Electoral volatility in Europe: Assessments and potential
explanations for estimate differences
Svante Ersson,
Department of Political Science, Umeå University, SE-901 87 Umeå, Sweden
e-mail: [email protected]
Paper to be presented to the 2012 Elections, Public Opinion and Parties (EPOP) Conference
Oxford University, September 7 - 9, 2012
Abstract:
The problem addressed in this paper is how to measure and calculate accurate aggregate
electoral volatility scores. This paper relies on two kinds of data sets to make comparisons of
different estimates possible; a first one is a collection of electoral data for 31 European
countries covering elections from after WWII up till the June 2012 elections in Greece and
France, and a second one collects estimates of aggregate electoral volatility as presented in the
research literature. For the electoral data the ambition has been to collect data for as many
parties participating at elections as possible, and based on this data set scores for aggregate
electoral volatility has been estimated for 395 elections. Besides mapping the variation in
volatility in space and over time comparisons of different estimates of volatility has been
conducted in order to gain an understanding of why estimates tend to differ for certain
countries and elections. The findings reported in this paper suggest that disaggregating the
group of “other parties” only have a limited impact on volatility estimates; the treatment of
“independents” matters more for the estimates for party systems where they are numerous; the
Pedersen index is still useful as a measure of aggregate electoral volatility, although other
measures like the lsq-index or the 1-cos-index may contribute with complementary
information; distinguishing between type A and type B volatility enriches the analysis of
electoral volatility, but it is not a solution for achieving more reliable estimates of volatility;
what is obvious from this paper is that the treatment of party splits and party mergers and the
defining of genuinely new parties are crucial for achieving more reliable volatility estimates.
To arrive at better volatility estimates it is necessary to reach agreement about the analytical
framework meaning how to define what is meant with genuinely new parties and how to treat
party splits and party mergers; to do so there is also a need to consult country expertise and
examine a number of critical cases where an analytical framework may be tested and further
refined.
“The occurrence of such events [splits and mergers] makes it very difficult to compare
electoral volatility over time in at least some of the European nations, most notably France
and Italy.” (Pedersen 1979: 5)
“Measuring volatility by means of a single index, such as the Pedersen index, appears to have
hardly any meaning.” (Grunberger 1985: 204)
“… comparing Total Volatility in post-communist countries to Total Volatility in established
democracies is essentially comparing apples to oranges.” (Powell & Tucker 2012: 27)
1. Introduction
The “earthquake election” of 1973 in Denmark was probably an impetus for Mogens Pedersen
to write his paper for the 1978 ISA World Congress in Uppsala, which was published in the
European Journal of Political Research 1979 and later republished as a book chapter in 1983.
One can say that the article from 1979 was the starting point for the research about electoral
volatility in Western Europe to be followed by other areas later on1. His conceptualization of
volatility as Aggregate Volatility (Vt; Pedersen 1980: 399)2 has become known as the
Pedersen index (Taagepera & Grofman 2003: 662).
Yet, the word volatility and its conceptualization was not unknown to Nordic political
scientists in the mid-1970s. When Erik Damgaard (1974: 21) prior to the 1973 election
analysed the stability of the Danish party system he characterized the party system of the
1960’s as featuring “increasing mobility and volatility”. In their analysis of the 1973
Norwegian election did Henry Valen and Stein Rokkan (1974: 217) note “increasing volatility
within national electorates” in Scandinavia. Commenting on the simultaneous election in
Sweden Olof Petersson (1974: 225) concluded: “Compared with the recent turbulences of the
Danish and Norwegian party systems, there were remarkably small changes in the 1973
Swedish election”. In his analysis of the 1976 election he added about the Swedish case:
“Aggregate stability is combined with a large and increasing individual volatility” (Petersson
1978: 113). It is also interesting to note that Carsten Jarlov and Ole P. Kristensen (1978: 62)
in a previous issue of Scandinavian Political Studies constructed a measure of aggregate
electoral mobility which equalled what Pedersen called Total Net Change (TNCt), but instead
of volatility they talked about “electoral fluidity”. It may also be something to note, that
according to Damgaard (1974: 108), Mogens Pedersen in his earlier research on legislative
recruitment in Denmark had employed the concept volatility then measuring “the extent of
change in the personal composition of the elite over time”. It was thus primarily in a
Danish/Scandinavian context where Pedersen’s theorizing and conceptualization of electoral
volatility took place, something that may be worth to remember.
1
Geographically the first studies of electoral volatility dealt with Western Europe, but the areas covered were
extended to Latin America (Coppedge 1998, Roberts and Wibbels, 1999) as well as Central and Eastern Europe
(Birch 2001, 2003; Sikk 2001, 2006; Tavits 2005, 2008a) and recently the focus is all-European (Dassonneville
& Hooghe 2011); among recent contributions one may cite: Shair-Rosenfield (2008), Mainwaring et al. (2010),
Epperly (2011), Bischoff (2012), Gherghina (2012), Jones (2012), Robbins & Hunter (2012).
2
This index was named an ”index of dissimilarity” by Duncan & Duncan (1955) and an ”index of distortion” by
Loosemore & Hanby (1971) (Taagepera & Grofman 2003: 662).
1
Pedersen’s focus was on electoral volatility measuring changes in party votes at the aggregate
level; the key concept employed in 1979 (p. 4) “volatility” was in 1980 (p. 399) and 1983 (p.
33) extended to “aggregate volatility” thus implicitly opening up for measurement of
volatility at the individual level (1980: 402). Crewe in 1985 (p. 8) added the analysis of seats
to the analysis of votes and he also introduced the distinction between net (aggregate level)
volatility and gross (individual level) volatility. A further step was taken by Bartolini & Mair
(1990: 22-23) when they broke down total (aggregate) volatility into block volatility and
within-block volatility. The problems associated with mergers and splits among parties have
been addressed in various ways; Birch (2003: 122-23; see also Janda & Kwak 2009:10)
introduced the concept party replacement which implied the exclusion of parties that did not
contest two consecutive elections; Mainwaring & Zocco (2007:123) established a few
calculating rules to accommodate for changes due to mergers and party splits; recently Powell
& Tucker (2012: 6-12) have distinguished between Type A (A new party) volatility between
nonstable parties and Type B (Between existing parties) volatility between stable parties3.
The aim of this paper is mainly empirical with a focus on measuring and calculating aggregate
electoral volatility among 31 European countries4 covering the period from after the Second
World War up till today. To be somewhat more precise the aims are5:
- to collect electoral data and compute new own estimates; the ambition is, for most elections,
to reduce the number of votes for other parties; make an effort to trace splits and mergers of
parties between the elections among the party systems; employ different measurement
formulas to capture aggregate electoral volatility;
- to map the variation in aggregate electoral volatility over time and in space based on the new
own estimates;
- to retrieve existing volatility scores/estimates for a reasonable number of elections reported
in the research literature;
- to detect deviating estimates in order to locate a number of critical elections to draw lessons
from;
- to draw conclusions from these enquiries with regard to how to improve the precision and
validity in measuring volatility scores.
2. Measuring aggregate volatility
In this section I will first present the aggregate volatility measures that I will employ; I will
then further discuss some of the rules I will apply for estimating these measures; the third step
involves discussing data employed, while the final step of this section will be a brief
presentation of the data sets constructed.
3
Type A volatility comes close to what Rose & Munro label demand volatility and type B volatility is similar to
what they name supply volatility (Rose & Munro 2009: 52; 2003: 77-86; see also Stískala 2012).
4
27 current EU-member countries plus Croatia, Iceland, Norway and Switzerland.
5
Among issues not dealt with in this paper I may name volatility in votes vs. seats and net volatility vs. gross
volatility. Neither will I contribute to an analysis of the causes behind the variation in volatility; before that there
is a need to establish reliable and valid scores of aggregate volatility in Europe.
2
a) Aggregate volatility measures6
The focus of this paper is on the Pedersen index as presented in his article from 1980 (p. 399);
(1) Total Aggregate Volatility (v) =
Two alternative measures have been suggested by Koppel & Diskin (2009: 282 & 284) here
labelled the lsq-index (or the Gallagher index) and 1-cosines index:
(2) lsq index =
(3) 1- cosines index (=1 – cos) =
To construct the type A and the type B volatility indexes Powell & Tucker (2012: 10) by
convention employ a 2%-threshold; “we defined all parties that received at least 2% of the
vote share in either the first or the second election to ‘count’ as a party. All other parties were
considered not to have been part of the party system …”. To compute these measures all
parties receiving less than 2% of the votes are deleted. Type A volatility captures volatility
from parties entering or exiting the party system:
(4) Type A volatility index (Powell & Tucker 2012: 6) =
Type B volatility captures volatility among stable parties which are contesting both elections:
(5) Type B volatility index (Powell & Tucker 2012: 7) =
To be able to compute these indexes decisions has to be taken about mergers and splits among
the parties in the party system, which will be discussed in the next section. Suffice here to say
that when computing these two indexes (type A and type B) I will follow the procedures
employed for computing the total aggregate volatility index (v).
b) Estimating aggregate electoral volatility: rules applied
For a pair of elections there may be changes among the parties participating at these two
elections.7 First, in the literature there are suggestions as how to classify the participating
6
The formulas listed below are copied from the articles/papers discussed in this section; later on in section
2.d)(iii)-2.d)(vii) I make an effort to explain in words how the formulas have been applied.
3
parties. Bartolini & Mair (1990:311) basically distinguishes between three types of new
parties, namely mergers, splits and truly new parties; Mair (1999: 216-17) makes a similar
distinction calling mergers for marriage, splits for divorce and “new” new parties for birth, i.e.
parties that “cannot be adequately regarded as having derived from either a merger among, or
a split from, pre-existing parties.” Hug (2001: 13) maintains these distinctions and adds the
category electoral alliances; to him genuinely new parties are those “which emerge without
the help from members of existing parties.” Sikk (2006: 57) also stick to these distinctions but
focussing on the genuinely new parties he defines them as parties that are “not successors of
any previous parliamentary parties, have a novel name as well as structure, and do not have
any important figures from past democratic politics among its major members.” Krouwel &
Lucardie (2008: 279) introduces a category “transformations” but otherwise keep with the
distinctions introduced by Bartolini & Mair.8 A more narrow and precise definition of a new
party, here I take it to refer to what is called genuinely new parties, is formulated by Barnea &
Rahat (2010: 311) to be a “party that has a new label and that no more than half of its top
candidates (top of candidate list or safe districts) originate from a single former party.”
Second, these distinctions are relevant when calculating the volatility scores, and in particular
this is so for the parties classified as mergers or splinter parties. Sikk (2005:393) identifies
three approaches for dealing with these categories of parties, namely: a) for a pair of elections
combined votes are calculated for split or merged parties9; b) for the split or merged parties a
larger and a smaller party is identified, and the smaller party is classified as the new party; c)
for the split or merged parties no predecessors are recognized, meaning that they will be
classified as new parties. I agree with Sikk that the first option a) is the least troublesome one,
and I will try to follow that approach when calculating the volatility scores. Third, choosing
this approach makes it extremely important to distinguish the genuinely new parties from the
split or merged parties. The definitions of Hug, Sikk and Barnea and Rahat gives some advice
for handling the complicated cases faced when calculating the volatility scores.10
Based on the deliberations discussed in the research literature there are some rules, or criteria,
that may be applied and which may facilitate the process of calculating the volatility scores;
the criteria I will follow are briefly commented upon here:
i. Continuous parties:
These are parties which are stable over at least one pair of elections, but often over a number
of pairs of elections. They are recognised as continuous parties in the literature, but in most
cases also in the election data; when a party has changed name but they still display an
obvious continuity with a previous party, they are counted as continuous parties (Mainwaring
& Zoco 2007: 173). Typical examples would be the Nordic Social Democratic Parties, but
only the Swedish SAP remains stable in a strict sense over the period studied.11 The scores for
these parties are generally easy to compute.
7
This means that I have made no effort to cover changes in turnout which may be captured through the use of
survey data.
8
Tavits (2008a: 122) closely follows the classification of Hug; Mustillo (2009: 317-18), however, differ
somewhat when applying the discussion of new parties to Latin America identifying five categories of new
parties.
9
Combined votes means that the comparison between the two elections refers to the vote for the parties in
question at election t0 and the same set of parties at election t+1.
10
Birch (2003: Appendix C pp. 185-186) discusses criteria for differentiating various types of parties although
not specifically addressing what constitutes a genuinely new party.
11
This means that neither “Nya Partiet” in 1979 nor “Junilistan” in 2006 are considered as splinter parties.
4
ii. Alliances:
For a particular election two or more parties may agree upon forming an electoral alliance;
such an alliance is recognised in the literature as well as noted in the election data. Examples
of such alliances are to be found in Finland (Kesk and Liberals 1983), Sweden (CP and KDS
in 1985) and the Czech Republic (Koalice 2002). The scores are based on comparisons of the
combined vote of the alliance partners at the election preceding the alliance (election t0) and
the elections following the end of a temporary alliance (election t+2) with the election when
the alliance is formed (election t+1).
iii. Mergers:
New parties are formed when two or more parties merge; these mergers are recognised in the
literature. Typical examples of mergers are to be found in the Netherlands (CDA formed by
KVP, ARP and CHU in 1977), Denmark (EL formed by DKP and VS in 1991) and the United
Kingdom (the Alliance, later Liberal Democrats formed by SDP and Liberals in 1983). The
scores are based on comparisons of the vote for the new party (at election t+1) with the
combined vote for the “merging parties in the election immediately preceding the merger”
(election t0) (Bartolini & Mair 1990: 311).
iv. Splits:
New parties are formed when a party splits into two or more parties; these splits are
recognised in the literature. Examples of splits are quite numerous also in the Nordic countries
as in Denmark (DKP split into DKP and SF in 1959), Norway (DNA split into DNA and SF
in 1960), Finland (PS split into PS and TPSL in 1959) but also in Austria (FPÖ split into FPÖ
and LF 1993, and later FPÖ into FPÖ and BZÖ 2003). The scores are based on comparisons
of the combined vote of the new parties (at election t+1) with that “of the original party in the
election immediately preceding the split” (election t0) (Bartolini & Mair 1990: 311).
v. Genuinely new parties:
Genuinely new parties are parties formed which are not merged or split up ones; these new
parties are often recognised in the literature. Genuinely new parties are in particular to be
found among party families like the green parties or populist parties of various shades.
