This project is aimed at collecting electoral data for all European
countries both in a historical and a regional perspective. Data were made
machine readable from the beginning of national elections in the 19th
century until the present in a disaggregated form down to the level of
electoral districts. The data will we published as an electronic data
handbook, comprising full documentation and introductory chapters.
1. The Project
The aim of this brief research note is to describe the Project on Comparative
European Electoral History (1830-1995).
This research programme consists in producing a computerised data handbook
on European electoral history from the beginning of competitive elections
in the course of the nineteenth century to the present. Data has been
collected for the following countries: Austria, Belgium, Denmark, Finland,
France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, the Netherlands,
Norway, Portugal, Spain, Sweden, Switzerland and the United Kingdom.1
The collection of data is therefore exhaustive work throughout both the
spatial and historical dimensions and is intended to go beyond the already
existing case- or period-oriented collections organised according to incompatible
and non-comparable criteria.
The project is divided into two parts corresponding to two distinct
sets of files.
First, the handbook consists in a systematic and standardised collection
of the results in absolute numbers of all general elections to the lower
houses of the different countries. Results were collected at the constituency
level or another relevant regional level according to the basic territorial
units of each country. Moreover, this collection records elections since
the very beginning of the European electoral history, that is as far back
as official and secondary sources allow for. This should improve the quality
of electoral data, rarely available at constituency level and on computerised
support, often limited to given countries and periods, and never presented
systematically for different countries. For the first time all electoral
results will be reunited within a set of files compiled according to standard
and consistent criteria.
A second set of files provides a quantitative and coded history of the
processes of democratisation (suffrage requirements, nature and level
of restrictions of the right to vote, steps of enfranchisement, etc.)
and of the systems of representation (voting procedures, electoral laws,
etc.) for each country. A series of variables corresponding to these aspects
has been defined and transformed into operational and numerical material.
The work done on the topics of democratisation and representation is meant
to transform the existing information into an operational, computerised,
and thus analysable form.2
The handbook which documents the data provides useful information on
a certain amount of topics for each country and for the entire 1830-1995
period:
1. the sources of the collection, both official and secondary;
2. the names of the constituencies, their changes through time, boundary
changes, the merging of several constituencies together as well as the
splittings of constituencies and patterns of gerrymandering;
3. the dates of all general elections and names of the chambers or houses;
4. the names of all political parties, changes in their names, the merging
of parties, the breakaways, as well as the dates of these modifications;
5. the names, addresses, phone and fax numbers and e-mail addresses
of all European and North American archives, centres, ministries and other
official offices whose collaboration was indispensable for the collection
of the data.
For each country the basic elements of the political and electoral history
are outlined in relation to the relevant changes in the type of electoral
system, levels of enfranchisement, national boundaries, organisation of
political forces, and so forth. This part of the work also provides a
description of the institutional features relevant for the understanding
of the data in each country as well as a critique of previous works. Finally,
this part proceeds to a detailed description of the computerised organisation
of the data in order to make the files user-friendly and straightforward.
Following the principle of a cumulative and interactive conception of
research which seeks the collaboration between research centres, a detailed
and exhaustive description of a set of standard rules for the collection
of electoral data has been laid out, thereby making it possible for further
research to be incorporated. These rules concern the way in which information
- structured according to very different electoral systems - can be transformed
into a computerised and analytically useful form. The computerised nature
of the collection follows a second principle on which the project is based,
that is, the possibility of making data available for transformation,
improvement and analysis, thereby reducing the gap which exists between
the possibilities offered by the technical means available and the actual
use made of them in the field of political science.
As far as temporal periods are concerned, the basic units are defined
according to legislative election years, both general and partial renewals
of the lower houses. The starting point is defined by the sources themselves
and not by independence, boundary changes or levels of democratisation.
As regards Austria, for example, the collection starts with 1873 and covers
the entire Austrian half of the Habsburg Empire after the Ausgleich
of 1867 (Cisleithania). Concerning the Irish case, data for the Republic
is presented starting with the first election after the Anglo-Irish Treaty
of 1922, but results for the 1832-1918 period have also been collected
and inputted in the files for the United Kingdom.
Table
1: Countries, period covered and levels of aggregation: summarised
information Countries Period Units N covered
Table
2: Electoral variables
The choice of the territorial unit for which data was collected varies
from country to country, in particular according to the electoral system.
In the case of plurality systems, the constituency represents the natural
unit, although results were also aggregated according to other regional
bases. The constituency also was the rule in the case of STV. For proportional
representation systems - given the bigger size of the constituencies -
a more disaggregated unit was sometimes preferred (e.g., Italy 1946-1992).
