Geography, not Socio-Demographics: Explaining the
PDS (Party of Democratic Socialism) Vote in Berlin
John
O’Loughlin
Frank
Witmer
Valerie
Ledwith
*
John O’Loughlin is Professor of Geography in the Institute of Behavioral
Science University of Colorado Boulder Frank
Witmer is a PhD student in the Department of Geography
and in the Institute of Behavioral
Science University of Colorado Boulder Valerie
Ledwith is a PhD student in the Department of Geography, University of California Los
Angeles Ron Johnston of the
University of Bristol (UK) for providing us with the EMax entropy maximizing
program. Tom Dickinson and Clionadh
Raleigh helped us with the preparation of the paper.
ABSTRACT
The PDS (Partei
Demokratische Sozialismus – Party of
Democratic Socialism) continues to gain electoral support in the pos-unification elections
in the former German Democratic Republic (
_____________________________________________________________________________________
The process of the unification of Germany Berlin Berlin Germany Hesse ), the governing
coalitions of Länder are seen as
tests of how well parties can cooperate and as predictors of a possible federal
government coalition. But, though the
Berlin SPD-PDS coalition governs the entire city-state, the geographic bases of
support of the coalition partners are clearly divided into two halves along the
line of the former Berlin Wall. The
German capital, though politically united, still shows a powerful wedge between
the two parts – “die Mauer in den Köpfen” (the wall in people’s heads) - as
seen in electoral statistics and public opinion polls.
In this
paper, we examine the PDS support in two recent elections, those for the Landtag (state parliament) in 1999 and
2001 using detailed data for the electoral districts of Berlin East Berlin , the
availability of a large representative sample allows a matching of survey and
aggregate data to probe the geographic contexts of the votes. We ask if the PDS support from specific
groups (e.g., the ratio of people over 60 who choose the party) varied
according the overall trends or if there are special circumstances that can
account for the relative success in certain districts. Though the survey sample was large (a sample
of 15,936 in 1999), the respondents are not adequately distributed across all
78 electoral districts (Wahlkreise)
to allow analysis by district. Instead,
we must infer the district-level estimates using aggregate data. Recent developments in ecological inference
enable such inferences and in this paper, we compare the results from EzI
(Ecological Inference – King, 1997) and EMax (entropy maximizing – Johnston and
Pattie, 2000). While the results are
similar, the results presented in this paper offer another check on the
consistency of the two methods and an evaluation of the comparative advantages
of each. The district estimates are then
used to gain insights into the variation of the kinds of support of the PDS
across socio-demographic groups.
The
growing electoral successes of the PDS in
The October 21, 2001
Table 1. Percent Vote by Party
Table 2. Percent Vote for PDS
Berlin
The ideological divisions that had
permeated post-war politics in the city were made chillingly clear with the
construction of the Berlin Wall in August 1961 (Zimmer, 1997; Elkins and
Hofmeister, 1988). Although an armed
boundary between East and
In hindsight,
the reunification of
On
Former
German Chancellor Willy Brandt's famous prediction, as the Berlin Wall fell,
that “what belongs together will now grow together” (quoted in Leiby, 1999,
109) shows the degree of political naivéte that permeated
Direct federal subsidies to
The German electoral system has been
termed a “personalized proportional law”, with half of the parliamentary
deputies elected by a plurality vote in single member districts (Erststimmen) and half by proportional
representation from land lists (Zweistimmen)
(Conradt, 1996, 155). The first vote is
given to individual candidates, while the second is a party vote. The second (party) vote is usually deemed
most important because it is used to determine the final percentage of
parliamentary mandates a party will receive.
