Content uploaded by Jan Rosset
Author content
All content in this area was uploaded by Jan Rosset on Jan 12, 2015
Content may be subject to copyright.
International Political Science Review
2015, Vol. 36(1) 78 –98
© The Author(s) 2013
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0192512113498830
ips.sagepub.com
Mind the gap: Do proportional
electoral systems foster a more
equal representation of women
and men, poor and rich?
Julian Bernauer
University of Berne, Switzerland
Nathalie Giger
University of Mannheim, Germany
Jan Rosset
University of Lausanne, Switzerland
Abstract
Female gender and low income are two markers for groups that have been historically disadvantaged
within most societies. The study explores two research questions related to their political representation:
(1) ‘Are parties biased towards the ideological preferences of male and rich citizens?’; and (2) ‘Does the
proportionality of the electoral system moderate the degree of under-representation of women and poor
citizens in the party system?’ A multilevel analysis of survey data from 24 parliamentary democracies indicates
that there is some bias against those with low income and, at a much smaller rate, women. This has systemic
consequences for the quality of representation, as the preferences of the complementary groups differ. The
proportionality of the electoral system influences the degree of under-representation: specifically, larger
district magnitudes help in closing the considerable gap between rich and poor.
Keywords
Gender inequality, ideological congruence, income inequality, multilevel regression, proportional representation
Introduction
In an ideal democracy, every citizen, regardless of differences in, say, income or gender, is endowed
with equal political voice. In reality, money and other unequally distributed resources heavily
Corresponding author:
Julian Bernauer, Institute of Political Science, University of Berne, Fabrikstrasse 8, 3012 Berne, Switzerland.
Email: julian.bernauer@ipw.unibe.ch
498830IPS36110.1177/0192512113498830International Political Science ReviewBernauer et al.
2013
Article
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 79
influence processes of political decision-making. Social status is related to political participation
(Lijphart, 1997), potentially further widening gaps in representation between ‘rich’ and ‘poor’.
Similarly, a gender gap could result from the continuing male dominance in political leadership,
despite decades or even centuries of movements for women’s rights. We join the resulting debate
that has emerged around the question of whether representatives are selectively responsive to spe-
cific subgroups of the citizenry.
Most of this kind of research, in particular, on income and political inequality, has heavily
focused on the US case (Bartels, 2008; Enns and Wlezien, 2011; Gilens, 2005; Griffin and Newman,
2007; Hacker and Pierson, 2010; Jacobs and Page, 2005; Schlozman et al., 2012; Soroka and
Wlezien, 2008; Ura and Ellis, 2008). The study at hand widens the scope both geographically,
covering 24 Western democracies, and thematically, addressing the role of the institutional context
for the quality of representation. Within the field of subordination, we study the ideological prefer-
ences of two historically disadvantaged groups (Phillips, 1995; Piven and Cloward, 1977): women
and less affluent citizens.
Not all citizens are able to make their voice heard in the political process, and economic or
class-based inequalities, in particular, are reinforced through biases in political representation,
organisation and policy (Bartels, 2008; Hacker and Pierson, 2010; Schlozman et al., 2012). US
case studies have documented that both policy output (Gilens, 2005, 2012) and the behaviour of
individual legislators (Bartels, 2008) tend to mirror the interests and preferences of rich citizens
more closely than those of the poor. Individual participation and organised interests are equally
biased towards the more educated and affluent (Schlozman et al., 2012). Similarly, there is a grow-
ing literature focusing on the substantive representation of women’s interests and the possible link
to the number of women in the legislature (see, e.g., Ruedin, 2013; Sawer, 2012; Schwindt-Bayer
and Mishler, 2005; Studlar and McAllister, 1996; Wängnerud and Sundell, 2012). These studies
provide initial support for the thesis that representative bodies tend to be biased towards the prefer-
ences of men and rich citizens.
In line with the literature on party–citizen linkages (Adams and Ezrow, 2009; Dalton, 1985;
Ezrow, 2010; Thomassen, 1994), we argue that one of the main channels for the representation of
citizens’ ideological or policy preferences is political parties and that the party system ‘makes it
possible for citizens to express their preferences’ (Powell, 2004: 93). Building on this reasoning,
we study the substantive representation of these subgroups by analysing their ideological congru-
ence with political parties. This approach has the advantage of understanding political parties as
intermediaries of preferences, which has been little explored so far in the study of inequalities in
representation (however, see Adams and Ezrow, 2009; Giger et al., 2012) despite the importance
of party–citizen linkages in the literature on substantive representation (see, e.g., Ezrow, 2010;
Thomassen, 1994). Furthermore, our broad comparative focus allows studying the role of political
institutions. We believe that electoral rules are important moderators in the translation of prefer-
ences into effective representation. While an extant literature is dealing with the general question
of whether proportional electoral rules foster closer congruence between citizens and elites (Blais
and Bodet, 2006; Golder and Stramski, 2010; Powell, 2000), less is known about the impact of the
institutional setting on the ideological representation of subgroups.1 Following a large body of lit-
erature emphasising the fairness and inclusiveness of proportional representation (PR; for a review,
see Blais, 1991), we argue that proportional electoral systems favour a more equal representation
of women and the poor. PR allows for more parties, which can and will spread across the whole
ideological range (Golder and Stramski, 2010), and, hence, for more (viable) parties on the Left –
the political ‘corner’ typically preferred by the groups of interest (see below). Furthermore, and
flying under the radar in this analysis, the larger district magnitudes under PR (and the size of a
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
80 International Political Science Review 36(1)
party’s district delegation; see Matland, 1993) encourage the nomination and election of more
female candidates (Kostadinova, 2007; Matland, 1998), and likely more balanced sets of candi-
dates in general.
The results of the quantitative-empirical analysis indicate that poor citizens, in particular, share
distinct left-leaning preferences, and suffer from bias against these. Most notably, proportional
electoral rules and, specifically, larger district magnitudes reduce the under-representation of the
poor. The picture is much less clear-cut in the case of women. Their level of ideological congru-
ence is only slightly and statistically insignificantly lower than that of men on average. Although
levels of relative representation of women differ clearly across countries, this limited variation
cannot be attributed to the influence of the electoral system. In sum, there appears to be some lee-
way for institutional design to reduce gaps in substantive representation between the rich and poor.
The ideological representation of women and the poor
One of the recent trends in the study of representation has been to disentangle constituency repre-
sentation and analyse how well smaller groups within the population, in other words, sub-constit-
uencies (Adams and Ezrow, 2009), are represented. This literature has focused mainly on the
substantive representation of two broadly defined groups of citizens: those who were thought of as
most influential (see Adams and Ezrow, 2009; Ezrow et al., 2011; Griffin and Newman, 2005;
Jacobs and Page, 2005) and disadvantaged ones (Bartels, 2008: 252–282; Gilens, 2005; Griffin and
Newman, 2007; Schwindt-Bayer and Mishler, 2005). In this article, we take the second perspective
and focus on the ideological representation2 of two core groups in this respect, namely, women and
poor citizens.3 They share features that make their study particularly relevant in terms of represen-
tation and democracy. These particular groups have historically been excluded from the democratic
process and, notably, from the right to vote. Beyond female disenfranchisement, many of the early
modern democratic institutions excluded the poor from elections. Examples are electoral rules that
required the payment of a tax for voting (Belgium) or being a property owner (UK, USA). Some
electoral systems also gave differential weight to citizens’ votes depending on the income group to
which they belonged (Prussia).