Units are usually consistent with the traditional presentation system
of the data used by official sources.
Parties were selected on the basis of their size, i.e., they are included
if they poll at least five per cent of the total nation-wide vote and/or
at least five per cent in at least one territorial unit (see Rose and
Urwin 1957, p. 18 and Urwin 1983, p. 228). Parties which fulfill none
of these criteria were added to the 'other parties' columns.3
Some gross indications of these choices are given in Table 1.
2. Files on election returns:
(Table 2)
Files are constituted by a simple rectangular structure in which the
cases are the constituencies and the variables the different identification,
electorate and voting variables.
As far as cases are concerned, the main problem was to find standard
solutions to boundary changes, practices of gerrymandering and redistricting
as well as to national boundary modifications, in particular for earlier
periods. Furthermore, cases were made identifiable by type of area (urban,
rural, etc.). Finally, national totals were also computed. These problems
were solved by defining two identification variables (the 'unit' and 'area'
variables). The different combinations of these two variables permit the
identification and the selection of constituencies and/or national values
for all kinds of areas. There is no space here to describe the technicalities
employed to deal with the problem of boundary changes: apparition (new
units), disappearance, merge, split (secession) or transfer of a part
of a unit to another unit. Details are given in the code book.
In their standard form, files on election returns contain the following
variables:
- the already mentioned 'unit' and 'area' identification variables,
- three electorate variables (number of seats to be returned in each
constituency, number of persons entitled to vote and the number of actual
voters),
- three general voting variables (the number of valid votes, minor parties
or dispersed votes and unknown votes),
- and a series of variables concerning the valid votes to parties and
lists.
As one can see, no 'time' or 'year' variable was defined. The reason
is that in these files time does not flow vertically but horizontally,
from the left to the right. Variable names are therefore composed of the
name of the variable itself (e.g.,'vv' for valid votes) and the year of
election concerned (e.g., vv1848).4
Obviously, things are not as simple as they appear here. In the case
of multiple voting (as many votes as there are seats to be returned within
a given constituency) an 'x' is added to the variable name and a 'y' when
estimations taking values back to the 'one (wo)man, one vote' equivalence
are available.5 Similarly, the extension 'p' appears in case of partial
elections (e.g., Belgium and Luxembourg). Problems such as unopposed constituencies,
elections by acclamation (or show of hands), uncontested constituencies
in which a single candidate is elected by a 'yes' or 'no' vote (e.g.,
Denmark), curial systems, etc. also had to be solved. The measures adopted
to solve these problems are described in detail in the code book of the
collection.
Table
3: Variables of democratisation and representation Variable
name Description of the variables
3. Files on democratisation and representation: Table 3
The information contained in the second set of files draws from a large
amount of literature on political development and from previous research.
The goal of the collection was to give this information an analytically
useful form. In other words, historical information needed to undergo
a systematic coding process and had to be transformed into operational
and numerical material through the identification of a set of indicators.
The data collected in this part of the work is not territorially disaggregated.
The cases are therefore not constituted by territorial units as in the
first set of files, but rather by countries at a given election year.
The time dimension, therefore, flows vertically for each country file.
The variables were grouped according to three concepts:
1. enfranchisement;
2. equality of voting conditions;
3. representation system.6
The coding of these three concepts implies the formulation of operational
definitions. This shows that the concepts of enfranchisement and of equality
- the process towards democratisation - are ordinal (some-times cardinal)
variables of historical development, while the concept of representation
is a categorical variable of typological nature. With respect to the first
two concepts it is, in effect, possible to speak at least in terms of
'levels' of democratisation, even though it is sometimes problematic to
determine 'how much' and therefore to build continuous variables. By contrast,
the categories which define different representation systems do not constitute
orders of more/less or better/worse, but rather types.
The path towards universal suffrage (enfranchisement) can be described
by following two dimensions. On the juridical side, three indicators show
the steps of enfranchisement and of reduction of restrictions denying
the right to vote: census, capacity and sex. On the demographic side,
three other quantitative indicators provide for measures of the levels
of enfranchisement: persons entitled to vote, persons entitled to vote
as per cent of population over 20 and persons entitled to vote as per
cent of population having the voting age (see Flora et al. 1983, pp. 89-151).
One further indicator determines the level of enfranchisement. This is
the voting age. Plural/equal voting, the provision of secrecy and (in)direct
voting are the indicators of the equality of conditions among citizens.