Examples of such genuinely new parties may be localised to Sweden (KDS in 1964, MP in
1981, ND in 1990 and SD in 1986; see Bolin 2012a), Austria (Greens in 1983), Germany
(Greens in 1983), Estonia (RP in 2003) and Bulgaria (NSDV in 2001) (Hug 2001b, Sikk
2005b, Tavits 2008b). The score for a genuinely new party is computed from scratch meaning
that the volatility score will be based on the election figure received at the first election when
it participated (at election t0) to be compared with the preceding election (election t-1) where it
did not take part. It is no doubt that a major problem when computing the volatility scores is
to distinguish new parties formed as splits and mergers (where combined votes is applied)
from the genuinely new parties (where no combined votes is applied). A few critical cases
will be discussed later in the paper – section 4.c).
vi. Death of parties:
Parties for which there are no further electoral results, unless they have not merged with other
parties or entered an electoral coalition, are considered dead parties, which have exited the
party system. The score for the dead party is calculated by comparing the result for the latest
election (election tn) where it participated with the following election (election tn+1) where it
did not participate.
5
c) Data employed
There are two sets of data employed in this paper. First, we have the electoral data which
forms the basis for the computation of the scores measuring alternative operationalisations of
total (net) aggregate electoral volatility (see section 2.a) above). Second, we have a set of
estimates of aggregate volatility for European countries reported in the research literature; a
reasonable number of estimates have been collected for the 31 European countries to make it
possible to identify critical cases where the classification of parties and the calculation of
volatility scores diverge.
i. Electoral data
One need not go that many years back to recognise that electoral data for comparative
research was not easily available (cf. Lijphart 2005). The first edition of Mackie & Rose
(1974) covering the Western world simplified this task, and it was made easier by later
editions (3rd edition from 1991) and an update (from 1997); Central and Eastern Europe is
covered first by Rose, Munro & Mackie (1998), and then by Rose & Munro in a first (2003)
and a second edition (2009). These data handbooks have been complemented by Nohlen &
Stöver (2010) as well as the yearly reporting of the Political Data Yearbook published by the
European Journal of Political Research.
These published sources may be complemented by a number of websites containing
information about electoral data with a comparative scope.12 The ones I have benefited from
are Bochsler, CIAS, Popescu & Hannavy and Nordsieck. In some instances Wikipedia, the
English as well as the national ones, may contain useful complementary information.
The main sources used are, however, the websites maintained by the national electoral
authorities. In some cases the data is complete and easily accessed (Austria, Germany, Spain,
Portugal, Czech Republic, Slovakia, Slovenia and Estonia, to cite some examples), whereas in
other cases only a brief period may be covered from these websites (Denmark, United
Kingdom, France, Ireland and Luxembourg).
All in all, a quite detailed data set has been assembled where the number of anonymous other
parties have been minimised; the data sources employed are listed in Appendix A.
ii. Aggregate volatility estimates in the research literature
Beginning with Crewe & Denver (1985) and followed by Bartolini & Mair (1990) a number
of estimates of total aggregate volatility scores for specific elections in various countries have
been presented in the research literature, or published online on the Internet. Through
different searches I have been able to retrieve volatility scores for 390 elections for the 31
countries covered in this paper; on an average some seven studies have been utilised for each
country. Some countries, however, like the Czech republic, Slovakia, Poland and Hungary are
covered more intensely, whereas countries like Croatia, Cyprus and Malta more seldom are
studied with respect to electoral volatility. In Appendix B the studies made use of are listed.
12
These and other websites have been accessed over the period from November 2011 to July 2012; a final check
of the addresses has been conducted by mid August 2012.
6
d) Data sets constructed13
To arrive at the different measures of volatility scores employed in this paper a number of
data sets have been constructed. Below I will briefly deal with these data sets covering thirtyone European countries:
(i) – election: Election wise from the first to the latest election, for each election, data on valid
votes received (sorted from highest to lowest figures) as well as in percent for each party has
been assembled; in addition, when completed, information about Mackie & Rose-code or
Rose & Munro-code, abbreviation for the party and party name are listed. To make sure of the
accuracy of the data, summaries of valid votes and votes in percent (summarizing to 100,00)
are added. There are thus eight columns and the number of rows equal number of parties
participating at an election (from a low number of 76 in Cyprus to the highest number of 816
in Spain).
(ii) – data: The data from the previous data set (i) is now sorted by party following relevant
codes and alphabetic orders; thus each party is located on a row, and on the columns votes
received in percent are reported; and for each column is asserted that it summarizes to 100,00.
The ambition has been to reduce the number of votes for “other parties” as far as possible.
The number of parties listed varies from a low 30 in Denmark to a high 453 in Spain.
(iii) - volatility: The data from data set (ii) is now organised so that it is possible to establish
pairs of elections from the first to the latest election, i.e. for the first election a column
(designated _fw which stands for forward) is added to match the following election, and for
the latest election another column (designated _bw which stands for backward) is added to
match the preceding election. Thus for each election for which a volatility score is to be
estimated there are two columns to be compared for each party, one _fw and one _bw. These
are the columns which are modified to adjust to what is demanded from the occurrences of
splits or mergers or the rise of genuinely new parties as well as the death of old parties. As far
as possible the necessary computations in these columns are indicated with different colours,
and the formulas used may be noted from the excel document (when completed the formulas
for the computations will be listed in a separate document). The volatility score, i.e. the
Pedersen index, is then arrived at through the following computation: for each pair of columns
for the relevant elections (_fw and _bw) the absolute sum is computed and then summarised
for all parties, and this sum is then divided by 2. It is this data set that forms the basis for the
other volatility measures computed.
(iv) – lsq-measure: Based on the previous data set (iii) the lsq-measure is arrived at through
the application of another computational formula; first for each pair of columns for the
relevant elections (_fw and _bw) the squared sum is computed and then summarised; the lsqscore is then arrived at by taking the square root from the previous sum divided by 2.
(v) – 1-cos-measure: Based again on the volatility data set (iii) this measure is arrived at
through the application of yet another computational formula; for each pair of columns for the
relevant elections (_fw and _bw) three sums are computed; first, the product of the pairs are
summarised (=sum 1); second, the square root of column_fw is computed and then
summarised (=sum 2); third, the square root of column_bw is computed and then summarised
(=sum 3); finally, the score is arrived at using this formula: 1 – (sum 1 / sq rt (sum 2 * sum3)).
13
Since this paper is a ”work in progress” no data has as yet been posted on the Internet, but anyone interested in
receiving the data sets, as they stand, may contact me on my e-mail: [email protected]; I expect to be
available on this address till around 2013-06-30.
7
(vi) - type-A volatility: Following Powell & Tucker (2012) the data set (iii) once again form
the basis for the calculation; for each pair of columns for the relevant elections (_fw and _bw)
figures lower than 2 are deleted as well as every stable pair with figures of 2 or higher; what
remains are pairs where for one of the elections the figure is lower than 2, and for the other
election the figure is higher than 2. The figures remaining in the data set forms the basis for
calculating type-A volatility in the same way as applied for the Pedersen index (see iii above).
(vii) – type-B volatility: Again following Powell & Tucker (2012) the data set (iii) is once
again used as the basis for the calculation; for each pair of columns for the relevant elections
(_fw and _bw) figures lower than 2 are deleted as well as pairs where for one of the elections
the figure is lower than 2, and for the other election the figure is higher than 2; what remains
are every stable pair with figures of 2 or higher. The figures remaining in the data set forms
the basis for calculating type-B volatility in the same way as applied for the Pedersen index
(see iii above).
(viii) – aggregate volatility estimates reported in the research literature: for each election
where volatility estimates are reported in the literature it is possible – in addition to number of
studies used (=N) - to compute the following statistical measures: median, mean, standard
deviation and coefficient of variability (=cv; computed as standard deviation/mean).
These are thus the data sets to be employed for mapping volatility in space and over time as
well as mapping variation in volatility estimates among 31 European countries.
3. Mapping variation in aggregate volatility in space and over time
Based on the data sets presented above the focus in this section is to map the variation in
volatility among the European countries in space and over time. But let us first look at how
the different measures and the different operationalizations of volatility are related to each
other.14
a) Volatility measures: v, lsq and 1-cos
In addition to the commonly employed Pedersen index (v) measuring total aggregate volatility
a few alternative measures has been suggested in the literature. Two alternative measures are
tested here, namely the least square measure (lsq) and the cosine measure (1-cos). It is
obvious that all three of them tend to go together which is indicated by correlations close to
,90, and these relations are valid in the set of countries making up Western Europe as well as
Central and Eastern Europe – see Table 1 below.
14
An effort has been made to reduce the size of support for “other parties” through the listing of as many parties
as possible available in election data. So far I have only conducted a preliminary analysis where I have separated
parties that have been classified as “other parties” to find out what impact they may have on the volatility scores.
On an average volatility scores for this group of parties summarises to 1,7 which is slightly more than 10% of the
total aggregate volatility scores estimated. For some countries like Spain, Czech Republic, Romania and
Slovakia this amounts to a certain impact, whereas for other countries like Sweden, Norway and Finland the
impact is negligible.
8
Table 1: Correlation between v, lsq and 1-cos by all and WE and CEE
v
All (n=395)
lsq
1-cos
r=,960 r=,893
WE (n=334)
lsq
1-cos
r=,956 r=,872
CEE (n=61)
lsq
1-cos
r=,919 r=,908
Generally the scores display higher values for v than for lsq, and this is also true when
comparing v and 1-cos. It is, however, only for a few elections that one may note deviating
estimates depending on the measure. In figure 1 the Bulgarian election 2001, the Polish one
for 1993, the Slovenian for 1992 and the Cyprus election of 1976 are noted; the Cyprus case
is affected by a small number of parties taking part in the election, whereas the opposite is
true for the Polish election indicating that the measures are slightly dependent on the number
of observations – parties in this case.15
Figure 1: Relation between measures of volatility: v and lsq
Also in Figure 2 do the two measures go quite well together, although here we need to move
from a linear to a quadratic relationship. The two Bulgarian elections of 2001 and 2009 are
distinguishing themselves through high volatility scores, whereas the v-scores are
distinctively higher than the 1-cos-scores for the 1947 Maltese election and the 1992
Romanian election; at both of these elections there is a major party losing a substantial part of
their vote.
15
The impact of number of parties on v and lsq is noted by Sikk (2001: 23).
9
Figure 2: Relation between measures of volatility: v and 1-cos
One conclusion from this exercise is that all three measures go together quite closely. Thus
one may infer that it is not a major issue which measure to employ. I thus tend to agree with
what Pedersen (1980: 397) maintained already from the beginning, although he was then
discussing fragmentation indices, namely that when estimating scores one should stick to the
established ones as long as there are no major reasons for abandoning them. Therefore, I
would say, it is not meaningful to replace v with any other general measure, not least since
most studies within the field employ the v measure when measuring total aggregate volatility.
b) Variation in space: by country and by region
When moving on to the mapping of volatility among the European countries the v measure
will be employed. To start with different operationalizations of volatility will be dealt with,
namely the relation between v and its subtypes, i.e. type A and type B. How are these three
types of volatility related to each other? The correlations displayed in Table 2 below refer to
all of Europe as well as to the two regions we have identified (WE and CEE).
Table 2: Correlation between v, type A and type B by all and WE and CEE
v
Type A
type A
r=,769
All (n=395)
type B Type A+B
r=,876
r=,986
r=,404
r=,803
type A
r=,655
WE (n=334)
type B Type A+B
r=,889
r=,984
r=,279
r=,701
type A
r=,769
CEE (n=61)
type B Type A+B
r=,676
r=,972
r=,109
r=,812
It seems to be the case that total volatility (v) correlates relatively strongly with type A, type
B and type A+B in all three contexts, somewhat stronger for all Europe than for CEE, but in
CEE v correlates slightly stronger with type A than with type B. As might be expected type A
and type B are not strongly correlated neither in WE nor in CEE. Thus, the volatility due to
the entry of new parties is something different from the volatility associated with stable
parties. Yet, it is still interesting to note, from this table, that total volatility (v) seems to
10
capture a variation in volatility that is not strongly different from what is captured by type A
and type B volatility.
Let us therefore take a closer look at the variation between countries for these different
operationalizations of aggregate volatility. The mean values for the four types of volatility are
displayed in Table 3 and in each column sorted from the lowest to the highest score. In
addition, the table also contains information about what amount of variation in these scores
that may be explained by country as it is summarized in the eta-scores.
Table 3: Variation in v, type A, type B and total A+B by country (sorted by size)
country
unitedking
switzerland
austria
finland
sweden
germany
malta
iceland
luxembourg
norway
belgium
denmark
italy
greece
portugal
nertherlands
cyprus
ireland
spain
france
slovakia
hungary
czech rep
poland
croatia
estonia
slovenia
romania
latvia
bulgaria
lithuania
Total
Eta sq
v
6,28
7,31
7,49
8,04
8,47
8,82
9,55
9,70
9,98
10,26
10,37
10,38
11,76
12,42
12,50
12,99
13,02
14,25
14,32
16,67
21,26
22,85
23,26
24,62
24,63
25,83
28,54
31,71
34,90
36,75
41,33
13,31
,550
country
unitedking
sweden
norway
austria
finland
germany
iceland
portugal
spain
denmark
belgium
greece
luxembourg
switzerland
nertherlands
italy
malta
france
hungary
croatia
cyprus
poland
romania
slovakia
ireland
slovenia
czech rep
estonia
bulgaria
latvia
lithuania
Total
type A
0,36
0,61
0,65
0,72
1,13
1,23
1,34
1,34
1,59
1,74
1,89
2,11
2,18
2,63
2,89
3,04
3,18
3,81
4,16
4,67
5,53
6,18
6,78
6,81
7,11
7,52
8,20
9,39
12,95
13,33
17,73
3,20
,410
country
switzerland
unitedking
austria
malta
germany
finland
ireland
belgium
luxembourg
cyprus
sweden
italy
iceland
denmark
greece
spain
nertherlands
norway
portugal
france
slovakia
czech rep
croatia
estonia
poland
slovenia
hungary
bulgaria
latvia
romania
lithuania
Total
type B
3,76
5,09
6,00
6,04
6,34
6,37
6,76
7,20
7,29
7,37
7,45
7,68
7,94
8,67
8,79
9,18
9,27
9,30
9,89
11,16
11,66
12,68
12,77
14,71
16,46
17,21
17,62
18,33
18,63
19,21
22,38
8,89
,390
country
unitedking
switzerland
austria
finland
germany
sweden
belgium
malta
iceland
luxembourg
norway
denmark
italy
spain
greece
portugal
nertherlands
cyprus
ireland
france
croatia
slovakia
czech rep
hungary
poland
estonia
slovenia
romania
bulgaria
latvia
lithuania
Total
total A+B
5,45
6,39
6,72
7,50
7,58
8,06
9,10
9,22
9,28
9,47
9,96
10,42
10,72
10,77
10,90
11,23
12,16
12,90
13,87
14,98
17,45
18,47
20,88
21,78
22,64
24,10
24,73
25,99
31,28
31,96
40,11
12,10
,491
What is obvious is that there is a distinct variation between countries as indicated by the etascores. Further, one may also note that the highest scores are to be found among the CEEcountries. It is only in terms of type A volatility that two non-CEE countries ranks high,
namely Cyprus and Ireland16. Still there are WE-countries which score high on volatility like
France and Spain but countries displaying low scores are all from Western Europe like United
16
The high scores for Ireland has to do with the fact that that the group of independents for each election are
treated as genuinely new parties; if treating the group of independents as stable parties the score for type A goes
down from 7,11 to 1,83.