Further, there is reason to expect that poor citizens and women differ in their preferences from
the rest of the population in the sense that they (to varying degrees and with exceptions) share simi-
lar (positive) attitudes towards redistribution and the welfare state, mapping on left-leaning ideo-
logical positions. Hence, their lack of representation in the party system can have severe
consequences for the quality of representation at the aggregate level (Griffin and Newman, 2007:
1037). Recent work about the preference distributions of the rich and poor underpins that those
with low income tend to differ in their preferences, especially with regard to welfare and redistribu-
tion (Cusack et al., 2008; Enns and Wlezien, 2011; Gilens, 2009; Headey and Muller, 1996). These
findings at least partly support the established Meltzer–Richard model (Meltzer and Richard,
1981), which assumes that poor citizens favour redistributive policies up to the point where the
benefit for them is outweighed by the efficiency costs of taxation.4 There is also some evidence that
despite the heterogeneity of women’s social status, differences between the genders in terms of
broad ideological orientations exist. Recent evidence suggests that in many industrialised coun-
tries, women share more left-leaning attitudes than men and also vote in higher proportions for
left-of-centre parties (Bergh, 2007; Giger, 2009; Inglehart and Norris, 2000; Manza and Brooks,
1998; Studlar et al., 1998). The literature further documents that women have more positive atti-
tudes towards redistribution and social security than men (Cusack et al., 2008; Gidengil et al.,
2003). Studies in a political-economy tradition explain this as an insurance against the high risk of
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 81
divorce and the resulting drop in income and dependency on the welfare state (Iversen and
Rosenbluth, 2006), while others point to the influence of structural factors or background variables
(Bergh, 2007; Inglehart and Norris, 2000; Studlar et al., 1998). In total, there is reason to expect
that low-income groups and – likely to a smaller extent5 – women have preferences towards more
redistribution and extensive social security. As these two claims are tightly linked to the Left–Right
cleavage in Western societies, we expect that they have preferences to the left of the rest of the
population. This creates the potential for bias in representation, depending on the ideological posi-
tions of political elites.
A number of recent studies address the question of whether US representatives are unequally
responsive towards citizens with different economic resources. While evidence is generally mixed,
a general conclusion is that when there are differences in policy preferences between income
groups, the views defended by the rich are better heard. Indeed, while some studies found limited
evidence for systematic differences in policy preferences across income groups (Soroka and
Wlezien, 2008; Ura and Ellis, 2008) and consequently little leeway for the thesis that the US leg-
islature is more responsive to the rich, Gilens (2005, 2012) and Bartels (2008: 252–282) unearth a
systematic bias of representatives towards citizens with higher income. The main explanations for
such a pattern include differences in political participation rates across income groups and their
diverse financial contributions to parties (Schlozman et al., 2012), as well as the fact that elected
representatives tend to be recruited from the upper social classes and understand better the needs
of their affluent voters, with whom they share a similar lifestyle (Mansbridge, 1999).
Recent studies dealing with the substantive representation of women have focused on the impact
of the proportion of women in parliaments on policymaking. Research has notably shown a link
between the gender of parliamentarians and their attitudes and behaviour in parliament (Evans,
2008; Studlar and McAllister, 1996) and has also demonstrated that the number of women in par-
liament has policy consequences (Schwindt-Bayer and Mishler, 2005; Wängnerud and Sundell,
2012). This literature provides evidence for a link between the descriptive and substantive repre-
sentation of women,6 indirectly showing that given the generally low number of women sitting in
parliament, women’s preferences are likely to be under-represented in policymaking. Less work is
directly concerned with the question of whether women suffer in general from lower ideological
congruence than men, although there is some affirmative evidence (Lovenduski and Norris, 2003).
In sum, based on the existing literature on the representation of these subgroups of the population,
we hypothesise a negative relationship between the individual features of female gender and low
income, on the one hand, and ideological congruence on a Left–Right dimension, on the other:
H1a: Female and poor citizens have lower levels of ideological congruence with political parties on a
Left–Right scale than male and rich citizens.
While the existing literature encourages the formulation of such hypotheses about a relationship
between low income or female gender and political representation, it generally lacks an integrated
perspective on the overlap between the two features and group heterogeneity (Dovi, 2007). This
study takes a different approach and acknowledges that the two subgroups covered here are neither
completely overlapping nor completely independent from one another. It is well known that gender
is a socially cross-cutting factor and while women are present in all societal classes, there is reason
to expect that women constitute a larger share of the poor than men. In fact, the differences in their
socio-economic background are listed as one of the driving forces behind the tendency of women
to have more leftist attitudes than men (see, e.g., Inglehart and Norris, 2000; Studlar et al., 1998).
Hence, we expect some of the potentially leftist positions of women to be explained by social
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
82 International Political Science Review 36(1)
factors such as income, but also maintain that, additionally, gender-specific life risks independent
of social status provide reasons for more leftist attitudes. In sum, it makes sense to account for
income and gender simultaneously in order to disentangle the relative effects of both features. We
see little reasons for postulating an interactive relationship, though, as both gender and income
induce similar preferences and neither offset nor necessarily reinforce each other. Still, the latter
possibility – that poor women are represented worse beyond a mere additive effect between income
and gender – appears slightly more likely and will be tested empirically:
H1b: Female poor citizens are particularly prone to have lower levels of party-ideological congruence on
a Left–Right scale than other voters.
Electoral rules and the ideological representation of women
and the poor
The purpose of this study is not only to describe the possible under-representation of certain groups
in the context of Western democracies, but also to examine whether the electoral system systemati-
cally moderates levels of under-representation. Such institutional factors can only be studied in a
cross-national perspective. We build on the debate as to whether PR systems foster ideological
congruence (Blais and Bodet, 2006; Golder and Stramski, 2010; Huber and Powell, 1994; Powell,
2000) between citizens and elites. The answer very much depends on the concept of congruence
chosen; in particular, there is increasing evidence that while PR fosters closer congruence between
citizens and legislatures compared with majority or plurality rule, there are little differences at the
government level due to the logic of coalition-building under PR (Blais and Bodet, 2006; Golder
and Stramski, 2010). Previous studies have mainly focused on congruence between governments
and the median voter (Esaiasson and Holmberg, 1996; Powell, 2000). We approach the topic from
a different angle, emphasising the role that parties and the party system play in the process of the
representation of interests. Particularly in a European context, parties are important vehicles for the
representation of citizens’ preferences, and the representation of citizens’ interests in the party
system is, hence, an important criterion to judge representational quality (Powell, 2000). Given our
party-centred focus on ideological congruence, we posit that PR systems perform better in the
representation of subgroups. While the current literature points towards the view that PR produces
a better descriptive representation of women (Kenworthy and Malami, 1999; Matland, 1998;
Matland and Studlar, 1996; Norris, 2006; Paxton, 1997; Ruedin, 2013; Siaroff, 2000; Studlar and
McAllister, 2002) or ethnic minority groups and their parties (Kostadinova, 2007), we expand this
argument to the ideological representation of women and the poor.