In order to ensure equal proportionality, the number of seats won in the
district elections are deducted from this total; thus the more district mandates
(first votes) a party wins, the fewer party (second vote) seats it will
receive. This electoral process can lead to a situation where there will be
more district votes (candidates) than the parliamentary mandate a party
receives through proportional representation (Edwards, 1998). This is known as
the Überhangmandate (excess mandate), and was an important
electoral loophole in the party strategy of the PDS in the federal elections of
1994 and 1998. In addition to its
personalized proportional law and its two vote system, the
The PDS: The
Regional-Alternative Party
Since the reunification of
The
purpose of this paper is to understand why the PDS as a post-Communist
political party continues to hold such strong regional appeal and to evaluate
the possible diffusion of its support. Among
the multiple explanations proffered, a key idea equates the success of the PDS
with its appeal as a regionalist-protest party.
Its success has surprised many
political commentators, who had assumed that it would not outlive the immediate
post-unification period in
Unlike many other former Communist
parties, the PDS confronted the challenge in 1990 of being immediately inserted
as a marginal actor into an externally directed, fully established functioning
political system (Phillips, 1999). While
the founding elections around 1990 were largely electoral disasters for
ex-Communist parties, the PDS made it to the Bundestag, by utilizing the
constitutional loophole afforded by the Überhangmandate.
This electoral success is even more
impressive since the party received no funding from the West German government
prior to the election. In contrast, all
three establishment West German parties, (CDU, SPD and FDP) contributed direct
financial support to their Eastern counterparts (Neugebauer and Stöss,
1996). The PDS consolidated their
success in the second round of national elections in 1994 by gaining a total of
30 seats. This reflects their continued
use of the Überhangmandate, but also
an increase of 2.4% of the national vote from 1990 to 4.4% (Kleinfeld,
1995). This was seen as strengthening
the party’s role as a reformed party and as further proof that they were a
resilient political force (Ziblatt, 1998).
This resilience is further confirmed by the party’s performance in the
1998 Federal election. For the first
time in its history, it cleared the electoral hurdle by polling 5.1% in the
Federal Republic as a whole (McKay, 2000).
The consolidation of the PDS as a
legitimate political party is not uniformly accepted by the German political
establishment, despite a growing academic consensus that ex-Communist parties
can have a stabilizing influence in the transition to a consolidated democracy. Post-Communist successor parties can
re-socialize an otherwise excluded and alienated segment of the electorate and
enable a more responsive party system (Mahr and Nagle 1995; Higley et al.,
1996). McKay (2000), however, asserts
that the persistence of an elite from Communist times threatens democratic
processes because these cadres are not interested in reform, but rather in
consolidating their elite position. Waller (1995) believes that the
organizational skills and resources of ex-Communist parties give them a
political advantage that is not necessarily democratic. Morueau (1998, 283) is
even harsher in his assessment of the PDS, defining them as a leftwing
extremist, anti-system party that promotes societal change through radical
democratization as camouflage. This view
is also the one most often taken by the centrist CDU party in Germany, who view
the historical ties of the PDS to the SED as good reason to isolate them from
mainstream politics (McKay, 2000).
Indeed, the CDU wants to put the PDS under the surveillance of the Verfassungschutz,
the agency in charge of protecting the constitution (Patton, 2000).
Ishiyama (1999c) takes an optimistic
view of the PDS, viewing “electoral demand” - issues relating to economic
conditions, traditional identities, uncertainty caused by transitions and the
related nostalgia for the Communist era - as important in explaining the
success of former Communist parties.
Ishiyama (1999b) makes further distinction between a successor party and
an adaptive party. The former claims
their successor status and retains members from the old Communist party, while
adaptive parties repudiate their Communist ideology and accept democratic
norms. It is clear that the PDS falls into the adaptive category more clearly
than the successor camp. Its critics
argue the impossibility of this, given the continued role of former SED members
in the PDS and the continued presence of a hard-line Stalinist faction (Patton,
2000). The SED had tried to position
itself on the left of the political spectrum as a contrast to the Kohl
government coalition. It soon realized
that this electoral strategy would not ensure its survival, due in a large part
to plummeting support and internal disagreements about the direction the party
should take (Barker, 1998; Hough, 2000b).
