Dataset of ideological polarization in Western Europe

Vincenzo Emanuele and Bruno Marino (Luiss, Rome)

Codebook (July 13, 2023)

Description
This dataset provides data on ideological polarization in Western Europe. It is based on parties’ left-right
placement provided by several expert surveys. Then, it uses Dalton’s polarization index (2008) to calculate
the polarization score in terms of votes and seats for each election. The dataset covers 20 Western
European countries since 1945, for a total of 398 parliamentary elections and legislatures (Lower House).
The dataset will be regularly updated to include the polarization scores of new elections and legislatures.

Content
Country: the country where the parliamentary election is held (in alphabetical order)
Election Year: the year in which the election is held
Election Date: exact date of the election
Polarization_votes: the polarization score of the election calculated from parties’ left-right placement
according to expert survey data and using Dalton’s polarization index (2008). The score ranges from 0 to
10, where higher values mean higher polarization.
Polarization_seats: the polarization score of the legislature calculated from parties’ left-right placement
according to expert survey data and using Dalton’s polarization index (2008). The score ranges from 0 to
10, where higher values mean higher polarization.

General rules and data sources

This dataset is based on parties’ left-right placement provided by a series of comparative and widely-
recognized expert surveys that allow for extended temporal coverage and reliable cross-country and
cross-time comparisons. Specifically, we use Castles and Mair (1983), Huber and Inglehart (1995), and all
the available Chapel Hill Expert Survey (CHES) waves (1999, 2002, 2006, 2010, 2014, 2019) (Jolly et al.
2022). The general rule is that, for each election and legislature (votes and seats), we assign left-right
values to parties according to the closer data source to the time of the parliamentary election. As a result,
the 1945-1987 period is covered by Castles and Mair (1983), the 1988-1995 period by Huber and Inglehart
(1995), the 1996-2000 by CHES 1999, the 2001-2003 period by CHES 2002, the 2004-2007 period by
1
CHES 2006, the 2008-2011 period by CHES 2010, the 2012-2016 period by CHES 2014, and the 2017-
2021 period by CHES 2019. In the case of Huber and Inglehart’s expert survey (1995), we have
considered the year of survey administration (1993) rather than the year of publication (1995). Moreover,
given that this latter is the only expert survey considered using a 1-10 range for parties’ left-right
placement rather than the 0-10 range used in all the other expert surveys, Huber and Inglehart’s data have
been rescaled to a 0-10 range to allow for comparability with the other surveys. Furthermore, in case of
an election year falling exactly in between two expert surveys (e.g., 2012 falling between CHES 2010 and
CHES 2014), we have opted for the more recent source. Following this rationale, the dataset does not
cover the 2022 and 2023 years yet, as the latter will be covered by the next CHES expert survey expected
for 2024. Data updates for 2022 and 2023 will be released as soon as CHES 2024 is available.
Following this general rule, in case of missing data for a party in a given expert survey, we attribute to
that party the left-right placement provided by the closer available source. For instance, if a party is
missing in CHES 2002, for the 2003 election, we take the party’s left-right placement attributed by CHES
2006, if available, or the following closer source. This rule follows a general principle of parties’
ideological stickiness, which argues that parties hardly change positions over time (Hooghe and Marks
2018).
Finally, in the case of missing data for a party in all surveys, we follow a principle of organizational
continuity (Bartolini and Mair 1990; Chiaramonte and Emanuele 2017). This means that if the party has
emerged by splitting from another party, we attribute to this party the left-right placement of its
predecessor (if available) in the election in which the splinter party appears for the first time. Moreover,
if the party is not the result of a split from a party for which a left-right placement is available but merges
into a party for which we have a left-right placement, the merging party assumes the score of its successor.
Furthermore, in the case of joint lists, if the left-right placement of the joint list is missing, we attribute
it by weighting the left-right placement of the parties forming the joint list by the respective vote share
in that election (if the constituent parties have also run individually in the election at time t) or in the
previous election (if the constituent parties have ran only in the joint list in the election at time t).
The main source for electoral data is Nohlen and Stöver (2010), for elections until 2009. For the elections
held since 2010, we have relied on official data provided by the pertinent electoral authority for each
country. Details about sources and other methodological choices are specified below in the notes on the
individual countries.