PR, we argue, moderates the extent to which inequalities in representation between individual
voters and parties arise, as such electoral rules can be understood as ‘permissive’ (Cox, 1997: 142;
Moser and Schreiner, 2012: 182) or ‘weak’ (Taagepera, 2002). They foster the entry and success of
smaller parties (Cox, 1997; Duverger, 1963), spreading the ideological range and yielding a legis-
lature closely mirroring the preferences of citizens (Golder and Stramski, 2010). On the basis of
models of spatial competition (Downs, 1957), we assume that party elites behave in an instrumen-
tally rational way in seeking votes by offering voters corresponding policy or ideological positions
(Cox, 1990; Strøm and Mueller, 1999). The structure of the electoral system is expected to influ-
ence party competition and levels of (subgroup) ideological representation by altering the incentive
structure for political elites (Blais and Bodet, 2006; Golder and Stramski, 2010; McDonald et al.,
2004; Powell, 2000). In this model, the electoral structure governs the number of parties competing
for voters and their relative ideological positioning.7 As long as a new party does not locate itself
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 83
exactly at the position of an existing party and stays within the distribution of voters, its addition
to the ideological space has the immediate mechanical effect of enhancing congruence between
voters and parties as one more position is available. We also expect a strategic effect under PR, as
parties will choose their positions in order to distinguish themselves from other parties by also tak-
ing extreme positions (see, e.g., Calvo and Hellwig, 2011). It is well known that unlike under
majoritarian rules, where two parties ideally converge towards the median voter (Duverger, 1963),
there is no clear equilibrium of positions in the multi-party context. Still, we expect the multiple
parties under PR to occupy alternative positions, hence covering the preferences of most citizens.
In this respect, in particular, disadvantaged groups benefit from PR. Given their weaker voice due
to lacking resources and lower levels of participation (Lijphart, 1997), these groups are less attrac-
tive compared with others for mainstream parties but arguably represent a worthwhile potential
electorate for smaller parties. Hence, the subgroups are not likely to be targeted unless the ideologi-
cal space gets crowded by parties seeking a niche. If the subgroups have distinct preferences, and
additional parties cater for these, their ideological representation is enhanced.
In addition to these considerations, poor and female citizens might also benefit from the change
in coalition patterns resulting from an increasing number of parties. Iversen and Soskice (2006)
have shown that moving from a two- to a multi-party system changes coalition strategies of centrist
actors in favour of the poor. In their model, the centrist voters who belong to the middle class have
less bargaining power under majority or plurality voting compared with PR. In majoritarian sys-
tems, because they cannot rely on a committed party, they are more inclined to vote strategically
and ally with the rich for fear of being exploited by the poor. This has as a consequence that the
‘political space in majoritarian systems is tilted to the right’ (Iversen and Soskice, 2006: 176).
Thus, PR not only results in parties filling the positions that are not yet taken in the ideological
space, but is also likely to produce a leftward readjustment of the party system – which is to the
favour of the two sub-constituencies studied if they have preferences to the left of the rest of the
population. In sum, the gap between women and men, and the poor and rich, should close to a
certain extent under PR. This explanation yields a cross-level interaction between subgroup status
and electoral context:
H2: The more proportional the electoral rules, the lower the under-representation of female and poor
citizens.
Design, data and model
To assess the ideological representation of sub-constituency interests, a multilevel analysis using
data from the Comparative Study of Electoral Systems (CSES) is performed. Twenty-four demo-
cratic, non-presidential systems from the first two waves of the CSES are chosen, covering 45
elections between 1996 and 2006.8 The countries covered are: Australia, Belgium, Bulgaria,
Canada, the Czech Republic, Denmark, Finland, Germany, Hungary, Iceland, Ireland, Israel, Italy,
the Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Slovenia, Spain, Sweden,
Switzerland and the UK. Within these countries, the CSES survey data cover several tens of thou-
sands of respondents, which are the basic units of analysis.
We rely on an established measure of ideological representation, which Achen (1978) calls the
‘proximity’ between political elites and constituencies. We are interested in whether the preferences
of subgroups are farther from the parties than are the rest of the population. This focus on relative
differences carries the advantage that it eases many of the problems associated with measurements
of ideological congruence and the interpretation of absolute values of ideological distance measures.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
84 International Political Science Review 36(1)
In our design, we only have to assume that the measures of ideological positions are likely to be
equally flawed across societal groups. Therefore, if we find that one societal group, for example,
women, has a larger distance to the party system compared with the rest of the population, we can
at least say that they are relatively worse represented.
The dependent variable in the analysis is the ideological congruence between individuals and
the party closest to their preferences (i.e. the party whose absolute distance to that individual is
smallest) on a Left–Right scale, and, hence, is measured at the individual level. The data on indi-
vidual ideological positions is taken from the CSES, where respondents are asked to place them-
selves on a scale from 0 to 10 where 0 means ‘Left’ and 10 means ‘Right’.9 For measuring the
preferences of parties, we use CSES expert judgements10 of party positions on the very same scale,
which are included in the CSES data set.11 We assume that both scales are quasi-continuous. The
individual ideological congruence Cij is defined as the absolute distance between the position of
citizen Pij and the position of the party most proximate to her own position Pj, multiplied by −1:
Cij = −min| Pj – Pij |
The multiplication by −1 eases the interpretation of the indicator. If the distance between a citizen
and the most proximate party gets smaller, Cij rises from negative values towards 0 and ideological
congruence increases.
The measure of ideological congruence as the distance between a citizen’s own position and that
of the closest party has several advantages compared to measures such as the distance to the party
voted for. First, it reflects the theoretical concept of ideological representation by highlighting its
‘systemic’ component (Pitkin, 1967).12 It allows us, second, to include the whole population in the
model and not only those citizens turning out to vote or indicating a party preference. And, third,
we also believe that by modelling congruence at the individual level (see also Griffin and Newman,
2007), we gain new leverage on representational quality as we include not only the average or
median preference, but also information about the distribution of the preferences. Golder and
Stramski’s (2010) measure of absolute congruence is similar to our measure as both capture the
distribution of preferences rather than relying on a point estimate. However, since we are interested
in group differences, our estimate is on the individual rather than the aggregate level.
The focal independent variables are the identifiers for sub-constituency status. Straightforwardly,
gender is coded 1 for women and 0 for men. Low-income groups are defined as the lower two
income quintiles (1) against the higher three quintiles (0), allowing cross-country comparisons (for
a similar approach, see Bartels, 2008).13 The last step in the analysis concentrates on system-level
explanations for gaps in representation between groups. The electoral structure has been discussed
as a moderating factor that increases ideological congruence in particular for women and the poor.
Instead of output measures such as the vote–seat proportionality of electoral results (Gallagher
index, see Lijphart, 2012), a more fine-grained input measure capturing the institutional ‘permis-
siveness’ (Moser and Schreiner, 2012) or ‘strength’ (Taagepera, 2002) of electoral systems is used
to operationalise the proportionality of electoral systems. To this end, we follow Cox (1997: 208),
who introduces two variables that capture the effects of lower and (if any) upper tiers of the elec-
toral system separately. The lower-tier variable is the log median legislator district magnitude.14
Cox (1997: 209) justifies this measure by arguing that the strength of electoral systems is best
captured using the district magnitude that is most common among legislators. A second factor
influencing the proportionality of the electoral system is possible upper tiers that add proportional-
ity to the electoral outcome. This variable equals the ‘percentage of all assembly seats allocated in
the upper tier(s) of the polity’ (Cox, 1997: 209). Typically, mixed electoral systems would appear
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 85
overly disproportional, relying only on district magnitude. District magnitude and the seat share
allocated in upper tiers are inversely related in the sample at hand. Systems with additional tiers of
varying size and effect are those of Denmark, Germany, Hungary, Iceland, Italy, New Zealand,
Norway, Poland and Sweden. Additionally, in line with previous work (see, e.g., Blais and Bodet,
2006), we control for the age of democracy, based on the observation that young democracies
might suffer from generally lower levels of ideological representation due to reasons such as that
the party system is yet to be consolidated.