By June 1990, the SED membership of 2.8 million in 1989 had plummeted to
350,000 members (Kleinfeld, 1999).
Recent membership figures of the PDS estimate the figure is
approximately 100,000, or less than 5% of the SED membership before the end of
East Germany (Kleinfeld, 1999). This
trend is also true in East Berlin.
Unlike its electoral support, membership has consistently declined
during the 1990’s. In March 1991, the
PDS had over 42,000 members in Berlin but by the end of 1998, membership stood
at under 17,000 (McKay, 2000).
The Electorate of the PDS
A dramatic contrast in age profile exists between the membership of
the PDS and those who support the party in elections. PDS members tend to be over 50 years of age
and overwhelmingly former SED members.
PDS voters in former East Germany, on the other hand, tend to be
younger; about 25% of voters between the ages of 18 and 24 voted for the PDS in
the 1994 Bundestag election(Krisch, 1996).
Despite an increasingly ageing membership, the party has been able to
register support equally from across all age categories, with the over 60 age
group being slightly the weakest (Hough, 2000a). This age distribution is important in terms
of the survival and future growth of the PDS as a small party within the German
electoral system. As party membership
ages, the appeal of the party to younger voters is extremely important for
ensuring the reproduction of its support (Kleinfeld, 1995). However, this is not a simple undertaking. In the Eastern Länder in general, party
membership appears to have little appeal for young people and the moderate
parties in particular look decidedly middle-aged (McKay, 2000). It is here that the historical ties of the
party may enable its continued success in the East. The PDS funds 23 Arbeitgemeinsaften (AGs: workgroups), to deal with different
interest groups within their constituencies.
These local level organizations within the party, a throwback to the
importance of party-organized civil society in the GDR, are run independently
from the party’s central structure (McFalls, 1998; Kleinfeld, 1995; Brie,
1995). It seems plausible that this
strategy can play a dual role of
attracting new members and voters for the party, and consolidating the party’s
objective of nurturing a more
‘bottom-up’ approach to democracy (Neugebauer and Stöss, 1996; Ziblatt,
1998).
Unemployment was
the most important issue motivating voters to choose the PDS in the 1994
elections (Dalton and Bürklin 1996, Gibowski 1995, Krisch, 1996). Support for
the party continues to be strongest in the outer areas of East Berlin where the
economic benefits of the redevelopment of the old city center and the transfer
to the federal seat of government from Bonn to Berlin have not yet created new
employment opportunities. The PDS cannot
be defined as merely the party of the unemployed; in fact, the PDS is certainly
not exclusively a working-class party, nor does it appeal only to those on the
lower rungs of the social ladder. It
attracts intellectuals and individuals with above-average interest in politics
(McKay, 2000). In effect, the PDS is
becoming the party of those employed who are concerned about unemployment, a
new worry for citizens of the former East Germany. As such, the party also benefits from Ostalgie, or East German memories of the
good elements of life in the former German Democratic Republic, among which was
guaranteed employment (Roth, 1999; Wiesentha, 1997; Zelle, 1998).
Three
sets of explanations are usually suggested to explain the post-1990 electoral
success of the PDS. First, the PDS vote
is characterized as “catch-all” in nature, made up of support of government
officials, civil servants, university students, workers and the unemployed of
the former GDR. The appeal of the PDS to
university-educated voters also holds true in the West, where the PDS did best
in districts with a high percentage of the population with a University or Hauptschule qualification, and a lower percentage of
technical school graduates (Gibowski, 1995).