11
Kingdom, Switzerland and Austria. It is also noteworthy that v generally, with the exception
of Denmark17, displays higher scores than for total A+B; the reason is that following the
coding principles of Powell and Tucker (2012) parties with less than 2% in voter support are
not counted.
The variation by region is mapped in Table 4 where the mean values for Western Europe
differs sharply from those for Central and Eastern Europe. Yet, somewhat surprisingly, the eta
scores are significant but not that high and the reason is that there is also a distinct variation
between countries within Western Europe.
Table 4: Variation in v, type A, type B and total A+B by region
grp
WE
CEE
Total
Eta sq
v
10,57
28,32
13,31
,430
type A
2,19
8,73
3,20
,229
type B
7,54
16,27
8,89
,279
total A+B
9,74
25,00
12,10
,363
To summarize, there is a substantial variation in aggregate volatility all over Europe, and this
is true for all kinds of operationalizations used, countries in CEE displaying higher scores
than countries in WE. Type B volatility is higher than type A volatility, although less so in
Western Europe; the two types of volatility (A and B) captures different aspects of volatility,
but they do also match the total scores. So the next step is going from variation in space to
variation in volatility over time.
c) Variation over time: by decade
Looking at the variation in volatility over time we may establish a pattern (Table 5). Volatility
goes down from the late 1940s till the 1970s when there is a rise, temporarily broken in the
1980s to increase again with the 1990s where the highest scores are to be noted for the present
decade, the 2010s. Thus this trend pattern captures the “earthquake elections” of the 1970s as
well as the entry of the CEE countries with the 1990s. The eta scores suggest that the
variation over time is most pronounced for the v measure whereas the type A volatility
displays a low eta score. It is also clear that the variation in volatility is stronger in terms of
spatial variation than with respect to the temporal variation.
Table 5: Variation in v, type A, type B and total A+B by decade
decade
1940s
1950s
1960s
1970s
1980s
1990s
2000s
2010s
Total
Eta sq
v
11,54
8,82
7,92
10,59
10,23
16,49
17,79
22,25
13,31
,188
type A
4,09
1,72
1,77
2,63
2,32
4,07
4,15
6,11
3,20
,058
type B
6,90
6,56
5,72
7,34
7,13
10,45
11,75
14,12
8,89
,173
total A+B
10,99
8,28
7,49
9,97
9,44
14,52
15,91
20,23
12,10
,156
17
To illustrate the Danish exception we may look at the votes for Retsforbundet in 1945 (= 1,86) and in 1947
(4,54) resulting in a v of 2,66/2 and a type A of 4,54/2.
12
The pattern of variation for aggregate volatility seems to fit with an expected pattern where
CEE countries has higher scores than the WE countries and the 1970s and the 2010s are
decades to distinguish. But how reliable are the scores estimated? May there be systematic
biases in the estimates presented in Table 3 to Table 5?18 One way to enquire into this
problem somewhat further is to compare these estimates with other estimates reported in the
research literature, and based upon discrepancies among estimates scrutinise and analyse a
few critical elections from which we may draw lessons and learn from mistakes. This is the
task for the next section of the paper.
4. Mapping variation in estimates of aggregate volatility
Let us now continue with mapping the variation in estimates of aggregate volatility as
reported in the research literature. The purpose with this section is first to identify groups of
elections, and countries, where there is a general agreement about the level of aggregate
electoral volatility as well as where there are major discrepancies in the volatility scores
estimated. Secondly, a comparison between my own estimates and those reported in the
literature makes it possible to select a number of elections where there are disagreements.
Thirdly, based on this comparison a few critical elections may be scrutinised more closely in
order to gain an understanding of what choices may have had an impact on the estimates
arrived at.
a) Variation in estimates in the research literature: standard deviation and coefficient of
variability
Two measures are employed to map the variation in estimates of volatility. The size of the
standard deviation is conditioned by the size of the volatility scores, whereas the size of the
coefficient of variability is normalised through the division of the standard deviation with the
mean. The two measures correlate quite strongly (r =,80), but as may be noticed in Table 6 the
order among the countries vary slightly when the two measures are sorted by size.
18
For my own curiosity I have compared these estimates with previous estimates I have made; the estimates
reported in this paper have been computed from a new dataset and, as far as possible, these estimates have been
made independent from those reported earlier. Depending on disposition the correlations reported may therefore
be characterized as worrying or encouraging: v and Ersson & Lane 1998 (r= ,92) and v and Lane & Ersson 2007
(r=,95).
13
Table 6: Variation in estimates in the research literature of total volatility by country
country
norway
denmark
sweden
malta
iceland
germany
finland
ireland
spain
austria
slovenia
unitedking
lithuania
switzerland
luxembourg
cyprus
nertherlands
poland
latvia
bulgaria
greece
estonia
czech rep
belgium
croatia
romania
slovakia
france
portugal
hungary
italy
Total
cv
0,04
0,04
0,05
0,07
0,08
0,08
0,09
0,10
0,11
0,12
0,12
0,13
0,14
0,14
0,15
0,16
0,17
0,18
0,19
0,20
0,22
0,22
0,24
0,27
0,28
0,29
0,30
0,30
0,34
0,36
0,37
0,16
country
malta
norway
denmark
sweden
germany
austria
unitedking
ireland
finland
switzerland
iceland
spain
luxembourg
nertherlands
greece
belgium
slovenia
cyprus
france
poland
portugal
hungary
lithuania
italy
czech rep
bulgaria
estonia
latvia
romania
croatia
slovakia
Total
stddev
0,27
0,41
0,42
0,43
0,68
0,69
0,74
0,74
0,83
0,93
1,26
1,53
1,88
2,29
2,36
2,99
3,16
3,39
4,56
5,59
6,00
6,18
6,33
6,45
6,76
6,80
7,07
7,30
8,33
8,59
9,36
2,56
What is obvious is that there is indeed a variation among the European countries with respect
to estimates in the research literature on aggregate volatility scores. There seems to be small
disagreements for the Nordic countries, whereas there is more of a discrepancy in the
estimates for countries in Central and Eastern Europe but this is also true for Italy, France and
Portugal. To gain a better understanding of these discrepancies it is necessary to enquire
closer into different estimates of the aggregate volatility scores.
b) Variation in estimates of aggregate volatility: median scores and new own scores (v)
The estimates to compare are on the one hand the ones I have presented earlier (v in Tables 3
to 5) and the median scores for volatility in the research literature as summarized above
(Table 6). I prefer the median scores to the mean scores since they bypass some extreme
outliers. However, to start with Table 7 displays the correlation between v and median and
mean scores for the whole sample (all) as well as for the WE-set and the CEE-set of countries.
14
Table 7: Correlation between v, median and mean by all and WE and CEE
v
All (n=390)
median
mean
r=,903
r=,855
WE (n=330)
median
mean
r=,877
r=,835
CEE (n=60)
median
mean
r=,71
r=,688
From the correlations we may note that there are discrepancies in the estimates of v and
median, less so in WE than in CEE; and the correlations between v and median are stronger
than what is the case for v and mean. Disaggregating the correlations between v and median
to the level of countries there are a few cases displaying distinct differences influencing the
overall correlations: Croatia (few observations), Slovakia (the 1992 and 2002 elections), but
also Cyprus (few observations), Iceland (the 1999 election) and Poland (the 2001 election).
To present an overview of differences in estimates Table 8 contains information about the
mean values for median (i.e. estimates from the research literature), difference (absolute)
between estimates in v and median (i.e. means for all elections); to normalize the figures for
presentation the column diff is divided with the column median, and the countries are sorted
by this variable (diff/median).
Table 8: Comparison of different estimates of aggregate volatility
country
median
8,38
diff: v vs. median
0,31
diff/median
nertherlands
12,59
0,51
0,040
germany
8,47
0,37
0,044
hungary
23,56
1,29
0,055
norway
10,56
0,64
0,061
lithuania
42,17
2,66
0,063
portugal
11,89
0,75
0,063
austria
7,41
0,51
0,069
denmark
11,02
0,79
0,071
finland
8,21
0,71
0,087
czech rep
22,38
2,09
0,094
latvia
36,51
3,66
0,100
spain
12,92
1,40
0,108
italy
12,69
1,57
0,124
greece
10,33
1,29
0,125
slovenia
29,65
4,09
0,138
belgium
9,14
1,31
0,143
unitedking
6,39
0,94
0,147
luxembourg
11,52
1,91
0,166
bulgaria
33,08
5,88
0,178
estonia
26,09
4,71
0,181
france
15,20
2,78
0,183
poland
29,97
5,87
0,196
iceland
11,69
2,63
0,225
slovakia
27,61
6,44
0,233
switzerland
6,12
1,44
0,236
malta
10,08
2,39
0,237
romania
25,10
6,76
0,269
ireland
9,75
4,50
0,461
cyprus
15,43
7,26
0,471
croatia
23,92
15,25
0,638
Total
13,14
2,14
0,163
sweden
0,037
15
Some of the countries ranking high may be disregarded either because of few observations
(Croatia and Cyprus), or due to the handling of independents (Ireland and Switzerland)
whereas there are other countries for which there are reasons to scrutinize some of the
elections in more detail (Slovakia, Iceland, Poland and Estonia). On the other hand it is also
the case that there are countries where disagreements are less pronounced, and this is true for
WE-countries as well as for CEE-countries, but also among these cases there may still be a
few elections to take a closer look at.
These comparisons indicate that by and large my estimates tend to go together with what has
been reported in the research literature, but there are also a lot of cases, countries as well as
elections, which require a closer look to draw conclusions about what may lay behind
different estimates of volatility scores. The next section will scrutinize a few critical elections
in these respects.
c) Critical elections to scrutinize
To gain an understanding for why there are differences in estimates of the volatility scores it
is necessary to go into details. By looking closer at elections that may be deemed as critical,
since there are sometimes huge differences in the scores estimated, and to find out which
decisions that lie behind a specific estimate may elucidate the more general decision
principles that should be applied. The elections, and countries, selected in this section is based
on the size of discrepancy identified (Table 8), but not exclusively so. There is one country
(Romania) which displays a relatively high score of difference but on closer inspection it is
difficult to pinpoint why the estimates differ. There are other countries (Denmark, Finland and
Norway) for which the difference score is small, but there are nevertheless some interesting
elections to look at. Overall the countries, and elections, presented in Table 9 below may still
illustrate some of the difficult decisions that have to be made when classifying changes in
party systems such as identifying party splits and party mergers as well as the occurrences of
birth of genuinely new parties and the death of old parties.
For a sample of 27 elections data on different kinds of volatility scores are presented; first,
there are my own original estimates (v) employed in this paper (and presented in Table 3 to
5); second, there are alternative estimates (denoted by an subscript _alt) inspired by what has
been reported in the research literature; third, a measure of absolute difference is reported in
the last column taking the absolute difference between v and alt_v, and the elections are
sorted according to the size of this measure; fourth, I have also added the mean scores (see
Table 9 below) which includes outliers and makes it possible to relate v, v_alt to the mean.
From the Table it is very clear that decisions about how to classify parties over a pair of
elections may implicate huge differences in the estimations of the volatility score. I will
comment these data country wise thus starting with the Czech Republic and end with
Norway.19
19
A crucial distinction that will be employed refers to what is counted as splinter parties and what is classified as
genuinely new parties. I have encountered a few attempts to list parties in Europe where this distinction is
applied (Mair 1999, Hug 2001b, Sikk 2005b, Bolleyer 2010, Bolin 2012b; unfortunately Tavits 2008b does not
fully apply this distinction). Within brackets ([]) I will add whether my classification applied agrees or disagrees
with the ones reported in these studies; it is important to stress that agreements/disagreements refers to my
interpretation of how the authors classify the parties in question as being splinter parties or genuinely new
parties, and nothing else/more.