Notes on individual countries

In the following paragraphs, we report specific issues related to data sources, missing information for
parties, and other national specificities. In particular, for each country, we list the elections for which
there is a lack of information on the left-right placement of parties totaling more than 10% of votes
2
and/or seats. Moreover, we also list the parties receiving at least 5% of the votes and/or seats in at least
one election for which the left-right placement is not available. For party names, we always report the
English name, with the acronym in the original language in brackets.

Belgium
As Nohlen and Stöver (2010) do not disaggregate ‘other parties’ and do not report the separate results
for Walloon and Flemish Christian-democratic, socialist, and liberal parties until 2003, we have followed
the official electoral authority as the electoral data source. Furthermore, the left-right placement of the
Christian Social Party (PSC-CVP), the Belgian Socialist Party (PSB-BSP), and the Liberal Party (PRL-
PVV) before their split is the average of the placements attributed to the respective Walloon and Flemish
splinter parties in Castles and Mair (1983). Moreover, the placement of the Cartel of Liberals and
Socialists (KLS) (1946-1958) is the weighted average of the placements of the Belgian Socialist Party and
the Liberal Party. Finally, the placement of the Catholic Lists (LC) (1954 and 1961) is the placement of
the Christian Social Party.

Cyprus
In 1976, the left-right placement of the alliance formed by the Democratic Party (DIKO), the Progressive
Party of the Working People (AKEL), and the Movement for Social Democracy (EDEK) is the average
of the placements of these three parties in Castles and Mair (1983), given no earlier electoral data are
available.
The 2021 Cypriot election reports a score for Polarization_votes that shall be treated with caution given
that the left-right placement available for parties represents 87.7% of the vote share.

Denmark
No left-right placement has been retrieved for the Justice Party (RF).

France
As reported by Nohlen and Stöver (2010, 671), electoral data sources are inconsistent. For this reason,
by comparing different sources (Nohlen and Stöver 2010, Siaroff 2019, Döring and Manow 2020, and
the official electoral authority), we have followed, election after election, the electoral results that were
more consistent across the various sources. As a result, the electoral data source for the Fourth Republic’s
elections is Siaroff (2019) and Döring and Manow (2020), for the Fifth Republic’s elections is Nohlen
and Stöver (2010) and Döring and Manow (2020), while, for the more recent elections, the official
electoral authority. The Radical Party (PR) and the related denominations across elections receive the
placement of the Union for French Democracy (UDF), in which the biggest part of the PR merges in
3
1978. Then, in 1967, the Center Democrats (CD), formed by parts of the Popular Republican Movement
(MRP) and the National Center of Independents and Peasants (CNIP), receives the placement of the
CNIP, given that – see below – no left-right placement is available for the MRP. Moreover, in some
elections, different sources report electoral results for ‘Extreme Left’ and ‘Extreme Right’ labels. We have
attributed to these lists the left-right placement of, respectively, the Workers Struggle (LO) and the
National Front (FN), as these are the first two recognizable parties emerging from these labels according
to the official electoral authority. Furthermore, to reconstruct the components of the UDF in 1978 and
1981, we have used Nohlen and Stöver (2010). Moreover, in different elections of the Fourth and the
Fifth Republic, Nohlen and Stöver (2010) and Siaroff (2019) report the labels ‘Conservatives’ or
‘Conservatives/Moderates’. These labels include various conservative lists, among which the most
important is the CNIP. We have attributed to these labels the left-right placement of the CNIP.
Furthermore, in 1978, 1981, and 1988, the list ‘Presidential Majority’ receives the placement of the
President’s party. In 1988, the votes and seats of the list Miscellaneous Right, following Nohlen and
Stöver (2010, 705), are attributed to the ‘Conservatives’. In 1993, the list Miscellaneous Greens receives
the score of The Greens (LV).
No left-right placement has been retrieved for the MRP and the ‘Poujade List’, namely, the Union for
the Defense of Traders and Artisans (UDCA).
The following elections report a score for Polarization_votes and Polarization_seats that shall be treated with
caution given that the overall vote and seat share of parties with an available left-right placement is below
90%: the 1945 (74.2% of votes and 72.6% of seats) and 1946 Constitutional Assembly elections (71.6%
of votes and 69.3% of seats), the 1946 (73.3% of votes and 71% of seats), 1951 (86.9% of votes and
84.9% of seats), 1956 (76.8% of votes and 77% of seats), and 1958 (86.8% of votes and 87.7% of seats)
elections.