In general, the Eastern electorate is more secular and liberal on social
issues (Roesler, 1998; Krisch, 1996; Dalton and Bürklin, 1996; Schmitt, 1995;
Kleinfeld, 1995;Gibowski, 1995; Bauer-Kaase, 1994; Rattinger, 1994). As a “catch-all protest party,” the PDS poses
a challenge to the Western political system by representing dissatisfaction and
frustration with the liberal democratic norms shaping contemporary German
society (Strom and Mayer, 1998; Oswald, 1996; Roesler, 1998). Established Western parties such as the CDU,
who initially had the support of many GDR citizens, are no longer seen as
legitimate representatives serving the needs of the Eastern electorate. Protest voting is not the expression of a
wish to return to the GDR, but rather an expression of distaste with the harsh
realities and uncertainties of capitalist life (Hough, 2000b).
The national
election of 1998 set a historic precedent since it was the first time in
Germany’s history that five parties achieved more than 5% of the Zweistimmen
and entered the Bundestag. In the
campaign, the CDU helped to determine the protest vote niche for the PDS by not
only ignoring the geographical dimension to contemporary German politics, but
further polarized the electorate by attempting to link the PDS with the
negative aspects of Berlin’s history such as the building of the Wall (McKay, 2000). The CDU strategy was to mobilize the more
conservative voters to vote against the PDS in the belief that a higher voter
turnout would weaken the PDS (Krisch, 1996; Dalton and Bürklin, 1996;
Kleinfeld, 1995). Instead, the 1998
election demonstrated a solid and expanding appeal of the PDS to the Eastern
electorate, reaching 19.5% (Rattinger and Kraemer, 1998). To sweeten the victory, the PDS breakthrough
came in a year of the highest voter turnout (82%) since unification (Phillips,
1999).
Since
unification, the PDS has seen its support in regional elections in the five
Eastern states go steadily upward, highlighting the electoral disunity of the
country. A second explanation for the
PDS success, political culture, claims that forty years of Communist rule led
to different social and cultural norms in the former Eastern state. The socialization process under the Communist
regime created a different political culture because, within East Germany,
national ideology and identity were closely bound with Communist principles of
egalitarianism and a working class culture (Segert, 1998; Mushaben, 1997;
Langguth, 1995). Citizens of the former
GDR are now searching for a new German identity, having been socialized in that
part of Germany where identities were framed in anti-West German sentiment.
Howard (1995) argues that East Germans constitute a separate ethnic group,
characterized as a self-perpetuating and territorially bounded group, with a
shared historical identity and political representation (i.e. the PDS). The identity is often typified by Ostalgie, where the past is increasingly
romanticized by Easterners as a critique of the Other, West Germans. Segert
(1998) characterstic this Ostalgie as
“problematic normalization”, where one’s sense of identity is tied to a deviant
sense of Other in order to
rationalize the former Eastern identity.
The utilization
of regional rhetoric gained a firm foothold in the party ideology of the PDS by
1994, centered on the idea that the mainstream West German parties were
ignoring East German ‘interests’.
Reflecting widespread East German dissatisfaction with the process of German
reunification, it also asserted that the PDS stands as the most authentic
representative for East Germans as a regionally distinct culture (Ziblatt,
1998). The PDS used a blend of charisma, personified in its leader Gregor Gysi,
and language, that emphasized territorial identity, shared regional history and
the social, cultural and economic differences that exist between East and West
Germany to appeal to an increasing number of alienated Eastern voters. Importantly, the PDS had more organizational resilience
than a typical protest party and its electoral tactics involved more than
assuming the mantle of an Eastern protectorate.
Preceding the 1990 Bundestag
election, the constitutional court ruled that the 5% threshold should apply
separately to East and West Germany. In
response, the PDS focused its campaign solely on the Eastern region, resulting
in seventeen PDS representatives in the united German Parliament (Ziblatt,
1998, 18). In 1994, thirty delegates were sent to the Bundestag in Bonn, though the PDS got only 4 per cent of the vote
but up to 40 per cent of the vote in some Eastern Berlin districts and 20 per
cent in Eastern Germany overall (Ziblatt, 1998). In addition, it highlighted the fact that the
PDS had established itself as the third strongest party in Eastern Germany
(Wittich, 1995).