16
Table 9: Differences in estimates of volatility for a number of elections due to changes in
classifications of party splits and party mergers (v vs. v_alt)
country
election
v
type A
czech rep
1992 19,44 8,51
iceland
1999 4,31
0,00
slovakia
1992 20,75 3,26
malta
1950 23,24 9,33
poland
2001 23,91 6,19
slovakia
2002 18,53 8,01
latvia
2006 17,83 1,05
iceland
1987 12,73 3,65
slovenia
2008 26,45 2,79
malta
1962 13,40 0,00
lithuania
2008 27,64 12,71
italy
1948 16,51 3,64
france
1973 23,81 10,28
france
1967 13,14 6,82
estonia
2003 34,68 18,33
iceland
1983 11,29 2,74
luxembourg
1968 4,93
0,00
greece
1993 8,80
0,00
iceland
1995 8,66
0,00
italy
1996 9,48
1,76
denmark
1998 8,21
1,26
hungary
2006 4,76
1,08
greece
2000 5,14
0,00
iceland
1953 6,90
3,01
denmark
1960 9,18
0,00
finland
1958 4,73
0,00
norway
1961 2,01
0,00
type B total A+B
10,67
19,17
3,15
3,15
16,11
19,37
13,47
22,79
15,75
21,94
7,92
15,94
12,98
14,03
12,01
15,66
21,20
24,00
13,14
13,14
13,99
26,70
10,85
14,49
12,46
22,74
5,83
12,65
14,97
33,29
6,81
9,55
4,72
4,72
6,89
6,89
6,11
6,11
6,39
8,15
6,04
7,30
3,34
4,42
3,86
3,86
3,88
6,90
8,61
8,61
5,24
5,24
1,93
1,93
v_alt mean
62,38 36,5
44,57 28,1
50,1 33,2
46,44 45,7
46,10 50,6
31,83 42,0
30,86 27,1
23,59 21,5
36,68 33,4
22,88 22,9
36,67 34,4
23,58 32,0
17,19 13,6
6,92 12,1
28,93 30,6
17,02 15,7
10,28 10,2
13,68 13,8
13,41 11,5
14,22 14,5
12,22 11,1
8,33 12,5
6,8
8,09
9,3
9,30
11,8
11,19
6,4
6,47
3,4
3,58
type_A_alt type_B_alt total_A-B_alt abs_diff
54,72
4,90
59,62
42,94
38,90
4,32
43,22
40,26
39,73
16,11
55,84
29,35
20,92
27,67
48,59
23,20
18,30
25,82
44,12
22,19
22,08
6,94
29,02
13,30
8,30
18,76
27,06
13,03
9,08
12,01
21,09
10,86
7,48
25,85
33,33
10,23
9,38
13,25
22,63
9,48
13,85
20,40
34,25
9,03
7,17
13,39
20,56
7,07
3,65
12,46
16,11
6,62
3,16
3,27
6,43
6,22
14,54
13,00
27,54
5,75
6,39
8,89
15,28
5,73
2,88
7,60
10,48
5,35
2,44
9,33
11,77
4,88
3,58
6,74
10,32
4,75
7,36
4,74
12,10
4,74
4,96
7,56
12,52
4,01
5,55
2,61
8,16
3,57
1,47
5,33
6,80
2,95
4,65
4,65
9,30
2,40
4,64
7,10
11,74
2,01
0,00
6,47
6,47
1,74
1,20
2,31
3,51
1,57
Note: In this sample v correlates with v_alt (r=,64) and with mean (r=,75); v_alt correlates
with mean (r=,87)
i. Czech republic
- 1990-1992: At issue for the election pair 1990 – 1992 is how to classify the Civic Forum
(OF) and the parties which emanated from OF in the 1992 election. If the new parties are
classified as splinter parties then one may arrive at my original estimate (v = 19,44); if the
new parties are classified as genuinely new parties, then the alternative estimate would be
more appropriate (v_alt = 62,38). One position is the one taken by Powell & Tucker (2012:
12) who argue:
To take an example, we can consider the fragmentation of the Civic Forum in the Czech
part of Czechoslovakia, which took place prior to the 1992 parliamentary election. In 1990,
the pro-democracy Civic Forum dominated the election taking 53:15% of the vote. The
movement subsequently divided into four parties to contest the 1992 election: the Civic
Democratic Party (33:9%), the Civic Democratic Alliance (5:0%), the Civil Movement
(4:4%) and the Club of Engaged Non Party Members (2:0%). In this case, no party kept
the original name and there was no single clear successor party. So all four parties are
considered new parties, and thus all contributed to Type A Volatility.
In a similar vein Mansfeldová (2004: 227) writes that the changes in the political system led
to the “disintegration of the Civic Forum into several new parties.”
17
On the other hand one could argue that the four parties in question were splinter parties
emanating from the Civic Forum. Rose and Munro (2009:101) writes that Civic Forum “[i]n
1991 split into Civic Democratic Party, Civic Movement, Civic Democratic Alliance, and
Club of Active Non-Partisans.” Tóka (1997:14-15) classifies OH as a successor of OF and
KAN as part of OF in 1990 but is in doubt about how to classify OF. Wightman (1993: 84)
classified ODS and ODA as the “main successor parties” to OF. Jehlička et al. (1993: 253)
discussing the 1992 election write: “The ‘new’ significant parties are /…/ direct successors to
Civic Forum (ODS, ODA, OH) /…/.” And Klíma (1998: 497) write about the succumbing
Civic Forum which was “thus split into three successor parties: the Civic Democratic Party
(ODS), Civic Democratic Alliance (ODA), and Civic Movement (OH).”
What it boils down to is what is meant with genuinely new parties, with successor parties, and
what implications there are if we identify splinter parties. Applying strict criteria these four
parties may be classified as genuinely new parties; yet, my reading of Bartolini & Mair (1990:
311) where they state that when “a party splits into two or more parties, the relevant electoral
volatility is computed by subtracting the combined vote of the new parties from that of the
original party in the election immediately preceding the split”, support the interpretation that
Civic Forum was splinted into four parties where the votes for these parties in 1992 should be
compared with the vote for Civic Forum in 1990. Thus, in my opinion, it is still an open issue
which of the two approaches that should be considered the “correct” one. [Sikk 2005 agrees]
ii. Iceland
Not less than five elections display relatively high figures for the difference score reported in
Table 9. I will briefly comment upon each of the five pairs of elections chronologically:
- 1949-1953 elections: Although the magnitude for difference is low, at issue is how to
consider the Republic Party [mr#18]20; as a splinter from the Progressive Party [mr#18] as
suggested by Mackie and Rose (1991: 211) or as a new party (as judged from the scores
arrived at by Pétursson 2005: 153)? My choice is splinter party which implies applying
combined votes for the 1949 and 1953 elections. [Hug 2001 agrees]
- 1979-1983 elections: Here I consider the Social Democratic Federation [mr#22] a splinter,
or a breakaway, party from the Social Democrats [mr#7] (Mackie & Rose 1991:211); if so the
v-measure is 11,29; if classified as a genuinely new party v_alt is 17,02. [Mair 1999, Hug
2001, Bolin 2012 agrees]
- 1983-1987 elections: At issue is, mainly, how to classify Citizens’ Party II [mr#24]; is it a
genuinely new party, or can it be looked upon as a splinter from the Independence Party
[mr#13]? My option is the second one. [Hug 2001 agrees and Mair 1999, Bolin 2012
disagrees]
- 1991-1995 elections: At stake here is People’s Movement [mr#29]; is it a genuinely new
party or a breakaway from the Social Democrats [mr#7]? Mackie and Rose characterize it as a
breakaway party, which I agree with, and consequently apply the combined votes for the two
elections for the two parties. [Bolin 2012 agrees]
- 1995-1999 elections: Here we have a complicated pair of elections and depending upon how
parties are classified volatility may go from the low 4,31 (=v) to a high 44,57 (=v_alt). The
1999 election involved a process of splits and mergers mainly on the left-wing scene. At the
1995 election there were four parties to the left, namely Social Democrats [rm#7), People’s
Alliance [rm#14], Women’s Alliance [rm#23] and People’s Movement [rm#29]. The aim was
to arrange a United Front (Social Democratic Alliance) of these four parties before the 1999
election but the left wing of the People’s Alliance [rm#14] withdraw and formed a Left20
Refers to the code adopted by Mackie & Rose (1991).
18
Movement-Greens (Hardarson & Kristinsson 2000: 412; Kristjánsson & Indridason 2011:
163). On one hand the two new parties may be classified as genuinely new parties or on the
other hand they may be classified as mergers where the two parties emanated from the four
parties; I have opted for the second alternative, but I am aware of that this may be a case
where the combining of votes from the two elections stretches the meaning of splits and
mergers a bit too far. [Bolin 2012 agrees]
iii. Slovakia
Here are two elections in need of a closer investigation:
- the 1990-1992 elections: The status of VPN is similar to that OF in the Czech republic;
should HZDS and ODÚ be classified as successor parties to VPN or classified as genuinely
new parties? If the first option is chosen v scores 20,75 whereas if the second option is
applied the score for v_alt is 50,1. Rose & Munro (2009: 116) classifies both parties as
splinters from VPN; Jehlička et al. (1993: 253) call the two parties as direct successors to
Public Against Violence; and Bureš & Just (2010: 75) write about VPN and “their successor
parties, the HZDS and ODÚ”. Further, following the detailed party family tree outlined by
Deegan-Krause (2011) these two parties are recognized as splits from VPN. Thus I find it
reasonable to treat VPN and its successor parties in the same way as is the case with the
Czech Civic Forum. [Sikk 2005 agrees]
- 1998-2002 elections: At this pair of elections the focus is on the Smer party; is Smer a
splinter from SDL or should Smer be considered a genuinely new party? The volatility score
varies from v=18,53 to v_alt = 31,83 whether Smer is classified as genuinely new party or
not. In his detailed analysis of this pair of elections Výtisk (2007: 35) treats Smer as a
genuinely new party. Yet, it is the case that one can call Smer a splinter party from SDL
(Deegan-Krause 2011). We also know that Fico had a leading position within SDL prior to
the formation of Smer in 1999; later in 2005 SDL merged with Smer (Passin 2012: 8). In my
opinion, it is therefore not unreasonable to treat Smer as a splinter party and apply a formula
for combining the votes of SDL and Smer in this pair of elections.
iv. Malta
Here are at least two pairs of elections that merit attention:
- 1947-1950: The issue for this pair of elections is Malta Workers’ Party (MWP); is it a
splinter party from the Labour Party or a genuinely new party? Mackie & Rose (1991: 313)
notes that MWP was formed “in October 1949 by Dr. Paul Boffa, the former leader of the
Labour Party”, and McHale & Skowronski (1983: 633) adds that he “resigned as leader of the
Malta Labour Party (MLP) following an MLP vote of no confidence.” My interpretation is
that it could be treated as a splinter party; if not the volatility scores goes up from v=23,24 to
v_alt=46,44.
- 1955-1962: The two parties at issue for this pair of elections are Democratic Nationalist
Party II (DNP) and Christian Workers Party (CWP). As for DNP II Mackie & Rose (1919:
313) notes that it was formed by “former members of the Nationalist Party” and it is classified
as a “progressive offshoot of the Nationalist Party” by McHale & Skowronski (1983: 631).
The CWP also had strong links to the Labour Party (MLP) as it was founded “in 1961 by a
former secretary-general of the Malta Labour Party” (McHale & Skowronski 1983: 630).
Thus, it seems reasonable also to treat these parties as splinter parties where the practice of
combining the vote may be valid; depending on which choice is made the volatility score may
vary between 13,40 (=v) and 22,88 (=v_alt). [Mair 1999 agrees with CWP but disagrees with
DNP]
19
v. Poland
For Poland a number of pairs of elections could have been selected for closer study, but the
pair 1997-2001 stands out as a deviating case:
- 1997-2001: What is the relation between the remnants of Solidarity, the AWS (Solidarity
Electoral Action) and PiS (Law and Justice) and PO (Citizens’ Platform)? Is it such a
connection where it is meaningful to treat PiS and PO as splinter parties, or should they rather
be seen as genuinely new parties? According to Rose & Munro (2009: 198) PiS “emerged
from Solidarity Electoral Action” and PO was “created by former members of Solidarity
Electoral Action and of Freedom Union”. My understanding is that most analysts regard PiS
and PO as genuinely new parties; yet, we here have a problem of drawing a proper line
between splinters forming new parties and actors forming a genuinely new party. I have opted
for the splinter version which gives an v=23,91, whereas choosing the genuinely new party
option gives an v_alt=46,10.
vi. Latvia
The pair of elections chosen for Latvia is 2002-2006:
- 2002-2006: The focus is here on how to treat Harmony Centre (SC) and its predecessors;
should SC be seen as a splinter from PCTVL (For Human Rights in a United Latvia) or as a
genuinely new party in 2006? According to Rose & Munro (2009: 162) parties belonging to
the PCTVL alliance “broke away, forming Harmony Centre”, and therefore it seems to be
reasonable to classify SC as a splinter party related to PCTVL, and consequently the principle
of combining the vote for SC and PCTVL may be applied; depending on what choice is made
v varies between 17,83 and 30,86 (=v_alt).
vii. Slovenia
A diverging pair of elections in Slovenia is the 2004-2008 pair:
- 2004-2008: Here two parties are in focus and at issue is how to treat Zares and Lipa and
their predecessors; are they splinter parties or genuinely new parties? Again, following Rose
& Munro (2009: 235) they treat both as splinter parties; Zares is a splinter from LDS whereas
Lipa is a splinter from SNS. Treating them as splinter parties and combining the votes v is
26,45, while treating them as genuinely new parties gives an v_alt = 36,68.
viii. Lithuania
The election pair selected for Lithuania is 2004-2008:
- 2004-2008: At issue here is whether there were splits within the Social Democratic Party
(LSdP) and the Liberal and Centre Union (LbCS) between the elections or birth of genuinely
new parties. Relying on Rose & Munro (2009: 178-79) LSdP in 2004 were joined in an
alliance with New Union – Social Liberals (NS-SL) but NS-SL contested the 2008 election
independently and the Frontas Party which took part in 2008 was a splinter from LSdP; the
Liberal Movement of the Republic of Lithuania (LRLS) was a splinter from LbCS.
Considering these parties as splinter parties and combining the vote for this pair of elections
results in an v=27,64, while treating them as genuinely new parties increases the volatility
score to 36,67 (=v_alt). In my opinion the splinter option is the more appropriate one.
ix. Italy
Calculating volatility scores for Italy is indeed tricky, especially for the post-1992 period, but
here only two pairs of elections will be looked upon:
- 1946-1948: The focus for this pair is on PSDI (Partito Socialista Democratico Italiano). In
the literature the PSDI is considered a splinter party emanating from the Socialist Party (PSI).
Mackie & Rose (1991: 263) write that it was “a centrist breakaway from the Socialist Party
20
led by Giuseppe Saragat.” Being a splinter party, according to my understanding, the principle
of combining votes for the old and the new parties for the pair of election should be applied.
Therefore I find it somewhat surprisingly that Bartolini & Mair (1990: [342]) treat PSDI as a
genuinely new party. My estimate of v=16,51 whereas treating PSDI as a genuinely new party
increases v to 23,59 (=v_alt).
- 1994-1996: The choice of focus for this pair of elections is the Dini List – Italian Renewal
(RI); is RI to be classified as a genuinely new party or does it have any predecessors? Partly
based on Prévost (2011) I have opted for a combination of the Segni Pact and parts of the
Socialist Party as predecessors; admittedly, this is questionable, and maybe the proper thing
would be to classify RI as a genuinely new party. Which choice is made matter, but not that
much, since the estimates varies between v= 9,48 and v_alt = 14,22. [Bolin 2012 disagrees]
x. France
As was recognised already by Pedersen, estimating volatility scores for the French party
system is not easy. Two pairs of elections will be chosen and here my estimates are higher
than the ones reported in the literature.
- 1962-1967: The conservative political scene in France is not easy to track down. Two
alternatives, among many, to face for this pair of elections refers to where to place the CNIP
of 1962 (mr#12); my option has been to keep the CNIP as part of the Modéres and let MRP
(mr#13) align with CD (mr#21) which results in a v of 13,14; the options chosen by Bartolini
& Mair (1990: [334]) was to align CNIP and MRP with CD which gives an v_alt of 6,92.