Multilevel models are the method of choice to adequately treat the data at hand, where individu-
als are nested in countries (Steenbergen and Jones, 2002). The method also enables us to adequately
test the interactive hypothesis, which postulates moderating effects of system characteristics on the
relationship between gender or income and ideological representation. The structure of our two-
level linear multilevel model is described as follows:
Level 1:
CA
ik kkik ik
=+ +
ππ ε
01
Level 2 random intercept:
πγγµ
00 0kkk
W=+ +
Level 2 random slopes:
πµ
10 1
kk
k
W=+ +
φφ
The model explains individual ideological congruence (Cik) of citizen i in country k by individ-
ual-level variables (Aik, their estimates π respectively) and country-level15 variables (Wk, their esti-
mates γ respectively). Furthermore, the model also includes variance terms on all levels (εik, µok)
which are assumed to be normally distributed. To model interactions, random slopes 1k for one or
several individual-level variables Aik are modelled as a function of a mean effect
φ
0, cross-level
interactions between country-level variables Wk and individual-level variables Aik (where
φ
is a
vector of coefficients), and a variance term µ1k. The descriptive and quantitative results are pre-
sented in the subsequent section.
Results
As a first step of our analysis, we focus on the question of whether the preferences of the two
groups studied are different from those of the rest of the population. Without differences in prefer-
ences between groups, there can be no under-representation of one group at the macro level. Our
theoretical expectation is that the two groups under consideration show more leftist attitudes than
the rest of the population due to their specific social positions. Figures 1 and 2 present the distribu-
tions of plain preferences of women versus men and the poor versus the rich, separated by
country.
The two figures show that the distributions of the preferences are not totally congruent with a
tendency towards more leftist preferences (negative gap). Women have on average more leftist
positions than do men; in our sample, this is the case in 17 out of the 24 countries. However, we
also find instances where the reverse is true and female voters on average are to the right of males,
for example, in the case of Finland. But largely, the findings are in accordance with the literature,
which predicts a modern gender gap regarding preferences and voting (i.e. women leaning more to
the left) in industrialised countries. Similarly, and slightly more pronounced, lower-income groups
are less rightist than the higher-income groups (about .17 on average, significant at p < .01). Again,
this is true for most cases in our sample, but not a universal phenomenon: in Slovenia, Ireland,
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
86 International Political Science Review 36(1)
Switzerland and Portugal, we even observe low-income voters to be slightly more rightist than
high-income voters. These diverging cases might be explained by the specific meaning of Left/
Right in a country (where Ireland is known as a special case), or by the strength of populist parties
finding support among the poor (e.g. note the difference in Switzerland to the representation of
women). In sum, though, we find support for our expectation of female and low-income group
preferences gravitating towards the Left, with the indication that an under-representation of these
societal groups is also relevant at the macro level.
The next step of our analyses is to test whether the two groups are also further away from the
party system than their counterparts. An ‘empty’ model without covariates (Table 1, Model 0)
Figure 1. Distribution of preferences by gender separated per country.
Note: Negative numbers indicate women more to the left than men; positive numbers the opposite. Grey bars indicate
non-significant differences between the two groups.
Figure 2. Distribution of preferences by income separated per country.
Note: Negative numbers indicate low-income citizens more to the left than high-income citizens; positive numbers the
opposite. Grey bars indicate non-significant differences between the two groups.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 87
shows a considerable variance share at the country level (9%), indicating that representation not
only is determined at the individual level, but also has a systematic context-level component.
Model 1 introduces the effect of being a female and of low income on ideological congruence. In
Model 1 (testing H1a), we see that low income reduces individual ideological congruence by .10
(statistically significant at p < .01), while female gender has a smaller negative effect of .01, which
is not significant across the whole sample. Notably, both variables have country-specific effects, as
their random slopes indicate. A more detailed assessment of the representation of the two sub-
groups is presented in Figure 3, where the average effects of being poor or female are displayed per
country.
Here, it becomes visible that the under-representation of the poor is a very consistent phenom-
enon: there is not a single country in our sample where low-income citizens are better represented
than their high-income counterparts. The situation is different for women: while we find a consid-
erable lower representation, that is, a negative coefficient of being female, in some countries (e.g.
Germany, Switzerland and Sweden), there is no universal pattern, and, in some countries, such as
Bulgaria or Norway, women are even slightly better represented than men by political parties on
the Left–Right dimension. This diverse situation of female representation is also the reason why
the coefficient on female gender in Model 1 (Table 1) is not pointing in one direction in a statisti-
cally significant fashion.
Although one might argue that the effects appear to be limited in absolute terms, we believe that
they should be regarded as substantive when related to the average distance between voters and the
closest party – particularly, if we take into consideration that citizens belonging to middle- and
high-income categories are on average only 0.57 points and 0.55 points, respectively, away from
the closest party. Low-income citizens are on average 18% further away from their party than the
more affluent. The effect is much less considerable in the case of women (around 4% further dis-
tance as compared with men).
In Model 2 (Table 1), an interaction between the status of low income and female is introduced
based on the discussion of a reinforcing pattern of being female and poor at the same time (H1b).
Table 1. Empty model and sub-constituencies (dependent variable: individual ideological congruence).
Model 0 Model 1 Model 2
Individual level
Female −.01 (.01) −.01 (.01)
Low income −.10 (.02)*** −.10 (.02)***
Female × low income .01 (.01)
Constant −.62 (.02)*** −.58 (.05)*** −.58 (.05)***
Random intercepts
Variance (countries) .062 (.018) .053 (.016) .053 (.016)
Variance (individuals) .564 (.003) .559 (.004) .559 (.004)
Random slopes
Variance (female) .003 (.001) .003 (.001)
Variance (low income) .005 (.002) .005 (.002)
N (countries) 24 24 24
N (individuals) 50675 50675 50675
Test vs linear reg. (p) .00 .00 .00
Notes: *** p < .01. Standard errors in parentheses. All calculations were performed in Stata 12, using the xtmixed
command with the reml (restricted maximum likelihood) algorithm.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
88 International Political Science Review 36(1)
Empirically, we find no such effects. The absence of an interactive pattern does not preclude the
existence of additive effects of under-representation. Model 2 predicts that female poor citizens
additively have the lowest scores of representation by parties.
Table 2 explores how subgroup representation depends on the country context, and more pre-
cisely, on the proportionality of the electoral system (H2). This is encouraged by the considerable
variation of the effects of both low income and female gender across countries. Model 3 reports the
corresponding findings with regard to the electoral system and its cross-level interactions with
subgroup status. Given the overlap between being poor and female, we estimate both cross-level
interactions in one model. Low income and female gender are interacted with the logged median
legislator district magnitude and the seat share allocated in upper tiers. The control for the age of
democracy indicates that levels of ideological congruence are generally much higher in established
systems. We now discuss the focal moderating effects of district size and upper-tier compensations
in more detail.