Germany
The Bayern Party (BP) receives the left-right placement of the Christian Social Union (CSU), as both
parties originate from the same predecessor, the Bavarian People’s Party (BVP). Moreover, in the 1949
and 1953 elections, we have attributed to the reconstituted Center Party (DZ) the placement of the
Christian Democratic Union of Germany (CDU), as the latter, as well as its Bavarian ally, the CSU, are
the successors of the pre-World-War-II Center Party (Zentrum).
No left-right placement has been retrieved for the All-German Bloc/League of Expellees and Deprived
of Rights (GB/BHE).
The 1949, 1953, and 1957 elections report a score for Polarization_votes that shall be treated with caution
given that the left-right placement available for parties represents, respectively, 85.1%, 88.2%, and 89.7%
of the vote share.
4
Greece
In 1974, the coalition United Left (EA) takes the left-right placement of the Communist Party of Greece
(KKE), which represents its main component. The same applies to Synaspismos (SYN) for the June
1989, September 1989, and 1990 elections.
No left-right placement has been retrieved for the Union of Democratic Center (EDIK).
The following elections report a score for Polarization_votes and/or Polarization_seats that shall be treated
with caution given that the overall vote and/or seat share of parties with an available left-right placement
is below 90%: 1974 (77.5% of the vote share and 80% of the seat share) and 1977 (87.1% of the vote
share).

Iceland
In 2013, Dawn (Organization of Justice, Fairness and Democracy) is created as a merge of three political
parties, The Movement, Citizens’ Movement, and the Liberal Party (Ff). As the latter is the only one with
an available left-right placement, we have attributed its placement to Dawn.
No placement has been retrieved for the National Preservation Party (Thva), the Union of Liberals and
Leftists (Sfvm), the Civic Movement – The Movement (B-H), and Bright Future (BF).
The 1971, 2013, and 2016 elections report a score for Polarization_votes that shall be treated with caution
given that the left-right placement available for parties represents, respectively, 89.1%, 83.1%, and 88%
of the vote share.

Ireland
For the 1989, 1992, and 1997 elections, the placement of the Sinn Fein (SF) is that reported by Huber
and Inglehart (1995) for Northern Ireland.
No placement has been retrieved for the Republican Party (CnP) and the Party of the Land (CnT).
The following elections report a score for Polarization_votes and/or Polarization_seats that shall be treated
with caution given that the overall vote and/or seat share of parties with an available left-right placement
is below 90%: 1948 (72.9% of votes and 80.3% of seats), 1951 (83.4% of votes and 85% of seats), 1954
(87.6% of votes), 1957 (89.4% of votes), 1961 (88.5% of votes), 2002 (89.9% of votes), 2011 (87% of
votes), 2016 (83.9% of votes and 88% of seats), and 2020 (87% of votes and 88.1% of seats).

Italy
The electoral data source is the official electoral authority. In 1992, the list Federalism (Federalismo) is a
joint list of various minor (mainly regionalist) parties, among which the Sardinian Action Party (PSDaZ)
is the only one with an available left-right placement. Therefore, we have attributed its placement to the
5
list Federalism. In 2008, the reconstituted Italian Socialist Party (PSI) takes the placement of the New
Italian Socialist Party (NPSI), one of its main founding members and the only one for which a placement
is available. In 2018, the joint list Power to the People (PAP) receives the placement of Communist
Refoundation (PRC), which represents its main component.
No placement has been retrieved for the Front of the Ordinary Man (UQ).

Malta
No placement has been retrieved for the Gozo Party (GP), the Democratic Action Party (DAP), the
Constitutionalist Party (PK), and the Jones Party (JP).
The following elections report a score for Polarization_votes and Polarization_seats that shall be treated with
caution given that the overall vote and seat share of parties with an available left-right placement is below
90%: 1947 (77.9% of votes and 77.5% of seats) and 1950 (81.4% of votes and 85% of seats).

Netherlands
The placement of the Christian Democratic Appeal (CDA) in Castles and Mair (1983) is also attributed
to its founding parties: the Catholic People’s Party (KVP), the Anti-Revolutionary Party (ARP), and the
Christian Historical Union (CHU).