Finally, five elements characterise representation systems: electoral
laws (the method used to transform votes into seats), the compulsoriness
of voting, the size of the constituency, the length of the legislature
and the size of the lower house. Further and more precise indicators can
easily be added to this initial core of information
4. Conclusive remarks
The spur to carrying out the project presented in this note stemmed
from the belief - on which scholars like Rokkan insisted - that political
science needs solid empirical bases, that is, the systematic collection
of information in the form of empirically measurable variables. Nowadays,
this should be done by means of computerisation. The low level of computerisation
of electoral data and the discrepancies between the sophistication of
the means at our disposal and the use which is made of them in the field
of political science are still surprising. The computerised form, in particular,
is a sine qua non condition for any kind of collaboration.
It is not possible to summarise the contents, the problems and the conceptual
and technical solutions which were adopted in a short article. This, indeed,
was not the aim of this note; instead, it was rather intended to inform
specialists that the systematic collection of election results is about
to be concluded, that a set of standard rules has been laid out, and that
new, better and more detailed information can easily be incorporated in
the existing body. The work done so far is a starting point. The collection
of this kind of data must be considered a continuous and cumulative process
of collaboration in which individual and separate efforts of scholars
and research centres converge. Ack-nowledgements are therefore addressed
in particular to all scholars, universities, official bodies, libraries
and other institutions which made the collection possible through their
collaboration.
Notes
1. The collection of the data is not limited to Western Europe. It will
include the results of several Eastern countries, notably Poland, Hungary
and Czechoslovakia (Czech Republic and Slovakia).
2. The collection does not have the ambition to improve the quality
of the outstanding comparative works carried out in the wake of Rokkan's
impulse, such as the inestimable collective researches of Rokkan and Meyriat
(1969) and Flora et al. (1983-87). By contrast, the present collection
certainly aims to improve the quality of electoral data, for which the
only existing comparative work presents results exclusively at the national
level (Mackie and Rose 1991).
3. This varies in the case of plurality systems for which results are
given for all candidates even when these belong to the same party and
do not reach either one of the two 'five per cent' criteria.
4. According to the number of cases and variables, the size of the files
varies from country to country. Files have been compiled to give the possibility
to switch the time dimension from its present horizontal position to the
vertical position. Election years are thereby transformed in cases and
time series techniques can easily be employed.
5. The term 'multiple voting' is used for cases in which voters are
allowed to cast as many votes as there are seats to be returned in a given
constituency (e.g., Belgium 1847-1897, Switzerland and Luxembourg). The
term 'plural voting', by contrast, is used for cases in which only some
voters are allowed to cast more votes than the rest of the electoral population
on the basis of census, capacity, indirect voting and curial systems.
6. The term 'representation system' is used instead of that of 'electoral
law' since the latter is preferred for a description of those mechanisms
through which votes are transformed into seats while the former indicates
the more general voting modalities.
References
Flora, P. et al., State, Economy, and Society in Western Europe 1815-1975.
A Data Handbook in two Volumes; Vol. I (1983), The Growth of Mass Democracies
and Welfare States; Vol. II (1987), The Growth of Industrial Societies
and Capitalist Economies; Frankfurt, Campus Verlag.
Mackie, T.T. and Rose, R. (1991, 3rd ed.), The International Almanac
of Electoral History, London, Macmillan.
Rokkan, S. and Meyriat, J. (eds.) (1969), International Guide to Electoral
Statistics; Vol. I, National Elections in Western Europe, The Hague-Paris,
Mouton.
Rose, R. and Urwin, D.W. (1975), Regional Differentiation and Political
Unity in Western Nations, London, Sage Professional Papers.
Urwin, D.W. (1983), Harbinger, fossile or fleabite? 'Regionalism' and
the West European party mosaic, in Daalder, H. and Mair, P. (eds.), Western
European Party Systems: Continuity and Change, London, Sage, pp. 221-256.
Further information about the project can be obtained from:
Daniele Caramani
Project on Comparative European Electoral History (1830-1995)
Research Archive Eurodata
Mannheimer Zentrum für Europäische Sozialforschung
Universität Mannheim
D-68131 Mannheim - Germany
Phone: +49 621 292 1723
Fax: +49 621 292 17 35
E-mail: daniele.caramani@mzes.uni-mannheim.de
Daniele Caramani is a researcher in Political and Social Sciences
at the European University Institute (EUI) in Badia Fiesolana, Florence,
Italy. He currently conducts his research activity at the MZES of Mannheim.
|