In models with interactions, the constitutive effects of variables refer to the situation where the
other variable in the interaction equals zero, while the interactive effect is a change in the effect of
either variable when the other variable changes by one unit (see Brambor et al., 2006). The coef-
ficient on low income in Model 3 refers to the effect of being poor and male under electoral sys-
tems with logged median legislator district magnitudes of 0 (and no seats allocated in upper tiers),
which equals single-member district systems. It shows that the poor have about a .13 lower ideo-
logical congruence than the rich under these conditions, again statistically significant at high lev-
els. To assess the effects of income and gender at different levels of electoral proportionality, both
the constitutive effects and the interaction effects are used to compute marginal effects. The under-
representation of individuals with low incomes decreases when electoral districts are larger, which
is visible in the positive coefficient of the interaction term between district size and low income.
For women, where there is no under-representation on average to start with, no moderation of the
district size is visible.
The upper panel of Figure 4 visualises the marginal effect of low income and gender, along
with stars indicating significance (see Brambor et al., 2006), as the log of the median legislator
Figure 3. Ideological representation of the poor and women separated per country.
Note: Negative numbers indicate lower levels of representation than the rest of the population.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 89
district magnitude changes from its minimum of 0 (equalling a district magnitude of 1) to its
maximum in the sample at about 5 (148 seats in the median legislator district). In the lower
panel of the figure, the corresponding empirical distribution of this variable is shown. The
negative effect of low income on ideological congruence loses its statistical significance (p <
.05) when the log of the median legislator district magnitude takes values just above 4, which
equals a district size of roughly 60 seats. However, there are only a small number of districts in
the sample larger than this quantity. Regarding gender, the figure confirms that women are on
average not much worse represented than men and that this pattern is not mediated by the elec-
toral rules.
The patterns are less clear for the variable ‘upper tier’ (controlling for district size). The coef-
ficients on female gender and low income refer to the situation with no compensation via the upper
tier of the electoral system. Here, the negative coefficient for low income again indicates that the
poor suffer from a lower ideological congruence. However, this effect remains rather stable across
the range of the variable ‘upper tier’. In other words, the under-representation of the poor is persis-
tent in systems with or without compensation of seats via the upper tier. Figure 5 displays the cor-
responding coefficients. Even for large portions of the seats allocated via upper tiers, the negative
effect of low income remains large, but is a little less pronounced than in systems without
Table 2. Cross-level interactions with the electoral system (dependent variable: individual ideological
congruence).
Model 3
Country level
Log of median legislator district magnitude (LML) .08 (.03)***
Seat share allocated in upper tier .08 (.02)**
Old democracy .20 (.06)***
Individual level
Low income −.13 (.04)***
Female −.01 (.03)
Cross-level interactions
LML × low income .02 (.01)
Upper tier × low income .00 (.00)
LML × female −.00 (.01)
Upper tier × female −.00 (.00)
Constant −.92 (.09)***
Random intercepts
Variance (countries) .019 (.007)
Variance (individuals) .493 (.003)
Random slopes
Variance (income) .004 (.002)
Variance (female) .003 (.001)
N (countries) 20
N (individuals) 44803
Test vs linear reg. (p) .00
Notes: ** p < .05; *** p < .01. Standard errors in parentheses. All calculations were performed in Stata 12, using the
xtmixed command with the reml (restricted maximum likelihood) algorithm. Missing countries are explained by the
missing variable ‘seat share allocated in the upper tier’, which is sometimes impossible to quantify due to varying com-
pensations depending on the actual electoral results (Belgium, Bulgaria, Czech Republic and Romania).
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
90 International Political Science Review 36(1)
compensation, though this effect is not statistically significant. For women, again, no systematic
pattern is visible.
In sum, the findings on the potential moderating effects of the electoral system are mixed but
notable. While larger electoral districts seem to be a remedy against the lower ideological congru-
ence of poor citizens, there is no significant improvement of women’s ideological representation in
larger electoral districts. The seat share allocated in upper tiers yields a similar conclusion, although
less clear-cut. Most of the empirical variation in the proportionality of electoral rules between the
countries in the sample occurs for district magnitude. While larger compensations of disproportion-
ality via seats allocated in upper tiers should have a similar effect to district magnitude, the results
only very tentatively support this expectation. Nevertheless, in the case of poor versus rich citizens,
electoral systems with larger district magnitudes provide a better environment for the representation
of their interests. The considerable gap in ideological representation between low and high incomes
is reduced to statistical insignificance when the district magnitude is large enough.
Figure 4. Graphical presentation of the interaction between low income/gender and the log of median
legislator district magnitude (top panel – marginal effect of low income/gender on the individual ideological
congruence Cij, conditional on the log of median legislator district magnitude; lower panel – distribution of
this moderating factor).
Notes: * Indicates significance at the 95% level. Baseline: high income, male. The figure is based on Model 3 in Table 2.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 91
Several reasons might explain the divergent findings between women and the poor. The pattern
of the representation of women’s interests is considerably more diverse from the outset than that of
poor citizens. Women’s preferences are not as distinct as those of the poor to start with, and in
several nations, their interests are not significantly worse represented by political parties than those
of their male counterparts on average. We still find considerable variation between countries, moti-
vating the analysis of cross-level interactions also for gender. Our argument about the moderating
effect of the electoral rules is largely based on the expectation of more leftist attitudes and a worse
representation of preferences, so it might not be surprising that electoral rules are not good in
explaining the opposite, namely, the over-representation of women’s preferences. Further, the
alignment of women’s interests and the Left–Right dimension is not as straightforward as for poor
citizens. In addition to the economic arguments discussed, women might distinguish themselves
from men more in cultural terms or on distinct women’s issues, such as abortion or maternity pol-
icy (see Schwindt-Bayer and Mishler, 2005). Generally, the Left–Right dimension of political
Figure 5. Graphical presentation of the interaction between low income/gender and share of seats
allocated in upper tiers (top panel – marginal effect of low income/gender on the individual ideological
congruence Cij, conditional on the share of seats allocated in upper tiers; lower panel – distribution of this
moderating factor).
Notes: * Indicates significance at the 95% level. Baseline: high income, male. The figure is based on Model 3 in Table 2.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
92 International Political Science Review 36(1)
competition does not fully capture the complexity of gender-related cleavages. It is more plausible
that redistributive preferences of the poor differ consistently from those of the rich and directly
translate to the Left–Right dimension. Finally, and reflecting critical discussions of groups’ repre-
sentation as unduly ‘essentialist’ (Phillips, 1995), women as a group might have less cohesive
preferences, and, as a result, might be less easily targeted by (small) parties.
Conclusion
In this article, we discuss the ideological representation of historically disadvantaged groups in a
comparative setting. The aim is to add a cross-national comparative perspective to the US literature
on the unequal responsiveness of politicians to different groups of citizens (Bartels, 2008; Gilens,
2005; Griffin and Newman, 2007; Jacobs and Page, 2005; Ura and Ellis, 2008). Our analysis con-
firms that citizens with low incomes are less well represented by political parties as compared with
the more affluent. The results are much less clear-cut in the case of women, as their preferences are
reflected by the party system in some countries better than the preferences of men.
However, our goal is not solely to document the potential under-representation of women and
the poor, but to explore the question of whether systemic factors are responsible for their levels of
representation. We find considerable support for our expectation that the context matters (including
clear variation in the effect of female gender across countries), but are not able to identify the rel-
evant variables equally well for all groups. PR creates party systems in which the preferences of
poor citizens are mirrored more equally. Following Cox (1997), and measuring the proportionality
of the electoral system in terms of the median legislator district magnitude and the share of seats
allocated in upper tiers of the system, multilevel analyses with cross-level interactions reveal that
higher district magnitudes close the gap in ideological representation between poor and rich citi-
zens. As an explanation, parties in more proportional electoral systems will have greater incentives
to take these subgroups of voters into account. In sum, our analysis shows the importance of insti-
tutional design in achieving a more equal political representation of minority groups, and, ulti-
mately, in achieving higher democratic quality (Lijphart, 2012).
