in post-communist Poland. Budapest: Central European University Press.
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research
866
IAIN McMENAMIN & ANNA GWIAZDA
Szczerbiak, A. (2002). Dealing with the communist past or the politics of the present? Lustration in post-communist Poland. Europe-Asia Studies 54(4): 553-572.
Szczerbiak, A. (2007). 'Social Poland' defeats 'liberal Poland'? The September-October 2005 Polish parliamentary and presidential elections. Journal of Communist Studies and Transition Politics 23(2): 203-222.
Zielinski, J., Slomczynski, K. & Shabad, G. (2005). Electoral control in new democracies: The perverse incentives of fluid party systems. World Politics 57(1): 365-395.
Address for correspondence: Iain McMenamin, Centre for International Studies and School of Law and Government, Dublin City University, Dublin 9, Ireland. Tel.: +353-1-7008072; Email: iain.mcmenamin@dcu.ie
© 2010 The Author(s)
European Journal of Political Research © 2010 European Consortium for Political Research