- 1968-1973: At issue for this pair of elections is, inter alia, how to treat the Reformateurs
(mr#26); should it be considered to be a genuinely new party (my option giving v=23,81) or
considered to be aligned with CD (alternative option giving v_alt = 17,19). Still one may note
that the volatility score estimated by Bartolini & Mair (1990: [334]) is distinctly lower with
10,9.
xi. Estonia
The pair of elections selected for Estonia is another case where my estimation is higher than
what is reported in the literature:
- 1999-2003: It is about how to treat the Coalition Party (rm#3) after the 1999 election; my
interpretation was that we here have a case of a party death, or exit, (Rose & Munro 2009:
133) resulting in a v of 34,68. However, a closer examination suggest that it is reasonable to
consider the Coalition Party as merging with the People’s Union (rm#28); this is how Pettai
(2004: 994) treats the party, and a similar interpretation may be drawn from the overview
available from Erakonnad (http://www.erakonnad.info/valimised.html; accessed in July
2012), which results in the alternative estimate of v_alt = 28,93.
xii. Luxembourg
- 1964-1968: The Popular Independent Movement (MIP; mr#18) entered parliament in 1964
but the major part of the party merged with the Democratic Party (rm#16) prior of the 1968
election; the alternative I have opted for is a merger of these two parties resulting in a v =
4,93; another alternative is to treat MIP as a stable party for this pair of elections giving an
alternative estimate of alt_v = 10,28.
xiii. Greece
For both pairs of elections the treatment of Polan (Political Spring; mr#50) is at issue; Polan
was formed by Samaras in 1993 as a breakaway from New Democracy (ND; mr#36) but
Polan did not participate in the 2000 election although expressing support for ND and joined
ND later in 2004.
21
- 1990-1993: Treating Polan as a splinter party from ND combining their votes in the pair of
elections means a v of 8,80; if not the alternative estimate goes up to alt_v = 13,68.
- 1996-2000: Treating Polan and ND as a forthcoming merger and combining their votes for
the pair of elections gives an v of 5,14; if not the alternative estimate goes up to alt_v = 8,09.
[Bolin 2012 agrees]
xiv. Denmark
In general different estimates of volatility in Denmark goes together; yet, here I single out the
treatment of the Socialist People’s Party (SF; mr#16) at the 1960 election and the treatment of
the Danish People’s Party at the 1998 election.
- 1957-1960: SF was formed in 1959 by the previous CP-leader Aksel Larsen who had been
excluded from DKP (mr#9), and he had the support from DKP members and voters but not
the DKP organisation. When SF entered parliament at the 1960 election the old CP was
ousted; a majority of the MPs elected were earlier members of DKP (Kragh 1976: 119-124,
Logue 1982: 83). To me SF is a case of a splinter party where the combining of votes for SF
and DKP would be needed to compute the volatility score; if so v = 9,18. If classifying SF as
a genuinely new party the alternative estimate alt_v = 12,22. It is interesting to note that
Bartolini & Mair (1990: [329]) treats SF as a genuinely new party. [Hug 2001 agrees; Mair
1999 disagrees]
- 1994-1998: Danish People’s Party (DF) was formed in 1995 when 4 MP:s left the
parliamentary group of Progress Party (FrP; mr#21). When DF was elected into parliament at
the 1998 election a majority of the MP:s were previous members of FrP (Pedersen 2006). For
me, this is another case of a splinter party where it is relevant to combine the two parties’
votes at this pair of elections; if so the estimate for v = 8,21; if DF is treated as a genuinely
new party the alt_v = 12,22. [Bolleyer 2010, Bolin 2012 agrees]
xv. Hungary
- 2002-2006: At issue is the treatment of the Centre Party (rm#19); the party took part in the
2006 election but the major part of the party joined the Fidesz-KDNP (rm#20) coalition in the
2006 election why I have opted for combing the vote of Fidesz, KDNP and Centrum for this
pair of elections resulting in a v = 4,76; if not the alternative volatility score would increase to
v_alt = 8,33.
xvi. Finland
- 1954-1958: During the end of the 1950’s there were disagreements within the Social
Democratic Party; a few MP’s broke with SSP (mr#1) and were returned in 1958, but a formal
break with SSP and the formation of TPSL (mr#15) waited till 1959 (Helenius 1977:281,
Berglund & Lindström 1978: 41). For this pair of elections the forthcoming TPSL has been
classified as a splinter party from SSP; such a treatment of TPSL gives a v = 4,73 while
another handling increases the volatility score to v_alt = 6,47.
[Hug 2001 agrees]
xvii. Norway
- 1957-1961: Inspired by the events in Denmark a new party SF (mr#14) reached the Storting
at the 1961 election; contrary to the Danish case the Norwegian SF had its origins in the social
democratic Norwegian Labour Party (DNA; mr#4) (Lorenz 1974: 156-157). Mackie & Rose
(1991: 359) classifies SF as a “breakaway group from the Labour Party” and can therefore be
treated as a splinter party resulting in a v = 2,01; disagreement on this point increases the
score to v_alt = 3,58. [Hug 2001 agrees; Mair 1999 and Bolin 2012 disagrees]
22
This review has highlighted the difference between genuinely new parties and new parties
formed through party splits or party mergers. Crucial for applying this distinction for concrete
cases is what is meant by a genuinely new party. I have tended mostly to agree with those
defining this group of parties quite narrow as when Hug suggests that genuinely new parties
are those “which emerge without the help from members of existing parties.” Following this
line of reasoning the number of genuinely new parties will be rather low for the sample
studied in this section of the paper. It may be too restrictive, but it implicates a continuity
within the party system which may be lost if the definition of genuinely new parties is
allowed to be too wide. Still it is obvious that what choice is made for handling the parties
participating in a pair of elections has consequences for the size of the volatility scores to be
estimated.
5. Conclusion
This attempt to map variation in volatility in Europe indicates that for a large number of
elections various estimates tend to go together. Yet, there is not a perfect match, and for some
groups of countries and some pairs of elections there are distinct divergences in estimates
presented in the research literature. Therefore it is wise yet another time to look at how to
arrive at estimates of the aggregate electoral volatility that may be considered to be valid and
reliable. Below a brief summary of what I have achieved in this paper:
a) Electoral data:
In this paper I have made an effort to collect electoral data for all parties participating at an
election thus minimising the number of voters allocated to a category of other parties. It has
not been a successful effort in every instance, but for many countries more or less all parties
are covered in the data sets employed. These data sets therefore are reasonably accurate, but
the increased level of accuracy has in general not had any great impact on the estimations of
the volatility scores; this is illustrated with scores reported for v which are slightly higher than
scores reported for median or mean (as reported in the research literature).
b) Independents and other parties:
The treatment of independents and other parties does matter slightly more; should
independents and other parties be treated as continuous parties or as genuinely new parties? In
the literature the continuous option is often chosen, but when it is possible to break down the
category of other parties it is often, not always, the case that we here face new parties which
many times properly would be classified as genuinely new parties. The sheer number of these
small parties appearing at elections in countries like Spain and Belgium makes it extremely
difficult to classify these parties properly, and to do so one need to have access to country
expertise.
c) Measuring volatility:
Three different measures of volatility scores have been employed and compared (see Table 1).
My impression from this comparison is that Pedersen’s index of aggregate volatility (v)
comes out quite well, and I see no need to abandon this measure for other alternative
measures; other measures like lsq and 1-cos are appropriate as complements but the vmeasure can still be employed as the core measure of aggregate electoral volatility.
23
d) Operationalizations of volatility:
The volatility score has been proposed to be broken down in various aspects; Bartolini & Mair
invented the distinction between block and within-block volatility; Powell & Tucker has
proposed breaking down with respect to volatility originating from new parties (type A) and
originating from stable parties (type B). My sense is that this is an interesting invention which
may elucidate different kinds of aggregate electoral volatility, thus enriching the description
of volatility, but this distinction is not a solution for achieving more reliable estimations of
volatility; crucial for estimating type A volatility is defining what genuinely new parties
stands for.
e) Rules for classifying parties participating in a pair of elections:
Again it is my sense, that crucial for achieving more reliable, and thereby also converging,
estimates is to reach agreement about how to classify parties contesting a pair of elections. I
have in the previous section highlighted the definition of genuinely new parties to be able to
distinguish them for new parties originating from splits or mergers of existing parties. At issue
is in particular whether a narrow or a broad definition of genuinely new parties should be
applied. Reaching more of an agreement on this issue would improve agreements among
various estimates of aggregate electoral volatility presented in the research literature.
f) Solutions proposed:
I can imagine that the following proposals may be conducive for arriving at more accurate
estimates of aggregate electoral volatility:
i. Common analytical framework: One way to arrive at a common framework for classifying
political parties attending a pair of elections would be the establishment of a research project
combining a team of project leaders with in-house expertise at their disposal and another team
of country experts familiar with the party system of their own country. What I have in mind is
the kind of project headed by T. Bergman, W.C. Müller and K. Strøm studying government
coalitions in Western Europe (Müller & Strøm 2000, Strøm, Müller & Bergman 2003, and
Strøm, Müller & Bergman 2008); another venue could be the approach taken in the
Handbooks edited by Sten Berglund and his colleagues (Berglund et al. 1998, 2004 and
forthcoming 2013).
ii. Examine critical cases: There is a need to assign time to a close and detailed examination
of critical cases of elections where there are divergent estimations, to be able to arrive at a
common analytical framework; this could be done within a research project, but not
necessarily so.
iii. Country expertise: There is a need of country expertise to be able to establish party family
trees which makes it possible to trace splits and mergers, births of new parties and deaths of
old parties to facilitate decisions about whether we are facing a genuinely new party or not;
one example of such an exemplary mapping of party families is the one produced by DeeganKrause (2011) for the Slovakian party system.
g) Comment on Mogens Pedersen’s achievement
I started this paper with Mogens Pedersen and therefore it may be appropriate to end it with a
few brief comments on his contribution to the study of aggregate electoral volatility. Tracing
the evolution of the Nordic party systems as well as many of the Western European ones has
not always been easy, but doing so one is seldom faced with major difficulties. Pedersen
himself was, however, clearly aware of the difficulties associated with tracing splits and
24
mergers among the French and Italian party systems; the difficulties associated with the
Italian party system has increased with the new political system of the 1990s. At the time of
writing his articles in the end of the 1970s the party systems of Central and Eastern Europe
was not yet the object of empirical studies, so he was not aware of the problems arising with
the study of these new party systems. It would of course still be possible to address this
question to Mogens Pedersen directly, but I see no reason to bother him now, about whether
he with the hindsight of today would have been eager to engage in the study of aggregate
electoral volatility in Europe. I can here only speculate, but I would expect him to have been
truly hesitant for such an endeavour. There are, after all, huge differences between the
relatively stable party systems of the Nordic countries when comparing them with the still
relatively unstable party systems of Central and Eastern Europe.
25
References:
Barnea, Shlomit & Gideon Rahat (2010) ‘Out with old, in with the “new”’: What constitutes a
new party? Party Politics, 17: 303-320.
Bartolini, Stefano & Peter Mair, Identity, Competition, and Electoral Availability: The
Stabilisation of European Electorates 1885-1985, Cambridge: Cambridge University Press.
Berglund, Sten & Ulf Lindström (1978) The Scandinavia arty System(s): A Comparative
Study, Lund: Studentlitteratur.
Berglund, Sten, Joakim Ekman & Frank H. Aarebrot (2004) (eds.) The Handbook of Political
Change in Eastern Europe, Second Edition, Cheltenham: Edward Elgar.
Berglund, Sten, Joakim Ekman & Frank H. Aarebrot (2013) (eds.) The Handbook of Political
Change in Eastern Europe, Third Edition, Cheltenham: Edward Elgar.
Berglund, Sten, Tomas Hellén & Frank H. Aarebrot (1998) (eds.) The Handbook of Political
Change in Eastern Europe, Cheltenham: Edward Elgar.
Birch, Sarah (2001) Electoral systems and party system stability in post-communist Europe,
Paper prepared for the APSA meeting in San Francisco.
Birch, Sarah (2003) Electoral Systems and Political Transformation in Post-Communist
Europe, Basingstoke: Palgrave Macmillan.
Bischoff, Carina S. (2012) Electorally unstable by supply or demand? – an examination of the
causes of electoral volatility in advanced industrial democracies, Public Choice, forthcoming;
http://www.springerlink.com/content/d2058l21j3436971/fulltext.pdf
Bochsler, Daniel: Database on sub-national results of national elections in post-communist
democracies (first chamber of national parliament); data available from:
http://www.bochsler.eu/ceedata/,
Bolin, Niklas (2012a) Målsättning riksdagen: Ett aktörsperspektiv på nya parties inträde I det
nationella parlamentet, Umeå University: Department of political science; http://umu.divaportal.org/smash/record.jsf?pid=diva2:525488
Bolin, Niklas (2012b) The New Party Entrance Dataset; appendix to Bolin (2012a);
http://www.erdda.se/index.php/projects/erd/nped
Bolleyer, Nicole (2010) Survival of the Fittest? The Persistence of New Parties in Western
Europe, Paper prepared for Etmaal Conference, Leuven, May 27th-28th 2010;
http://soc.kuleuven.be/web/files/11/72/W07-02.pdf
Bureš, Jan & Petr Just (2010) The Origins of the Czech and Slovak Pluralist Party Systems,
Politics in Central Europe, 6, 1: 41-82.
CIAS (The Center for Integrated Area Studies, Kyoto University): The Database on the
election and political system in East Central European Countries; data available from:
http://www.seinan-gu.ac.jp/~sengoku/database/
Coppedge, Michael (1998) The Dynamic Diversity of Latin American Party Systems, Party
Politics, 4: 547-568.
Crewe, Ivor & David Denver (1985) (eds.) Electoral Change in Western Democracies:
Patterns and Sources of Electoral Volatility, London: Croom Helm.
Crewe, Ivor (1985) Introduction: Electoral Change in Western Democracies: A Framework
for Analysis, in Ivor Crewe & David Denver (eds.) Electoral Change in Western
Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm, pp. 1-22.
Damsgaard, Erik (1974) Stability and Change in the Danish Party System over Half a
Century, Scandinavian Political Studies, 9: 103-125.
Dassonneville, Ruth & March Hooghe (2011) Mapping Electoral Volatility in Europe: An
analysis of trends in electoral volatility in European democracies since 1945, Paper presented
at the 1st European Conference on Comparative Electoral Research, Sofia (Bulgaria), 1-3
December 2011.