Yet, this moderating effect of the electoral system only becomes manifest for subgroups clearly
defined by their low economic status; women’s preferences are not represented in a better way
when the electoral system is more proportional. This result can be seen either as indicating that
men and women are on average equally well represented or as shedding some doubt on the meth-
odology used. Tentatively, the non-finding might be explained by differences between gender and
income groups regarding the relevance of the Left–Right dimension, the cohesiveness of women’s
preferences and, consequently, the incentives for parties to target women directly. This invites
additional research, possibly also in the form of case studies to underpin (or reject) the causal
mechanisms suggested. Along these lines, future studies could explore the role of other mediating
factors (at the macro level) for the relation between subgroup status and representation, including
relative group turnout (Lijphart, 1997), economic inequality (Rodríguez, 2004; Rosset et al., 2013)
or descriptive representation (Griffin and Newman, 2007). These context-level factors could shed
light on the contents of the black box located between electoral rules and the ideological congru-
ence of subgroups or establish independent explanations. Additionally, one of the limitations of
this study is its rather narrow focus on electoral representation. It is clear that women and the poor
also use other channels, including protest movements, to make their specific claims heard (Phillips,
1995; Piven and Cloward, 1977), and that electoral representation by parties can only be related to
government–citizen congruence to a limited extent, let alone explain changes in public policy (for
much more encompassing treatments on the recent rise of economic inequality in the US, see
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 93
Bartels, 2008; Hacker and Pierson, 2010; Schlozman et al., 2012). We believe that our findings still
have some implications for government–citizen congruence. If parties are systematically biased
and represent some groups better than others, it is likely that governments elected from among
these parties will also marginalise some groups.
Notwithstanding these limitations, the contribution of our study is its comparative focus with an
emphasis on non-presidential (mostly European) democracies, providing insights into the moderat-
ing power of political institutions able to absorb some of the under-representation of (certain) disad-
vantaged societal subgroups. If segments of society perpetually have a lower political voice, this
carries the risk of policies deepening already-existing inequalities, which can have important conse-
quences for levels of trust in political institutions and for their perceived legitimacy. As reflected in
our findings, proportional electoral rules appear to make it more attractive for political parties to
also attract groups with fewer resources and lower levels of political participation, as office and
policy rewards are more easily achieved. From a normative democratic perspective, it is worrisome
that sub-constituencies are systematically under-represented. Our (partial) finding of the moderating
power of electoral rules introduces some leeway for institutional design as a potential remedy.
Acknowledgements
All authors contributed equally and appear in alphabetical order. Earlier versions of this article were presented
at the Council for European Studies’ Seventeenth International Conference, Montreal, Canada, 15–17 April
2010, and at the Annual Meeting of the Midwest Political Science Association, Chicago, 22–25 April 2010.
We would like to thank the participants for their useful comments, in particular, André Blais, Delia Dumitrescu,
Peter Esaiasson, Adam Luedtke, Thomas Meyer, Irene Menendez, Shane Singh and Catherine de Vries. We
are also grateful for the valuable advice from the editors and reviewers of IPSR.
Funding
This research was supported by the European Science Foundation (grant number 07-HumVIB-FP-007_
REPCONG).
Notes
1. Gender scholars have long engaged with the question of whether certain institutional settings favour
the representation of women in parliament (see Paxton et al., 2007). This study, however, focuses on
ideological representation and not descriptive representation, which is still assumed to be relevant, but is
black-boxed here.
2. In this study, we focus on ideological representation as a special case of ‘substantive’ or ‘policy’ repre-
sentation (for the concept of representation, see Pitkin, 1967; Schwindt-Bayer and Mishler, 2005).
3. We treat women and the poor as (overlapping) societal groups based on the expectation that they have
distinct preferences. We note that these assumptions could be challenged as false ‘essentialism’ (Phillips,
1995), as multiple identifications exist (which we allow for in our analysis, at least for income and
gender). Group boundaries, especially in the case of the poor, can be blurred and preferences are not
necessarily both distinct from each other and homogeneous within groups (which we control for in our
individual-level analysis to some extent). Still, we regard the identification of groups as an analytically
crucial simplification for our model.
4. The recent rise of right-wing populist parties and evidence that these parties draw their support in par-
ticular from low-income and lower-educated individuals (see, e.g., Ford and Goodwin, 2010) might
make this argument less empirically convincing in the future, though.
5. We note that the so-called ‘modern gender gap’, with women leaning more to the Left than men, is not
visible in all Western countries to the same extent. For example, it is well known that the voting pattern
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
94 International Political Science Review 36(1)
in Finland is not showing any sign of a modern gender gap; in contrast, Finnish women are still voting
more often for conservative parties (Hart et al., 2009; Inglehart and Norris, 2000).
6. Please note that there is also a critical debate over whether such a link can be assumed, for example, see
the debate in Politics & Gender (2006), 2(4): 491–530.
7. As this variable is highly dependent on the electoral design, we do not explicitly include the number of
parties in the analysis, but assume a connection between electoral rules, the number of parties and indi-
vidual policy congruence.
8. Our sample comprises elections from a period of 10 years. Controlling for possible temporal effects does
not alter the results presented here and has very little substantial influence on the dependent variable.
9. The exact wording of this question is: ‘In politics people sometimes talk of left and right. Where would
you place yourself on a scale from 0 to 10 where 0 means the left and 10 means the right?’ (A3031
and B3045). Other policy positions of voters are not available in the CSES. As argued earlier, there is
a theoretical connection linking the preferences of both women and the poor to the general (economic)
Left–Right scale.
10. More precisely, the experts were asked about their judgement of the party position on the Left–Right
scale (A5004, B5018). Alternatively, the party positions can also be estimated using the Benoit–Laver
expert survey data (Benoit and Laver, 2006). These two measures of party positions correlate very highly
(Pearson’s correlation coefficient Module 1 (1996–2001): 0.89; Module 2 (2001–2006): 0.92). The CSES
expert data that are part of the contextual survey accompanying the individual CSES surveys carry the
advantage that they are measured on the same scale as the citizen preferences and time variant. To avoid
additional (problematic) rescaling, we decided to use the CSES expert data. The results are robust against
the use of Benoit–Laver expert party positions.
11. We note that other options to calculate the positions of parties exist. First, we can rely on party manifesto
data, as collected by the Comparative Manifesto Project (CMP) (Budge et al., 2001; Klingemann et al.,
2006). The use of this type of data carries the large disadvantage that by focusing on the percentage of
Left–Right statements, we capture more the salience a party puts on Left–Right issues than a party’s
substantial Left–Right position. Another option is to estimate the party placement as the mean voters’
evaluation of a party, available in the CSES survey. Our results are also robust against this variation of
the operationalisation. However, as we deem likely that the individual idiosyncratic factors do not cancel
out completely – as, for example, the position of a voter’s own party is often evaluated differently from
the rest of the parties – we prefer to operationalise party positions with expert data. A last option to esti-
mate party positions is the use of MP surveys (see, e.g., Esaiasson and Holmberg, 1996). Here, we again
run into the problem of data availability and cross-national comparability.