Prior to the self-identification of
PDS party leaders as regional guardians of East German interests (McFalls,
1998), the party was overtly concerned with the notion of “renewed socialism”
(Ziblatt 1998). The PDS adoption of its
role as defender of Eastern interests had its roots in a “grassroots”
organization called “the committee for social justice,” of which the PDS
chairman Gregor Gysi was a cofounder. As
this “committee” lost its political relevance in the years between 1990 and
1994, many of its ideas were co-opted by the PDS to ensure their appeal to the
Eastern electorate (Lang et al., 1995).
Furthermore, it was only eight months before the 1994 Federal election
when "Eastern interests"
became the rallying call of the PDS leadership.
In effect, the public shift of attitude towards viewing the unification
process as colonization (or Kohl-inization,
after the former German Chancellor
Helmut Kohl) by the Wessies
and the Treuhandshalt (the
privatization organization that sold off East German state assets) molded the
party’s new role and purpose as a regional defender by its perceived harsh
actions (Ziblatt, 1998).
Since the 1998 federal election,
there has been a further shift in the self-definition of the PDS as a regional
party. The official party line now
claims that they are a legitimate social-democratic party that will appeal to
the entire German electorate. However,
their electoral adversaries continue to pigeonhole the PDS as a party with
questionable roots, not at all interested in embracing the ideals of the
West. However, it is exactly this type
of attitude that is seen in the East as Western chauvinism and will continue to
define the PDS as the only party able and willing to represent the eastern
electorate. By continuing to serve a
sizeable group of people within an imperfect democratic system, the PDS is an
alternative to voters who would either abstain or support more radical and
politically-questionable right-wing parties.
The PDS will continue to offer a crucial stabilizing and integrating
factor in the East (Ertman, 2000).
A third possible
explanation for the role of the PDS is dissatisfaction with the democratic
system of government that is increasingly linked to economic performance in
post-Communist societies (Bauer-Kaase, 1994).
This connection is also seen in East Germany, especially because East
Germans place far more importance on social policy and social justice than do
West Germans (Eith, 2000). The syndrome
of post- unification dissatisfaction for many East Germans resulted from
adjustments to the social, economic and cultural challenges of West
Germany. Overall, East Germans have
experienced an increase in affluence, highlighted by rises in income,
retirement pensions and access to goods and services (Segert, 1998). However, income levels in Eastern Germany are
still only 73% of those in the West and pensions are approximately four-fifths
of a typical West German pension.
Economic restructuring has led to a complete transformation of the job
Kitschelt’s “winners–losers” hypothesis
(1992, 1994, 1995) holds that individuals in the former Communist East pay
close attention to their personal economic situation when deciding about their
vote (See also O’Loughlin et al., 1997 and Rattinger and Kramer,
1998). According to Kitschelt, voting
choice in post-Communist societies is driven by prospective and egotistic
considerations, while in consolidated democracies, there is a greater level of
retrospective voting. Increasing
economic development and affluence in former Communist countries induces a
general shift of voter preferences toward libertarian, participatory systems. These claims were initially validated by the
general acceptance in the former GDR of the liberal democratic model in
Germany’s first national election of 1990.
However, when the expectations did not match reality, a retrenchment
towards traditional values and disillusionment with the liberal economic system
of West Germany led to an increase in support for the PDS. The PDS appealed to the losers of
A
related hypothesis to the “winners-losers” model focuses upon the importance of
system performance in determining voter choice; in the East, voters take a more
pragmatic approach to democracy, judging it in terms of the comparative
performance of a socialist system (Conradt, 1997). Citizens expect to convert their individual
resources into economic benefits in the
Prior
to the 1994 national German elections, 84% of PDS supporters felt more like GDR
citizens than Germans (Neugebauer and Stöss, 1996). While this can overlap with the idea of
economic losers of unification, it could also be that there are distinct
strands of support within the PDS support base, one based on the loss of
identity, the other on economic loss.