26
Deegan-Krause, Kevin (2011) Family tree of political parties in Slovakia, 1990-2011;
http://www.pozorblog.com/wp-content/uploads/2011/11/Slovak-Party-Tree-20111.pdf
Duncan, Otis Dudley & Beverly Duncan (1955) A Methodical Analysis of Segregation
Indexes, American Sociological Review, 20: 210-217.
Epperly, Brad (2011) Institutions and Legacies: Electoral Volatility in the Postcommunist
World, Comparative Political Studies, 44: 829-853.
Erakonnad.info; 20 aastat erakondasid Eestis 1988-2008; http://www.erakonnad.info/
Ersson, Svante & Jan-Erik Lane (1998) Electoral instability and party system change in
Western Europe, in Paul Pennings and Jan-Erik Lane (eds) Comparing Party System Change,
London: Routledge, pp. 23-39.
Gherghina, Sergiu (2012) Explaining electoral volatility in Central and Eastern Europe: A
Party Organizational Approach, Department Political Science, Faculty of Social and
Behavioural Sciences, Leiden University; PhD Thesis;
https://openaccess.leidenuniv.nl/handle/1887/18567
Grunberg, Gerard (1985) France, in Ivor Crewe & David Denver (eds.) Electoral Change in
Western Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm,
pp. 202-229.
Hardarson, Ólafur Th. & Gunnar Helgi Kristinsson (2000) Iceland, European Journal of
Political Research, 38: 408-419.
Helenius, Ralf (1977) The Finnish Social Democratic Party, in William E. Paterson &
Alastair H. Thomas (eds.) Social Democratic Parties in Western Europe, London: Croom
Helms, pp. 272-291.
Hug, Simon (2001a) Altering Party Systems: Strategic Behavior and the Emergence of New
Political Parties in Western Democracies, Ann Arbor, MI: The University of Michigan Press.
Hug, Simon (2001b) Data appendix to Hug (2001a);
http://www.unige.ch/ses/spo/static/simonhug/newparty/parties.html
Janda, Kenneth & Jinyoung Kwak (2009) Competition and Volatility in Parliamentary Party
Systems for 212 Polities, Prepared for Delivery at the 2009 Annual Meeting of the Midwest
Political Science Association, Chicago, Illinois;
http://www.janda.org/bio/parties/papers/Janda%20(2009a).pdf
Jarlov, Carsten & Ole P. Kristensen (1978) Electoral Mobility and Social Change in
Denmark, Scandinavian Political Studies,1: 61-78.
Jehlicka, Petr, Tomás Kostelecky & Ludek Sykora (1993) Czechoslovak parliamentary
elections 1990: old patterns, new trends and lots of surprises, in John O’Loughlin & Herman
van der Wusten (eds.) The New Political Geography of Eastern Europe, London: Belhaven
Press, pp. 235-254.
Jones, Benjamin Rowe (2012) Electoral Volatility in Old and New Democracies: Comparing
Causes of Party System Institutionalisation, Central European University, Department of
Political Science, Masters of Arts Thesis; http://www.etd.ceu.hu/2012/jones_benjamin.pdf
Klíma, Michal (1998) Consolidation and Stabilization of the Party System in the Czech
Republic, Political Studies, 66: 492-510.
Koppel, Moshe & Abraham Diskin (2009) Measuring disproportionality, volatility and
malapportionment: axiomatization and solutions, Social Choice and Welfare, 33: 281-286.
Kragh, Jens (1976) Opbrud på venstrefløjen 1956-1960: Striden i DKP og SFs dannelse,
København: SP Forlag.
Kristjánsson, Svannur & Indridi H. Indridason (2011) Iceland: Dramatic Shifts, in Torbjörn
Bergman & Kaare Strøm (eds.) The Madisonian Turn: Political Parties and Parliamentary
Democracy in Nordic Europe, Ann Arbor, MI: The University of Michigan Press, pp. 158199.
27
Krouwel, André & Paul Lucardie (2008) Waiting in the Wings: New Parties in the
Netherlands, Acta Politica, 43: 278-307.
Lane, Jan-Erik with Svante Ersson (2007) Party System Instability in Europe: Persistent
Differences in Volatility between West and East? Democratization, 14, 1: 92-110.
Lijphart, Arend (2005) The new institutionalism and constitutional design, Perspectives on
Politics, 3: 322-23.
Logue, John (1982) Socialism and Abundance: Radical Socialism in the Welfare State: A
Study of the Danish Socialist People’s Party, København: Akademisk Forlag.
Loosemore, J. & V. Hanby (1971) The Theoretical Limits of Maximum Distortion: Some
Analytical Expressions for Electoral Systems, British Journal of Political Science, 1: 467477.
Lorenz, Einhart (1974) Arbeiderbevegelsens historie II: 1930-1973, Oslo: Pax.
Mackie, Thomas T. & Richard Rose (1974) The International Almanac of Electoral History,
New York, NY: The Free Press.
Mackie, Thomas T. & Richard Rose (1982) The International Almanac of Electoral History;
Second Edition, London: The Macmillan Press.
Mackie, Thomas T, & Richard Rose (1991) The International Almanac of Electoral History;
fully revised third edition, Basingstoke: Macmillan.
Mackie, Tom & Richard Rose (1997) A decade of election results: updating the international
almanac, University of Strathclyde: Centre for the Study of Public Policy.
Mainwaring, Scott & Edurne Zoco (2007) Political Sequence and Stabilization of Interparty
Competition, Party Politics, 13: 155-178.
Mainwaring, Scott, Carlos Gervasoni & Annabella España-Nájera (2010) The Vote Share of
New and Young Parties, University of Notre Dame, Kellogg Institute (Working Paper # 368);
http://kellogg.nd.edu/publications/workingpapers/WPS/368.pdf
Mair, Peter (1999) New Political Parties in Established Party Systems: How Successful are
They?, in Erik Bekkel, Kurt Klaudi Klausen & Poul Erik Mouritzen (eds), Elites, Parties and
Democracy: Festschrift for Professor Mogens N. Pedersen, Odense, Odense University
Press, pp. 207-224.
Mansfeldová, Zdenka (2004) The Czech Republic, in Berglund, Sten et al. (eds.) The
Handbook of Political Change in Eastern Europe, Second Edition, Cheltenham: Edward
Elgar, pp. 223-253.
McHale, Vincent E. & Sharon Skowronski (1983) (eds.) Political Parties of Europe,
Westport, CO: Greenwood.
Müller, Wolfgang C. & Kaare Strøm (2000) (eds.) Coalition Governments in Western Europe,
Oxford: Oxford University Press.
Mustillo, Thomas J. (2009) Modelling New Party Performance: A Conceptual and
Methodological Approach for Volatile Party Systems, Political Analysis, 17: 311-332.
Nohlen, Dieter & Philip Stöver (eds) (2010) Elections in Europe: a data handbook, BadenBaden: Nomos.
Nordsieck, Wolfram: Parties & Elections: The database about parliamentary elections and
political parties in Europe; data available from: http://www.parties-and-elections.eu/
Passin, Christian (2012) Wahlen Slowakei 2012: Analyse der Parlamentswahlen vom 10.
März 2012, Wien: Politische Akademie der ÖVP;
http://modernpolitics.at/uploads/media/slowakei_wahl_2012_update.pdf
Pedersen, Karina (2006) Driving a Populist Party: The Danish People’s Party; University of
Copenhagen, Department of Political Science; Arbejdspapir, 2006/6;
http://curis.ku.dk/ws/files/15288675/http___polsci.ku.dk_bibliotek_publikationer_2006_ap_2
006_06.pdf
28
Pedersen, Mogens (1979) The Dynamics of European Party Systems: Changing Patterns of
Electoral Volatility, European Journal of Political Research, 7: 1-26.
Pedersen, Mogens (1980) On Measuring Party System Change: A Methodological Critique
and a Suggestion, Comparative Political Studies, 12: 387-403.
Pedersen, Mogens (1983) Changing Patterns of Electoral Volatility in European Party
Systems, 1948-1977: explorations and explanations, in Hans Daalder and Peter Mair (eds)
Western European Party Systems: Continuity and Change, Beverly Hills, CA: Sage, pp. 2966.
Petersson, Olof (1974) The 1973 General Election in Sweden, Scandinavian Political Studies,
9: 219-228.
Petersson, Olof (1978) The 1976 Election: New Trends in the Swedish Electorate,
Scandinavian Political Studies, 1: 109-121.
Pettai, Vello (2004) Estonia, European Journal of Political Research, 43: 993-999.
Popescu, Marina & Martin Hannavy: Political Transformation and the Electoral Process in
Post-Communist Europe; data available from: http://www.essex.ac.uk/elections/
Powell, Eleanor Neff & Joshua A. Tucker (2012) Revisiting Electoral Volatility in PostCommunist Countries: New Data, New Results, and New Approaches; forthcoming at the
British Journal of Political Science;
http://www.eleanorneffpowell.com/uploads/8/3/9/3/8393347/powelltucker_2012.pdf
Prévost, Damien (2011) Une petite histoire des partis politiques italiens, La Clé des Langues,
Lyon: ENS LYON/DGESCO, ISSN 2107-7029. Mis à jour le 10 juillet 2012;
Consulté le 3 août 2012; http://cle.ens-lyon.fr/italien/une-petite-histoire-des-partis-politiquesitaliens-116056
Robbins, Joseph W. & Lance Y. Hunter (2012) Impact of electoral volatility and party
replacement on voter turnout levels, Party Politics, forthcoming.
Roberts, Kenneth M. and Erik Wibbels (1999) Party Systems and Electoral Volatility in Latin
America: A Test of Economic, Institutional, and Structural Explanations, American Political
Science Review, 93: 575-590,
Rose, Richard & Neil Munro (2003) Elections and Parties in new European democracies,
Washington, DC: CQ Press.
Rose, Richard & Neil Munro (2009) Parties and elections in new European democracies,
Colchester: ECPR Press.
Rose, Richard, Neil Munro & Tom Mackie (1998) Elections in Central and Eastern Europe
since 1990, University of Strathclyde: Centre for the Study of Public Policy.
Shair-Rosenfield, Sarah (2008) Assessing the causes and effects of electoral volatility : party
system fragmentation, time, and executive turnover; A thesis submitted to the faculty of the
University of North Carolina at Chapel Hill in partial fulfillment of the requirements of the
degree of Master of Arts in the Department of Political Science, Chapel Hill;
https://cdr.lib.unc.edu/record?id=uuid%3a249ecd6b-2a73-457b-9b5b-c9fcdd687ae4
Sikk, Allan (2001) Stability of Post-Communist Party Systems, University of Tartu,
Department of Political Science (MA Thesis); http://lepo.it.da.ut.ee/~allans/ma/ma.doc
Sikk, Allan (2005a) How Unstable? Volatility and the Genuinely New Parties in Eastern
Europe, European Journal of Political Research, 44: 391-412.
Sikk, Allan (2005b) Genuinely New Parties in Eastern Europe; Appendix to Sikk, Allan
(2005a); http://www.homepages.ucl.ac.uk/~tjmsasi/genp.pdf
Sikk, Allan (2006) Highways to power: New party success in three young democracies, Tartu:
Tartu University Press; http://www.homepages.ucl.ac.uk/~tjmsasi/
Stískala, Jozef (2012) Party System of Slovak Republic and its Stability after 2010 and 2012
Elections in Comparative Perspective, Slovak Journal of Political Sciences, 12: 224-251;
http://revue.kpol.ff.ucm.sk/rocnik-12/cislo-3/stiskala_studia.pdf
29
Strøm, Kaare, Wolfgan C. Müller & Torbjörn Bergman (2003) (eds.) Delegation and
Accountability in Parliamentary Democracies, Oxford: Oxford University Press.
Strøm, Kaare, Wolfgan C. Müller & Torbjörn Bergman (2008) (eds.) Cabinets and Coalition
Bargaining: The Democratic Life Cycle in Western Europe, Oxford: Oxford University Press.
Taagepera, Rein & Bernard Grofman (2003) Mapping the Indices of Seats-Votes
Disproportionality and Inter-Election Volatility, Party Politics, 9: 659-677.
Tavits, Margit (2005) The Development of Stable Party Support: Electoral Dynamics in PostCommunist Europe, American Journal of Political Science, 49: 283-298.
Tavits, Margit (2008a) Party Systems in the Making: The Emergence and Success of New
Parties in New Democracies, British Journal of Political Science, 38: 113-133.
Tavits, Margit (2008b) Supplemental materials for Tavits (2008a);
http://tavits.wustl.edu/SupplementalMaterial.pdf
Tóka, Gábor (1997) Political Parties and Democratic Consolidation in East Central Europe,
University of Strathclyde: Centre for the Study of Public Policy (Studies in Public Policy;
279).
Valen, Henry & Stein Rokkan (1974) Norway: The Election to the Storting in September
1973, Scandinavian Political Studies, 9: 103-125.205-218.
Výtisk, Radek (2007) Otázka “užitečnosti” vybraného způsobu kvantitativního hodnocení
výstupu voleb, Brně, Masarykova Univerzita; http://is.muni.cz/th/65506/fss_m/
Wightman, Gordon (1993) The Czechoslovak Parliamentary Elections of 1992, Electoral
Studies, 12, 1: 83-86.
30
Appendix A: Election results data
Listed below are the main sources used to collect electoral data for the 31 European countries.
General sources:
Books & periodicals:
European Journal of Political Research (1992 - ) Political data yearbook.
Mackie, Thomas T, & Richard Rose (1991) The international almanac of electoral history;
fully revised third edition, Basingstoke: Macmillan.
Mackie, Tom & Richard Rose (1997) A decade of election results: updating the international
almanac, University of Strathclyde: Centre for the Study of Public Policy.
Nohlen, Dieter & Philip Stöver (eds) (2010) Elections in Europe: a data handbook, BadenBaden: Nomos.
Rose, Richard & Neil Munro (2009) Parties and elections in new European democracies,
Colchester: ECPR Press.