Norway
The Christian Unity Party (KSP) is the result of the merge between the New Future Coalition Party (SNF)
and the Christian Conservative Party (KKP) in 1998. As the only available left-right placement is that of
the SNF, we have attributed it to the KSP.

Portugal
For the sake of comparability, electoral results have been recalculated whenever including blank and
invalid votes in the calculus of valid votes. Since 1979, the United People Alliance (APU) and the Unified
Democratic Coalition (CDU) have been given the left-right placement of their main component, the
Portuguese Communist Party (PCP).
No placement has been retrieved for the Democratic Renewal Party (PRD).
The 1985 Portuguese election reports a score for Polarization_votes and Polarization_seats that shall be
treated with caution given that the left-right placement is available for parties representing 80.3% of votes
and 82% of seats.

Spain

6
Similarly to Portugal, electoral results have been recalculated whenever including blank votes in the
calculus of valid votes. For the 1982 election, electoral sources report inconsistent data for parties. We
have decided to use the official electoral authority as the data source. Moreover, given that Nohlen and
Stöver (2010) do not disaggregate the ‘other parties’ category, we have followed the official electoral
authority for the vote shares of parties in the ‘other parties’ category.

Switzerland
Also in this case, given that Nohlen and Stöver (2010) do not disaggregate the ‘other parties’ category,
we have followed the official electoral authority for the vote shares of parties in the ‘other parties’
category.
No placement has been retrieved for the Swiss Party of Labour (PdA) and the Independents Alliance
(LdU/AdI).
The following elections report a score for Polarization_votes and/or Polarization_seats that shall be treated
with caution given that the overall vote and/or seat share of parties with an available left-right placement
is below 90%: 1947 (89.5% of votes), 1951 (88.7% of votes), 1955 (88.9% of votes), 1959 (88.7% of
votes), 1963 (89.1% of votes), 1967 (83.7% of votes and 87% of seats), 1971 (81.1% of votes and 85.5%
of seats), 1975 (84.5% of votes and 89.5% of seats), 1979 (89.3% of votes), 1983 (88.7% of votes), and
1987 (89.8% of votes).

United Kingdom
The votes and the seat won by the Speaker are attributed to the party to which the Speaker belongs. For
the 1945-1987 general elections and the 2001-2019 general elections, the left-right placement of the Sinn
Fein (SF) is that of the party contesting the elections in Ireland.

How to cite this dataset?

Emanuele, V. and Marino, B. (2023), ‘Dataset of ideological polarization in Western Europe’, Rome: Italian
Center for Electoral Studies, https://doi.org/10.7802/2592.

References
Bartolini, S. and Mair, P. (1990) Identity, Competition, and Electoral Availability. Cambridge: Cambridge
University Press.
Castles, F. G. and Mair, P. (1984) Left–Right Political Scales: Some ‘Expert’ Judgments. European Journal
of Political Research, 12 (1), pp. 73-88.

7
Chiaramonte, A. and Emanuele, V. (2017) Party system volatility, regeneration and de-institutionalization
in Western Europe (1945-2015). Party Politics, 23 (4), pp. 376-388.
Dalton, R. J. (2008) The quantity and the quality of party systems: Party system polarization, its
measurement, and its consequences. Comparative Political Studies, 41 (7), pp. 899-920.
Döring, H. and Manow, P. (2020) Parliament and Government Composition Database. https://parlgov.org
Hooghe, L. and Marks, G. (2018) Cleavage theory meets Europe’s crises: Lipset, Rokkan, and the
transnational cleavage. Journal of European Public Policy, 25 (1), pp. 109-135.
Huber, J. and Inglehart, R. (1995) Expert Interpretations of Party Space and Party Locations in 42
Societies. Party Politics, 1 (1), pp. 73-111.
Jolly, S., Bakker, R., Hooghe, L., Marks, G., Polk, J., Rovny, J., Steenbergen, M. and Vachudova, M. A.
(2022) Chapel Hill Expert Survey trend file, 1999-2019. Electoral Studies, 75,
https://doi.org/10.1016/j.electstud.2021.102420
Nohlen, D. and Stöver, P. (eds) (2010), Elections in Europe: A Data Handbook. Baden-Baden: Nomos.