From this perspective, the appeal of the PDS is not merely to the “old
guard” who are nostalgic for Communism, but equally its defense of the old way
of life that offers some stability to those most affected by the complexities
of post-unification society. This
approach allows conceptual space for both the regional and protest aspects of
the PDS party, which cannot be treated as clearly separate definitions. It also offers an explanation of the
geographic differences that exist not only between East and West; but also
within the East itself.
Data and Methods
Berlin was divided
into 23 Bezirke (districts) until the municipal reform of 2000 that
combined some of the smaller districts.
Each of the Bezirke has its own council and though the Berlin
Wall is now removed, its line
Figure 1 - Locations of the 23 Bezirke (districts) of Berlin prior to the redistricting in 2000
that reduced the number to 12 by combining Friedrichshain and Kreuzberg;
Charlottenburg and Wilmersdorf; Steglitz and Zehlendorf; Tempelhof and
Schöneberg; Treptow and Köpenick; Marzahn and Zellersdorf; Mitte, Tiergarten and Wedding; Hohenschönhausen and Lichtenberg; Pankow,
Prenzlauer Berg and Weißensee while Spandau, Neukölln and Reinickendorf
remained as single districts.
The 23 Bezirke
are divided into 78 Wahlkreise (electoral regions) and the results for
these regions provided the basic data for this paper. We concentrate on the party vote (Zweistimmen)
rather than the candidate races in the individual 78 regions. As can be seen from Table 1, these two vote
percentages are typically very close and though some candidates, like the party
leaders, receive a personal vote in excess of the usual party support in a
region, the second vote is a better representative index of party
preferences. The second major data
source for the paper consists of demographic data for the electoral regions but
unfortunately, these data are only available for age and gender
categories. The third source consists of
a large mail survey (a total of 15936
respondents) at the time of the 1999 election from which we generate regional
level estimates of the support of age and gender groups for the PDS (Bömmerman,
1999).
A comparison of the distribution of
the PDS first and second votes in 1999 and 2001 in the four maps in Figure 2
shows both strong correlations and a diffusion of the PDS zone of
dominance. The biggest difference in the
maps is the relative size of the regions of the two votes. In 1999, the number of Wahlkreise with
over 40.6% support for the PDS was similar for the first and second votes; by
2001, the number of Wahlkreise with more than 40.6% for the PDS second
(party) vote was dramatically larger than the first vote, though both showed
gains over 1999. These trends, though
only for a two year period, match wider developments in the PDS diffusion. Early successes came in the form of
individual candidates winning in the single-member constituencies (first
votes). Over time, the party as a whole
is gaining new adherents, while continuing to maintain their individual seats
and even increase them. In the 2001
second vote distribution, the PDS gained more than 40.6% in every one of the 32
Wahlkreise in former East Berlin and won a further 20-30% of the vote in
the older, central districts of the unified capital, the heartland of the Green
party support in the 1990s (Doud, 1995; Lawson, 1998; Ledwith, 2000). More general comparison of the four maps
shows a classic diffusion pattern; the key question after 2001 is whether the
PDS can continue to spread their success to the traditional SPD areas of
working-class West Berlin and thus becoming a city-wide party like their SPD
coalition partners.
To estimate the composition of the
PDS vote across the 78 Wahlkreise, we turned to two procedures that have
been developed to infer individual voter behavior from aggregate statistics and
which can generate small area estimates for the variables of interest. While the voter exit polls are valuable in
providing “global”
or city-wide values
(e.g., the percentage of women who voted for the PDS), these are not available
in a reliable manner for the individual Wahlkreise. Survey data have only been published for the
1999 election, the focus of the remainder of this paper. A total of 116,646 voters were sent a mail
survey questionnaire in 124 Stimmbezirke
(electoral precincts) across Berlin for a sample of 4.77%. This poll was re
Figure 2 - Distribution of the First and Second Votes for the PDS in the Berlin
Landtag elections of 1999 and 2001 by the 78 Wahlkreise (election districts).
indicated by the
close fit of the polling data and the actual numbers. For example, survey results indicated that
38% of East Berliners voted for the PDS (the actual proportion was 40%) and the
respective values for West Berlin were 4.5% and 4.4% (Bömermann, 2001). Since the EMax method used for small-area
estimates uses the global survey data, it is imperative that these values stand
as an accurate reflection of the electoral preferences.