12. One could argue that the downside of this measure is that we lose the direct connection between indi-
viduals and parties as the party with the closest distance is not always the party voted for. However,
measures based on the distance between the party chosen and individual positions are even more prone
to flaws, such as lower probabilities among the poor of choosing a party based on distance.
13. The models are robust against a specification of the lowest income quintile as a ‘low-income group’.
14. Note that this differs from the median district magnitude, as Cox (1997) weights the district magnitudes
by their size (i.e. the number of legislators elected, hence the name median ‘legislator’ district magni-
tude), and then picks the median value.
15. Specifying elections instead of countries as the second level does not affect our results.
References
Achen, Cristopher H. (1978) Measuring representation. Americal Journal of Political Science 22(3): 475–510.
Adams, James and Lawrence Ezrow (2009) Who do European parties represent? How Western European par-
ties represent the policy preferences of opinion leaders. The Journal of Politics 71(1): 206–223.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 95
Bartels, Larry Martin (2008) Unequal Democracy: The Political Economy of the New Gilded Age. New York,
NY: Russell Sage Foundation.
Benoit, Kenneth and Michael Laver (2006) Party Policy in Modern Democracies. London: Routledge.
Bergh, Johannes (2007) Explaining the gender gap – a cross-national analysis of gender differences in voting.
Journal of Elections, Public Opinion and Parties 17(3): 235–261.
Blais, André (1991) The debate over electoral systems. International Political Science Review 12(3): 239–
260.
Blais, André and Marc André Bodet (2006) Does proportional representation foster closer congruence
between citizens and policy makers? Comparative Political Studies 39(10): 1243–1262.
Brambor, Thomas, William Roberts Clark and Matt Golder (2006) Understanding interaction models:
Improving empirical analyses. Political Analysis 14(1): 63–82.
Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens et al. (2001) Mapping Policy Preferences. Estimates
for Parties, Electors, and Governments 1945–1998. New York, NY: Oxford University Press.
Calvo, Ernesto and Timothy Hellwig (2011) Centripetal and centrifugal incentives under different electoral
systems. American Journal of Political Science 55(1): 27–41.
Cox, Gary W. (1990) Centripetal and centrifugal incentives in electoral systems. American Journal of
Political Science 34(4): 903–935.
Cox, Gary W. (1997) Making Votes Count: Strategic Coordination in the World’s Electoral Systems.
Cambridge: Cambridge University Press.
Cusack, Thomas R., Torben Iversen and Philipp Rehm (2008) Economic shocks, inequality and popular sup-
port for redistribution. In: Pablo Beramendi and Christopher J. Anderson (eds) Democracy, Inequality and
Representation. New York, NY: Russell Sage Foundation, 203–231.
Dalton, Russell J. (1985) Political parties and political representation: Party supporters and party elites in nine
nations. Comparative Political Studies 18(3): 267–299.
Dovi, Suzanne (2007) Theorizing women’s representation in the United States. Politics & Gender 3(3): 297–
319.
Downs, Anthony (1957) An Economic Theory of Democracy. New York, NY: Harper and Row.
Duverger, Maurice (1963) Political Parties: Their Organization and Activity in the Modern State. New York,
NY: Wiley.
Enns, Peter K. and Christopher Wlezien (2011) Who Gets Represented? New York, NY: Russell Sage
Foundation.
Esaiasson, Peter and Soren Holmberg (1996) Representation from Above: Members of Parliament and
Representative Democracy in Sweden. Aldershot: Dartmouth Publishers.
Evans, Elizabeth (2008) Supply and demand: Women candidates and the Liberal Democrats. British Journal
of Politics and International Relations 10(4): 590–606.
Ezrow, Lawrence (2010) Linking Citizens and Parties: How Electoral Systems Matter for Political
Representation. Oxford: Oxford University Press.
Ezrow, Lawrence, Catherine de Vries, Marco Steenbergen et al. (2011) Mean voter representation versus
partisan constituency representation: Do parties respond to the mean voter position or to their supporters?
Party Politics 17(3): 275–301.
Ford, Robert and Matthew J. Goodwin (2010) Angry white men: Individual and contextual predictors of sup-
port for the British National Party. Political Studies 58(1): 1–25.
Gidengil, Elisabeth, André Blais, Neil Nevitte et al. (2003) Women to the Left? Gender differences in politi-
cal beliefs and policy preferences. In: Manon Tremblay and Linda Trimble (eds) Women and Electoral
Politics in Canada. Don Mills, Ontario: Oxford University Press, 140–159.
Giger, Nathalie (2009) Towards a modern gender gap in Europe? A comparative analysis of voting behavior
in 12 countries. Social Science Journal 46(3): 474–492.
Giger, Nathalie, Jan Rosset and Julian Bernauer (2012) The poor representation of the poor in comparative
perspective. Journal of Representation 48(1): 47–61.
Gilens, Martin (2005) Inequality and democratic responsiveness. Public Opinion Quarterly 69(5): 778–796.
Gilens, Martin (2009) Preference Gaps and Inequality in Representation. Political Science & Politics 42:
335–341.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
96 International Political Science Review 36(1)
Gilens, Martin (2012) Affluence & Influence. Economic Inequality and Political Power in America. Princeton:
Princeton University Press
Golder, Matt and Jacek Stramski (2010) Ideological congruence and electoral institutions: Conceptualization
and measurement. American Journal of Political Science 54(1): 90–106.
Griffin, John D. and Brian Newman (2005) Are voters better represented? The Journal of Politics 67(4):
1206–1227.
Griffin, John D. and Brian Newman (2007) The unequal representation of Latinos and whites. The Journal of
Politics 69(4): 1032–1046.
Hacker, Jacob S. and Paul Pierson (2010) Winner-Take-All Politics: How Washington Made the Rich Richer
– and Turned Its Back on the Middle Class. New York, NY: Simon & Schuster.
Hart, Linda, Anne Kovalainen and Anne Maria Holli (2009) Gender and power in politics and business in
Finland. In: Niskanen Kirsti and Anita Nyberg (eds) Kön och makt i Norden: Del I Landsrapporter.
Copenhagen: Nordiska ministerrådet, 65–130.
Headey, Bruce and Denis Muller (1996) Policy agendas of the poor, the public and elites: A test of Bachrach
and Baratz. Australian Journal of Political Science 31(3): 347–368.
Huber, John D. and G. Bingham Powell (1994) Congruence between citizens and policymakers in two visions
of liberal democracy. World Politics 46(3): 291–226.
Inglehart, Ronald and Pippa Norris (2000) The developmental theory of the gender gap: Women’s and men’s
voting behavior in global perspective. International Political Science Review 21(4): 441–463.
Iversen, Torben and Frances Rosenbluth (2006) The political economy of gender: Explaining cross-national
variation in household bargaining, divorce, and the gender voting gap. American Journal of Political
Science 50(1): 1–19.
Iversen, Torben and David Soskice (2006) Electoral institutions and the politics of coalitions: Why some
democracies redistribute more than others. American Political Science Review 100(2): 165–181.
Jacobs, Lawrence R. and Benjamin I. Page (2005) Who influences U.S. foreign policy? American Political
Science Review 99(1): 107–123.
Kenworthy, Lane and Melissa Malami (1999) Gender inequality in political representation: A worldwide
comparative analysis. Social Forces 78(1): 235–268.