Websites:
Bochsler, Daniel: Database on sub-national results of national elections in post-communist
democracies (first chamber of national parliament); data available from:
http://www.bochsler.eu/ceedata/,
CIAS (The Center for Integrated Area Studies, Kyoto University): The Database on the
election and political system in East Central European Countries; data available from:
http://www.seinan-gu.ac.jp/~sengoku/database/
Nordsieck, Wolfram: Parties & Elections: The database about parliamentary elections and
political parties in Europe; data available from: http://www.parties-and-elections.eu/
Popescu, Marina & Martin Hannavy: Political Transformation and the Electoral Process in
Post-Communist Europe; data available from: http://www.essex.ac.uk/elections/
Austria:
Bundesministerium für Inneres: Wahlen: Nationalratswahl, Historischer Rückblick; data
available from: http://www.bmi.gv.at/cms/BMI_wahlen/nationalrat/NRW_History.aspx
Belgium:
DB Election: Le site officiel des résultats électoraux belges; data available from:
http://www.ibzdgip.fgov.be/result/fr/main.html
Elections federales 2003; data available from:
http://elections2003.belgium.be/electionshome/fr/result/chamber/table_top.html
Elections federales 2007; data available from:
http://elections2007.belgium.be/fr/cha/results/results_tab_etop.html
Elections federales 2010; data available from:
http://elections2010.belgium.be/fr/cha/results/results_tab_CKR00000.html
Bulgaria:
Electoral Commission of Bulgaria: Bulgarian election 2009; data available from:
http://pi2009.cik.bg/results/proportional/rik_00.html
Electoral Commission of Bulgaria: Bulgarian election 2005: data available from:
http://pi2005.cik.bg/results/
Croatia:
Arhiva Izbora: Election results; data available from: http://www.izbori.hr/arhiva/arhiva.html
31
Cyprus:
Ministry of the Interior: Parliamentary elections 2011; data were available from:
http://live.elections.moi.gov.cy (now dead link)
Parliament of Cyprus: Parliamentary Elections; data available from:
http://www.parliament.cy/parliamentgr/002_10.htm
Czech Republic:
Czech Statistical Office: Results of elections; data available from:
http://www.volby.cz/index_en.htm
Denmark:
Danmarks statistik: Valgoversigt; data available from: http://www.dst.dk/valg/index.htm
Folketinget: Folketingsvalgene 1953-2011; available from:
http://www.ft.dk/Folketinget/Oplysningen/Valg/~/media/Pdf_materiale/Pdf_download_direkt
e/Folketingets%20Oplysning/Folketingsvalgene%201953-2011.pdf.ashx
Ministeriet for Sundhed og Forebyggelse: Resultater fra tidligere folketingsvalg; data
available from: http://valg.im.dk/Valg/Folketingsvalg/resultat-seneste-valg.aspx
Estonia:
Vabariigi Valimiskomisjoni: Riigikogu valimised; data available from:
http://www.vvk.ee/arhiiv/riigikogu-valimised/
Elections in Estonia 1992-2011; available from: http://www.vvk.ee/past-elections/electionsin-estonia-1992-2011/
Finland:
Official Statistics of Finland (OSF): Parliamentary elections [e-publication]. ISSN=17996279. Helsinki: Statistics Finland; available from: http://www.stat.fi/til/evaa/index_en.html
Statistics Finland: Tables on the subject area of: Parliamentary elections - Time series; data
available from: http://pxweb2.stat.fi/database/StatFin/vaa/evaa/evaa_as/evaa_as_en.asp
France:
Ministère de l'Intérieur, de l'Outre-mer, des Collectivités territoriales et de l'Immigration:
Législatives 2012 - Résultats France entière, métropole et outre-mer - Tour 1; data available
from: http://www.data.gouv.fr/var/download/027091f0c429e219ca91adc93358731a.xls
Ministère de l'Intérieur: Résultats électoraux en France; data available from:
http://www.interieur.gouv.fr/sections/a_votre_service/elections/resultats/
de Boissieu, Laurent: France Politique: Élections législatives; data available from:
http://www.france-politique.fr/elections-legislatives.htm
Greece:
Ministery of Interior: Election results; data available from:
http://www.ypes.gr/en/Elections/NationalElections/Results/
Germany:
Der Bundeswahhleiter: Bundestagswahlen; data available from:
http://www.bundeswahlleiter.de/de/bundestagswahlen/fruehere_bundestagswahlen/
32
Hungary:
National Election Office: Election results; data available from:
http://www.valasztas.hu/dyn/pv10/outroot/vdin1/en/l22x.htm
Vokscentrum: Parliamentary election results; data available from:
http://www.vokscentrum.hu/index.php?jny=hun
Iceland:
Statistics Iceland: Election results; data available from:
http://www.statice.is/Statistics/Elections/General-elections
Wikipedia: Alþingiskosningar; http://is.wikipedia.org/wiki/Al%C3%BEingiskosningar
Ireland:
Department of the Environment, Community and Local Government: Voting; data available
from: http://www.environ.ie/en/Publications/LocalGovernment/Voting/
Italy:
Ministerio dell’Interno: ARCHIVIO STORICO DELLE ELEZIONI - Consultazione dati; data
avilable from: http://elezionistorico.interno.it/
Latvia:
Centrālā vēlēšanu komisija: Saeima elections; data available from:
http://web.cvk.lv/pub/public/28757.html
Statistics Latvia: Statistics Database: Politics and Religion; data available from:
http://data.csb.gov.lv/DATABASEEN/visp/Annual%20statistical%20data/24.%20Politics%2
0and%20religion/24.%20Politics%20and%20religion.asp
Lithuania:
The Central Electoral Commission of the Republic of Lithuania: Previous elections; data
available from: http://www.vrk.lt/en/pirmas-puslapis/previous-elections/sorted-by-thedate.html
Luxembourg:
Ministère de l'Intérieur: Le site officiel des élections au Grand-Duché de Luxembourg ; data
available from: http://www.elections.public.lu/fr/index.html
Malta :
Lane, John C.: Elections in Malta: Parliamentary Election Results, 1921 – 2008; data
available from: http://www.maltadata.com/
Netherlands:
Kiesraad: Databank verkiezingsuitslagen; data available from:
http://www.verkiezingsuitslagen.nl/Default.aspx
Nederlandse verkiezingsuitslagen 1918-nu; http://www.nlverkiezingen.com/
Parlamente & Politiek: Verkiezingen Tweede Kamer 1918-2010; data available from:
http://www.parlement.com/9291000/modulesf/g87hwdf0
Norway:
Statistisk sentralbyrå: Statistikkbanken: Stortingsvalg; data available from:
http://statbank.ssb.no/statistikkbanken/Default_FR.asp?Productid=00.01&PXSid=0&nvl=true
&PLanguage=0&tilside=selecttable/MenuSelP.asp&SubjectCode=00
33
Wikipedia: Stortingsval; http://nn.wikipedia.org/wiki/Stortingsval
Poland:
Panstwowa Komisja Wyborcza: Elections results; data available from:
http://pkw.gov.pl/wybory-sejm-senat/
Portugal:
Comissão Nacional de Eleições: Resultados eleitoras; data available from:
http://eleicoes.cne.pt/raster/index.cfm?dia=05&mes=06&ano=2011&eleicao=ar
Romania:
Permanent Electoral Authority: Polls results; data available from:
http://www.roaep.ro/en/section.php?id=66
Spain:
Ministerio del Interior Consulta de resultados electorales; data available from:
http://www.infoelectoral.mir.es/min/home.html
Slovakia:
Statisticky urad Slovenskej republiky: Parlamentné voľby; data available from:
http://portal.statistics.sk/showdoc.do?docid=4490
Slovenia:
Ministry of Public Administration: Decision-making by citizens; data available from:
http://volitve.gov.si/en/index.html
Sweden:
Valmyndigheten: Tidigare val; data available from: http://www.val.se/tidigare_val/
Statistics Sweden: Demokrati: Allmänna val; data available from:
http://www.scb.se/Pages/SubjectArea____12261.aspx
Switzerland:
Bundesamt für Statistik: Wahlen: Nationalrat; data available from: http://www.politikstat.ch/nrw2011KT_de.html
Schweizerische Bundeskanzlei: Nationalratswahlen; data available from:
http://www.bk.admin.ch/themen/pore/nrw/index.html?lang=de
Wikipedia: Wahl in der Schweiz;
http://de.wikipedia.org/wiki/Kategorie:Wahl_in_der_Schweiz
United Kingdom:
Boothroyd, David: United Kingdom Election Results; data available from:
http://www.election.demon.co.uk/
House of Commons: Elections results: UK; data available from:
http://www.parliament.uk/topics/Election-results-UK.htm
The Electoral Commission: General elections; data available from:
http://www.electoralcommission.org.uk/elections/results/general_elections
34
Appendix B: Aggregate volatility estimates in the research literature
Studies/data sets with a comparative focus covering more than one country:
Abuş, Murat (2003) Democratization in the Balkans, 1990-2002, Alternatives: Turkish
Journal of International Relations, 2, 3&4: 86-105.
Bakke, Elisabeth & Nick Sitter (2005) To Stay the Course or Cross the Floor: Members of
Parliament, Parties and Party Systems Change in Central Europe, Oslo: The Centre for
European and Asian Studies at Norwegian School of Management.
Bartolini, Stefano & Peter Mair (1990) Identity, competition, and electoral availability: The
stabilisation of European Electorates 1885-1985, Cambridge: Cambridge University Press.
Bågenholm, Andreas & Andreas Johansson Heinö (2009) Still not stable after all these years:
The logic of path dependence in party system stability in Central and Eastern Europe, Paper
prepared for the ISA Annual Convention, New York, February 15-18, 2009.
Catón, Matthias (2009) Wahlsysteme und Parteiensysteme im Kontext: Vergleichende Analyse
der Wirkung von Wahlsystemen unter verschidenen Kontextbedingungen, Heidelberg
http://www.caton.de/wahlsystem-kontext
Dassonneville, Ruth & March Hooghe (2011) Mapping Electoral Volatility in Europe: An
analysis of trends in electoral volatility in European democracies since 1945, Paper presented
at the 1st European Conference on Comparative Electoral Research, Sofia (Bulgaria), 1-3
December 2011.
Enyedi, Zsolt & Fernando Casal Bértoa (2010) Pártverseny – Mintázatok és Blokk-Politika
Kelet-Közép-Európában (1990-2009), Politikadományi Szemle, 19, 1: 7-30;
http://www.poltudszemle.hu/szamok/2010_1szam/2010_1_enyedi.pdf
Gunther, Richard & José R. Monteor (2001) The Anchors of Partisanship: A Comparative
Analysis of Voting Behavior in Four Southern European Democcracies, in P. Nikiforos
Diamandouros & Ricahrd Gunther (eds.) Parties, Politics and Democracy in the New
Southern Europe, Baltimore, MD, The Johns Hopkins University Press, pp. 83-152.
Gunther, Richard (2005) Parties and Electoral Behavior in Southern Europe, Comparative
Politics, 37: 253-275.
Hamann, Kerstin & John Kelly (2006) Voters, Parties and Social Pacts in Western Europe,
Paper Prepared for delivery at the 15th International Conference of the Council for European
Studies, Chicago, IL, March 29-April 2, 2006.
Hansen, Kasper M. (2009) Voters and the Elections Campaigns: Late deciders who are they?
Paper presented at the Two-Day seminar on Voting Behaviour, Centre for Voting and Parties,
University of Copenhagen, 24-25 August, 2009.
Hirata, Takeshi (2000) The emergence of the party system and the electoral volatility in
Central Europe, Central European Political Science Review,1: 272-291.
Horwath, Attila (2008) Stabilization and Concentration in the Political Party Systems of the
Visegrád Countries, in Attila Ágh & Judit Kis-Varga (eds) New perspective for the EU team
presidencies: New members, new candidates and new neighbours, Budapest: ‘Together for
Europe’ Research Centre, pp. 251-301.
Janda, Kenneth (1994) Restructuring the Party Systems in Central Europe, Paper presented
for the International Symposium, Democratization and Political Reform in Korea, Sponsored
by the Korean Political Science Association, Seoul, Korea, November 19, 1994.
Jäckle, Sebastian: Government Survival Example – Dataset; data available from:
http://www.sebastianjaeckle.de/teaching.html
Knutsen, Oddbjørn: Western Europe: Parties, voters and elections; data available from:
http://folk.uio.no/stvok1/
Kopecky, Petr (2007), Building Party Government: Political Parties in the Czech and Slovak
35
Republics, in Paul Webb and Stephen White (eds.), Party Politics in New Democracies,
Oxford: Oxford University Press, pp. 119-146.
Nordsieck, Wolfram: Parties & Elections: The database about parliamentary elections and
political parties in Europe; data available from: http://www.parties-and-elections.eu/21
O’Dywer, Conor (2006) And the Last Shall be First: Party Systems Institutionalization and
Second-Generation Economic Reform in Postcommunist Europe, Paper submitted for the
2006 Comparative Politics Workshop at Duke University “Post-Communist Political
Economy and Democratic Politics,” April 7-8, 2006.
Saarts, Tõnis (2009) Political parties and party systems in the Baltic States: Part I;
http://www.tlu.ee/opmat/ri/rit6006/partysystem/4_1slaid.pdf
Saarts, Tõnis (2011) Comparative Party System Analysis in Central and Eastern Europe: the
Case of the Baltic States, Studies of Transition States and Societies, 3, 3: 83-104;
http://www.tlu.ee/stss/wp-content/uploads/2011/11/stss_nov_2011_saarts.pdf
Siaroff, Alan (2000) Comparative European Party Systems: An analysis of Parliamentary
Elections since 1945, New York, NY: Garland Publishing.
Sikk, Alan (2001) Stability of Post-Communist Party Systems, University of Tartu,
Department of Political Science (MA Thesis); http://lepo.it.da.ut.ee/~allans/ma/ma.doc
Sikk, Alan (2006) Highways to power: New party success in three young democracies, Tartu:
Tartu University Press; http://www.homepages.ucl.ac.uk/~tjmsasi/
Tóka, Gabor & Andrija Henjak (2007) Party Systems and Voting Behaviour, in Pavel Šaradin
& Eva Baradová (eds.) The Visegrad Countries 15 years After the Transition, Olomouc,
Palacky University Press, pp. 210-244.
Van Kessel, Stijn Theodoor (2011) Supply and demand: identifying populist parties in Europe
and explaining their electoral performance, Doctoral thesis, University of Sussex;
http://sro.sussex.ac.uk/7521/
Studies/data sets with a focus on one country:
Austria:
Duncan, Fraser (2007) The End of the Wende? The 2006 Austrian Parliamentary Election,
Perspectives on European Politics and Society, 8, 1: 13-30.
Haerpfer, Christian (1985) Austria, in Ivor Crewe & David Denver (eds.) Electoral Change in
Western Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm,
pp. 264-286.