The first procedure used to generate
the Wahlkreis estimates is the
ecological inference model from King (1997).
EzI (ecological inference) has been widely used in political science to
estimate vote-splitting, racial bloc-voting, and electoral choices for
socio-demographic groups. In geography,
the procedure offers great prospects for revitalizing the discipline of electoral
geography to allow individual-aggregate analyses and it has been the subject of
favorable review by geographers (Johnston, 1998; Fotheringham, 2000;
Davies-Withers, 2001) and used to understand the variation of the vote in
Weimar Germany (O’Loughlin, 2000, 2002) and Ukraine (O’Loughlin, 2001). Because of this growing familiarity, the method is only briefly
summarized here – further details are available in King (1997).
In
order to calculate the PDS support in Berlin among older voters (aged 60 and
over), we use the proportion of the population over 60 (Xi) and the
PDS percentage (Ti) for each geographic unit, in this case, the 78 Wahlkreise geographic units of Berlin in 1999. Using King’s notation, in the current
example, the independent variable, X, is the aged 60 population and the
dependent variable is PDS support, T. An
identity is used for combinations of the district values for Ti (PDS)
and Xi (over 60), Ti = bib
Xi + biw
(1 – Xi). The purpose of the
EzI modeling is to estimate βb (the aggregate rate of the PDS
vote among the over 60 population for the whole city) as well as the estimates
for the individual Wahlkreise (78 units in all), bib. Combined with information about the bounds
for each district, found by projecting the regression line onto the horizontal
(bib
, the PDS vote) and the vertical (biw,
the non-PDS vote) axes, the EzI method combines two earlier inference methods
(King, 1997; see also Figure 4 in this paper).
When the bounds on the axes are narrow, the stronger are the chances of
a plausible and accurate solution.
Reliable estimates for the turnout rate of Kuchma voters in Ukraine were
presented by O’Loughlin (2001) using this methodology.
The second procedure, termed EMax,
differs from King’s in two main ways (Johnston and Pattie, 2000). First, EMax uses the mathematical
entropy-maximizing approach rather than a statistical approach. Second, EMax
requires additional data that describe global patterns to constrain the
estimates further. These additional data
are usually obtained from sample surveys.
This latter requirement means, in effect, that EMax can be used in fewer
circumstances then EzI since public opinion surveys became commonplace only in
the past half-century. Historical
studies, such as the composition of the support for the 1930s Nazi party, must
rely on EzI for estimates (O’Loughlin, 2002).
The entropy-maximizing procedure produces maximum-likelihood estimates
given a set of initial constraints. The
EMax procedure is best explained by example.
For a fictional district of Berlin with 9 voters, consider the
following gender and PDS vote row and column sum constraints:
|
|
PDS |
not PDS |
|
|
Female |
? |
? |
5 |
|
Male |
? |
? |
4 |
|
|
3 |
6 |
|
In this example, we are interested in
finding the most likely allocation of votes for the PDS, which requires
calculating the value of each cell denoted by the question
The EMax
procedure starts from the macrostate and calculates all possible values for
each question
|
|
PDS |
not
PDS |
PDS |
not
PDS |
PDS |
not
PDS |
PDS |
not
PDS |
|
Female |
0 |
5 |
1 |
4 |
2 |
3 |
3 |
2 |
|
Male |
3 |
1 |
2 |
2 |
1 |
3 |
0 |
4 |
Since these four combinations are the
only possible solutions that satisfy the given constraints, one of them must
reflect the actual breakdown of votes for this district. EMax selects the mesostate based on the
number of associated microstates, where each microstate represents a
different distribution of voters. For
instance, in the first mesostate, the five female voters can only be allocated
one way: all voted for a party other than the PDS. Each one of the four males, however, could
have voted for a non-PDS party.