Klingemann, Hans-Dieter, Andrea Volkens, Judith Bara et al. (2006) Mapping Policy Preference II: Estimates
for Parties, Electors and Governments in Eastern Europe, the European Union and the OECD, 1990–
2003. Oxford: Oxford University Press.
Kostadinova, Tatiana (2007) Ethnic and women’s representation under mixed election systems. Electoral
Studies 26(2): 418–431.
Lijphart, Arend (1997) Unequal participation: Democracy’s unresolved dilemma. American Political Science
Review 91(1): 1–14.
Lijphart, Arend (2012) Patterns of Democracy. Government Form and Performance in Thirty-Six Countries
(2nd edn). New Haven, CT: Yale University Press.
Lovenduski, Joni and Pippa Norris (2003) Westminster women: The politics of presence. Political Studies
51(1): 84–102.
McDonald, Michael D., Silvia M. Mendes and Ian Budge (2004) What are elections for? Conferring the
median mandate. British Journal of Political Science 34(1): 1–26.
Mansbridge, Jane (1999) Should blacks represent blacks and women represent women? A contingent “yes”.
The Journal of Politics 61(3): 628–657.
Manza, Jeff and Clem Brooks (1998) The gender gap in U.S. presidential elections: When? Why? Implications?
American Journal of Sociology 103(5): 1235–1266.
Matland, Richard E. (1993) Institutional variables affecting female representation in national legislatures: The
case of Norway. The Journal of Politics 55(3): 737–755.
Matland, Richard E. (1998) Women’s representation in national legislatures: Developed and developing
countries. Legislative Studies Quarterly 23(1): 109–125.
Matland, Richard E. and Donley T. Studlar (1996) The contagion of women candidates in single-member
district and proportional representation electoral systems: Canada and Norway. The Journal of Politics
58(3): 707–733.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
Bernauer et al. 97
Meltzer, Allan H. and Scott F. Richard (1981) A rational theory of the size of government. Journal of Political
Economy 89(5): 914–927.
Moser, Robert G. and Ethan Schreiner (2012) Electoral Systems and Political Context: How the Effects of
Rules Vary Across New and Established Democracies. Cambridge: Cambridge University Press.
Norris, Pippa (2006) The impact of electoral reform on women’s representation. Acta Politica 41(2): 197–213.
Paxton, Pamela (1997) Women in national legislatures: A cross-national analysis. Social Science Research
26(4): 442–464.
Paxton, Pamela, Sheri Kunovich and Melanie M. Hughes (2007) Gender in politics. Annual Review of
Sociology 33: 263–284.
Phillips, Anne (1995) The Politics of Presence. Oxford: Oxford University Press.
Pitkin, Hanna F. (1967) The Concept of Representation. Berkeley, CA: University of California Press.
Piven, Frances Fox and Richard Cloward (1977) Poor People’s Movements: Why They Succeed, How They
Fail. New York, NY: Pantheon.
Powell, G. Bingham (2000) Elections as Instruments of Democracy: Majoritarian and Proportional Visions.
New Haven, CT: Yale University Press.
Powell, G. Bingham (2004) The chain of responsiveness. The Journal of Democracy 15(4): 91–105.
Rodríguez, Francisco (2004) Inequality, redistribution, and rent-seeking. Economics & Politics 16(3):
287–320.
Rosset, Jan, Nathalie Giger and Julian Bernauer (2013) More Money, Fewer Problems? Cross-Level Effects
of Economic Deprivation on Political Representation. West European Politics 36(4): 817–835.
Ruedin, Didier (2013) Why Aren’t They There? The Political Representation of Women, Ethnic Groups and
Issue Positions in Legislatures. Colchester: ECPR Press.
Sawer, Marian (2012) What makes the substantive representation of women possible in a Westminster parlia-
ment? The story of RU486 in Australia. International Political Science Review 33(3): 320–335.
Schlozman, Kay Lehman, Sidney Verba and Henry E. Brady (2012) The Unheavenly Chorus: Unequal
Political Voice and the Broken Promise of American Democracy. Princeton, NJ: Princeton University
Press.
Schwindt-Bayer, Leslie A. and William Mishler (2005) An integrated model of women’s representation.
Journal of Politics 67(2): 407–428.
Siaroff, Alan (2000) Women’s representation in legislatures and cabinets in industrial democracies.
International Political Science Review 21(2): 197–215.
Soroka, Stuart N. and Christopher Wlezien (2008) On the limits to inequality in representation. PS: Political
Science & Politics 41(2): 319–327.
Steenbergen, Marco R. and Bradford S. Jones (2002) Modeling multilevel data structures. American Journal
of Political Science 46(1): 218–237.
Strøm, Kaare and Wolfgang C. Mueller (1999) Political parties and hard choices. In: Kaare Strøm and
Wolfgang C. Mueller (eds) Policy, Office or Votes? How Political Parties in Western Europe Make Hard
Decisions. Cambridge: Cambridge University Press, 1–35.
Studlar, Donley T. and Ian McAllister (1996) Constituency activity and representational roles among
Australian legislators. Journal of Politics 58: 69–90.
Studlar, Donley T. and Ian McAllister (2002) Does a critical mass exist? A comparative analysis of women’s
legislative representation since 1950. European Journal of Political Research 41(2): 233–253.
Studlar, Donley T., Ian McAllister and Bernadette C. Hayes (1998) Explaining the gender gap in voting: A
cross-national analysis. Social Science Quarterly 79(4): 779–798.
Taagepera, Rein (2002) Nationwide threshold of representation. Electoral Studies 21(3): 383–401.
Thomassen, Jacques (1994) Empirical research into political representation: Failing democracy or failing
models. In: M. Kent Jennings and Thomas E. Mann (eds) Elections at Home and Abroad: Essays in
Honor of Warren E. Miller. Ann Arbor, MI: University of Michigan Press, 237–264.
Ura, Joseph Daniel and Christopher R. Ellis (2008) Income, preferences, and the dynamics of policy respon-
siveness. PS: Political Science & Politics 41(1): 785–794.
Wängnerud, Lena (2000) Testing the politics of presence: Women’s representation in the Swedish Riksdag.
Scandinavian Political Studies 23: 67–91.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
98 International Political Science Review 36(1)
Wängnerud, Lena and Anders Sundell (2012) Do politics matter? Women in Swedish local elected assemblies
1970–2010 and gender equality in outcomes. European Political Science Review 4(1): 97–120.
Author biographies
Julian Bernauer is a postdoctoral researcher at the Institute of Political Science, University of Berne,
Switzerland. His research interests include political representation and the study of empirical patterns of
democracy. Recent work has appeared in Comparative Political Studies, the European Journal of Political
Research and Electoral Studies.
Nathalie Giger is a postdoctoral researcher and lecturer at the Mannheim Centre for European Social
Research (MZES) and the Department of Political Science at the University of Mannheim, Germany. Her
research interests lie in the linkage between citizens and political elites, in particular, in political representa-
tion and the electoral consequences of public policy. Her work has appeared in the European Journal of
Political Research, West European Politics and European Sociological Review, among others.
Jan Rosset is a PhD student at the University of Lausanne, Switzerland, and a research associate at the Swiss
Centre of Expertise in Social Sciences (FORS). His research centres on the relationship between economic
and political inequality and on the political representation of income groups. His work has been published or
is forthcoming in the Journal of Legislative Studies, Representation and West European Politics.
at Universitaet Mannheim on January 12, 2015ips.sagepub.comDownloaded from