Müller, Wolfgang C. (2000) Elections and the Dynamics of the Austrian Party System since
1986, English version of a contribution to: Fritz Plasser, Peter A. Ulram, and Franz Sommer
(eds.): Das österreichische Wahlverhalten Vienna: Signum, 2000;
http://members.chello.at/zap-forschung/download/dynamics.pdf
Müller, Wolfgang C., Fritz Plasser & Peter A. Ulram (2004) Party Responses to the Erosion
of Voter Loyalties in Austria: Weakness as an Advantage and Strength as a Handicap, in Peter
Mair, Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and Electoral Change:
Party Responses to Electoral Markets, London: Sage, pp. 145-178.
Belgium:
Deschouwer, Kris (2002) The Colour People, in Paul Webb, David Farrell & Ian Holliday
(eds.) Political Parties in Advanced Industrial Democracies, Oxford: Oxford University
Press, pp. 151-180.
Deschouwer, Kris (2004) Political Parties and Their Reactions to the Erosion of Voter
Loyalty in Belgium: Caught in a Trap, in Peter Mair, Wolfgang C. Müller & Fritz Plasser
21
I have employed this data set as it stands to calculate the volatility scores.
36
(eds.) Political Parties and Electoral Change: Party Responses to Electoral Markets, London:
Sage, pp. 179-206.
Mughan, Anthony (1985) Belgium, in Ivor Crewe & David Denver (eds.) Electoral Change in
Western Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm,
pp. 319-341.
Croatia:
Taleski, Dane (2010) Elections and parties in post-conflict countries, Paper prepared to be
presented at the 3rd Graduate Conference 20 August – 1 September, 2010 Dublin City
University.
Czech Republic:
Deegan-Krause, Kevin (2010) Czech Election Update: Time for the Bigger Picture;
http://www.pozorblog.com/2010/06/czech-election-update-time-for-the-bigger-picture/
Šedo, Jakub (2011) Výzkum volatility a proměny stranického spektra ve volbách do
Poslanecké sněmovny v roce 2010, Central European Political Studies Review, 13, 2-3: 246259; http://www.cepsr.com/clanek.php?ID=455
Denmark:
Bille, Lars & Karina Pedersen (2004) Electoral Fortunes and Responses of the Social
Democratic Party and the Liberal Party in Denmark: Ups and Downs, in Peter Mair,
Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and Electoral Change: Party
Responses to Electoral Markets, London: Sage, pp. 207-233.
Bille, Lars (1989) Denmark: The Oscillating Party System, West European Politics, 12, 4: 4258.
Borre, Ole (1985) Denmark, in Ivor Crewe & David Denver (eds.) Electoral Change in
Western Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm,
pp. 372-399.
Estonia:
Mikkel, Evald (2006) Patterns of Party Formation in Estonia: Consolidation Unaccomplished,
in. Susanne Jungerstam-Mulders (ed.) Post-Communism EU Member States: Parties and
Party Systems, Aldershot: Ashgate, pp. 23-49.
Finland:
Arter, David (2009) From a Contingent Party System to Party System Convergence? Mapping
Party System Change in Postwar Finland, Scandinavian Political Studies, 32: 221-239.
Pajala, Antti & Matti Wiberg (2009) Vakaata murskavoitoista ja rökäletappioista huolimatta:
Puljonko on paljon vaalivoitossa tai – tappiossa? Politiikka, 51, 2: 128-136.
Sundberg, Jan (1996) Partier och Partisystem i Finland, Helsingfors: Schildts.
France:
Knapp, Andrew (2002) France, in Paul Webb, David Farrell & Ian Holliday (eds.) Political
Parties in Advanced Industrial Democracies, Oxford: Oxford University Press, pp. 107-150.
Knapp, Andrew (2004) Ephemeral Victories? France’s Governing Parties, the Ecologists, and
the Far Right, in Peter Mair, Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and
Electoral Change: Party Responses to Electoral Markets, London: Sage, pp. 49-85.
37
Germany:
Klingemann, Hans-Dieter (1985) West Germany, in Ivor Crewe & David Denver (eds.)
Electoral Change in Western Democracies: Patterns and Sources of Electoral Volatility,
London: Croom Helm, pp. 230-261.
Scarrow, Susan E. (2002) Party Decline in the Parties State? The Changing Environment of
German Politics, in Paul Webb, David Farrell & Ian Holliday (eds.) Political Parties in
Advanced Industrial Democracies, Oxford: Oxford University Press, pp. 77-106.
Scarrow, Susan E. (2004) Embracing Dealignment, Combating Realignment: German Parties
Respond, in Peter Mair, Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and
Electoral Change: Party Responses to Electoral Markets, London: Sage, pp. 86-110.
Hungary:
Enyedi, Zsolt & Gábor Tóka (2007) The Only Game in Town: Party Politics in Hungary, in
Paul Webb & Stephen White (eds.) Party Politics in New Democracies, Oxford: Oxford
University Press, pp. 147-178.
Iceland
Harðarson, Olafur (2009) The parliamentary election in Iceland, April 2009;
http://www.uio.no/studier/emner/sv/statsvitenskap/STV4332B/v10/Hardarson.pdf
Pétursson, Sigtryggur (2005) Electoral instability in Iceland 1931-95: The impact of aggregate
electoral volatility and block volatility on the Icelandic party system, Stjórnmál og
stjórnsýsla, 1: 159-186; http://skemman.is/en/item/view/1946/8728
Ireland:
Mair, Peter & Michael Marsh (2004) Political Parties in Electoral Markets in Postwar Ireland,
in Peter Mair, Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and Electoral
Change: Party Responses to Electoral Markets, London: Sage, pp. 234-263.
Mair, Peter (2011) One of Europe’s Most Volatile Elections;
http://politicalreform.ie/2011/02/28/one-of-europe%E2%80%99s-most-volatile-elections/
Marsh, M.A. (1985) Ireland, in Ivor Crewe & David Denver (eds.) Electoral Change in
Western Democracies: Patterns and Sources of Electoral Volatility, London: Croom Helm,
pp. 173-201.
Murphy, R.J. & David M. Farrell (2002) Party Politics in Ireland, in Paul Webb, David Farrell
& Ian Holliday (eds.) Political Parties in Advanced Industrial Democracies, Oxford: Oxford
University Press, pp. 217-247.
Italy:
Allum, Percy & Renato Mannheimer (1985) Italy, in Ivor Crewe & David Denver (eds.)
Electoral Change in Western Democracies: Patterns and Sources of Electoral Volatility,
London: Croom Helm, pp. 287-318.
Bardi, Luciano (2002) Italian Parties, in Paul Webb, David Farrell & Ian Holliday (eds.)
Political Parties in Advanced Industrial Democracies, Oxford: Oxford University Press, pp.
46-76.
Bardi, Luciano (2004) Party Responses to Electoral Dealignment in Italy, in Peter Mair,
Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and Electoral Change: Party
Responses to Electoral Markets, London: Sage, pp. 111-144.
Bedock, Camille (2009) Quinze ans dans le laboratoire italienUne analyse des interactions
partisanes entre les règles électorales et la structure de la compétition politique, 1993-2008,
Master « Politiques et sociétés en Europe », Spécialité « Sociologie politique », IEP de Paris –
38
2008-2009; http://www.astrid-online.it/rassegna/Rassegna-28/04-122009/Bedock_mmoirefeuilledestyle-IEP_de_Paris_2008.pdf
Bartolini, Stefano (2002) The Political Consequences of the Italian Mixed Electoral System
(1994-2001), Paper prepared for the Conference on ‘Elections and Democracy’, Social
Science Institute (ICS), University of Lisbon, Lisboa, 1-2 February 2002;
http://www.ics.ul.pt/ceapp/english/conferences/fulbright/4SBartolini.pdf
Latvia:
Ijabs, Ivars (2009) Latvia, in KAS Democracy Report 2009, pp. 135-149;
http://www.kas.de/upload/Publikationen/2009/dr2009/dr_latvia.pdf
Lithuania:
Jastramskis, Mažvydas (2010) Party system stability concept and measurement: the situation
of Lithuania in 1990-2010, Parlamento Studijos;
http://www.parlamentostudijos.lt/Nr9/9_politika_1.htm
Novagrockiené, Jūraté (2001) Elections to the Seimas 2000: Party System Evolution or its
Transformation? Lithuanian Political Science Yearbook 2000, pp. 127-139;
http://www.tspmi.vu.lt/lt/mokslas/moksliniai-darbai/leidiniai/lithuanian-political-scienceyearbook-16
Malta:
Lane, John C.: Elections in Malta: Parliamentary Election Results, 1921 – 2008; data
available from: http://www.maltadata.com/22
Netherlands:
Anker, Hans, Cuperus, René & Pim Paulusman (2011) A Systemic Meltdown? Demographic
Change and Progressive Political Strategy in the Netherlands, Washington, DC: Center for
American Progress;
http://www.americanprogress.org/issues/2011/04/progressive_synthesis.html
Deschouwer, Kris (2002) The Colour People, in Paul Webb, David Farrell & Ian Holliday
(eds.) Political Parties in Advanced Industrial Democracies, Oxford: Oxford University
Press, pp. 151-180.
Eijke, C. Van der & B. Niemöller (1985) Netherlands, in Ivor Crewe & David Denver (eds.)
Electoral Change in Western Democracies: Patterns and Sources of Electoral Volatility,
London: Croom Helm, pp. 342-371.
Mair, Peter (2008) Electoral Volatility and the Dutch Party System: A Comparative
Perspective, Acta Politica, 43: 235-253.
Nederlandse verkiezingsuitslagen 1918-nu; http://www.nlverkiezingen.com/23
Norway:
Aardal, Bernt (2006) How to lose a walk-over election? A preliminary analysis of the 2005
Parliamentary Election in Norway, Oslo: Institutt for samfunnsforskning (Report; 2006:006);
http://www.socialresearch.no/Publications/Reports/2006/2006-006
Poland:
Gwiazda, Anna (2009) Poland’s Quasi-Institutionalized Party System: The Importance of
Elites and Institutions, Perspectives on European Politics and Society, 10: 350-376.
22
I have employed this data set as it stands to calculate the volatility scores.
This databank presents data on gains and losses for the parties participating at a pair of elections which I have
used for calculating a volatility score.
23
39
Jasiewicz, Krzysztof (2007) Poland: Party System by Default, in Paul Webb & Stephen White
(eds.) Party Politics in New Democracies, Oxford: Oxford University Press, pp. 85-118.
Markowski, Radoslaw (2008) The 2007 Polish Parliamentary Election: Some Structuring,
Still a Lot of Chaos, West European Politics, 31: 1055-1068.
Szczerbiak, Aleks (2006) Power without Love? Patterns of Party Politics in Post-1989
Poland, in Susanne Jungerstam-Mulders (ed.) Post-Communism EU Member States: Parties
and Party Systems, Aldershot: Ashgate, pp. 91-123.
Portugal:
Freire, André (2005) Eleções de segunda ordenm e ciclos eleitorais no Portugal democrático,
1975-2004, Análise Social, 40: 815-846.
Jalali, Carlos (2007) Partidos e Democracia em Portugal, 1974-2005: da Revolução ao
Bipartidarismo, Lisboa: ICS.
Romania:
Birnir, Jóhanna Kristín (2004) Institutionalizing the Party System, in Henry Carey (ed.).
Romania since 1989: Politics, Economics and Society, Lanham: Lexington books, pp. 139158.
Bochsler, Daniel & Sergiu Gherghina (2008) The shakedown of the urban-rural division in
post-communist Romanian party politics, Paper to be presented at the BASEES Annual
Conference, 29-31 March 2008, Cambridge.
Ciobanu, Ionuţ (2006) Sisemul românesc de partied: de la competiţie spre coluziune, Sfera
Politicii, 14, 123-124: 5-20.
Slovakia:
Výtisk, Radek (2007) Otázka “užitečnosti” vybraného způsobu kvantitativního hodnocení
výstupu voleb, Brně, Masarykova Univerzita; http://is.muni.cz/th/65506/fss_m/
Slovenia:
Fink-Hafner, Danica (2006) Slovenia: Between Bipolarity and Broad-Coalition Building, in
Susanne Jungerstam-Mulders (ed.) Post-Communism EU Member States: Parties and Party
Systems, Aldershot: Ashgate, pp. 203-231.
Spain:
Medina, Lucia & Mariano Torcal (2007) La institucionalización del sistema de partidos
español: El peso de los anclajes de classe, religión e ideología en la competencia PSOE/PP:
1988-2004, Trabajo presentado en el Congreso de la ACEPA del 21-23 de septiembre en
Madrid;
http://www.aecpa.es/uploads/files/congresos/congreso_07/area04/GT12/Lmedina_.pdf
Ocaña, Francisco A. & Pablo Oñate (1999) Índices e indicadores del sistema electoral y del
sistema de partidos: Una propuesta informática para su cálculo, Reis: Revista española de
investigaciones sociológicas, 86: 223-245;
http://dialnet.unirioja.es/servlet/articulo?codigo=760050
Torcal, Mariano & Ignacio Lago (2008) Electoral Coordination Strikes Again: The 2008
General Election in Spain, South European Society & Politics, 13: 363-375.
Sweden:
Oscarsson, Henrik & Sören Holmberg (2011) Åttapartivalet 2010, Stockholm: SCB;
http://www.scb.se/statistik/_publikationer/ME0106_2010A01_BR_ME05BR1101.pdf
40
Switzerland:
Ladner, Andreas (2004) The Political Parties and the Party System, in Ulrich Klöti et. al.
(eds.). Handbook of Swiss Politics. Zürich: Neue Zürcher Zeitung Publishing pp. 197-242.
Nabholz, Ruth (1998). Das Wählerverhalten in der Schweiz: Stabilität oder Wandel? Eine
Trendanalyse von 1971-1995, in: Hanspeter Kriesi, Wolf Linder & Ulrich Klöti (eds.).
Schweizer Wahlen 1995, Bern: Verlag Paul Haupt, pp. 17-40.
United Kingdom:
Denver, David (2006) Elections and votes in Britain, Basingstoke: Palgrave Macmillan.
Webb, Paul (2002) Political Parties in Britain, in Paul Webb, David Farrell & Ian Holliday
(eds.) Political Parties in Advanced Industrial Democracies, Oxford: Oxford University
Press, pp. 16-45.
Webb, Paul (2004) Party Responses to the Changing Electoral Market in Britain, in Peter
Mair, Wolfgang C. Müller & Fritz Plasser (eds.) Political Parties and Electoral Change:
Party Responses to Electoral Markets, London: Sage, pp. 20-48.
41