Therefore, there are a total of four microstates associated with the
first mesostate. For the second
mesostate, any of the five females could have cast the one vote for the PDS,
yielding five microstates. For the
males, any of the four could have cast the first PDS vote, leaving only three
to cast the other PDS vote. This yields
12 possible combinations, but since order is not important, these are reduced
to six microstates for the males.
Assigning letters to each male voter – a, b, c, and
d – the PDS and non-PDS voters may be enumerated: (1) ab, cd
(2) ac, bd (3) ad, bc (4) bc, ad (5) bd,
ac and (6) cd, ab.
Combining the five female allocations with the six male allocations
produces a total of 30 microstates for the second mesostate. There are 40 microstates associated with the
third mesostate (ten ways to allocate the five females and four for the males)
and ten associated with the fourth (ten for the females and one for the
males). Since the third mesostate has
the most possible microstates, the EMax procedure chooses it is the maximum
likelihood solution. A more detailed
description of the EMax procedure is found in Chapter 5 and Appendix 3 of
Johnston (1985)
This example
above, however, only uses two constraints, the row and column sums as in King’s
EzI method. To increase the accuracy of
the results, EMax requires an additional sum over all districts in the study
area to obtain its estimates; it generates these numbers from the survey data,
in this case, the ratio of men (and women) who voted for the PDS. EMax applies the same maximum likelihood
procedure as in the example, but ensures that none of the mesostates violates the
added global constraints. Since we are
interested in who voted for the PDS not just in one district, but for each of
the 78 Wahlkreise of Berlin, the
above example can be extended by including row and column sums for each
district, and global sums from Berlin-wide survey data. The three sums required are thus (i) the
votes for the PDS and for all other parties within each of the 78 Wahlkreise, (ii) the voter turnout by
gender for each Wahlkreis, and (iii)
the Wahlkreis totals of PDS and
non-PDS votes by gender for all of Berlin.
Note again that items (ii) and (iii) require additional information not
used by King’s method. Item (iii) can be
derived directly from Berlin survey data but item (ii) must be estimated by
performing an additional EMax calculation since no such turnout data by gender
are available for each district. The
sums for this additional EMax procedure are (i) the voter turnout for each
district, (ii) the number of registered voters by gender, and (iii) the
citywide turnout by gender. Again, item
(iii) is available from the survey data.
The estimated results from this EMax calculation were then used as input
to the second estimation. The
implementation of the EMax procedure uses an iterative process to estimate the
results. It terminates after the maximum
number of iterations are complete or when the threshold parameters are
met.
The same double EMax technique
was also used to determine the spatial distribution of PDS support by age. Citywide polling data (Bömermann, 2001)
provided total turnout by age group (18-34, 35-59, 60+) for the first EMax
calculation, and Statistisches Landesamt Berlin (1999) provided total
PDS votes by age for all of Berlin.
Though this double EMax technique introduces additional error, the
aggregate results of the procedure are very similar to the survey results, with
EMax estimates within two percent of survey results (Table 3). In order to get the global input indexes for
the EMax, the values from the postal survey were recalculated by weighting the
disaggregated group values by the number of voters in each age group. This weighting produced estimates of turnout
for the 18-35 age group of 52.32%, for 35-60 years of 67.23%, and for over 60
of 75.10%. Similar weighting adjustments
were made for the PDS vote for the same age groups which yielded city-wide
estimates of 19.2% support for the party among 18-35 year olds, 19.9% for the
35-60 year olds, and 15.6% for those over 60 (Table 3).
Ecological Inferences of Turnout and PDS Vote in
Berlin, 1999
One of the major drawbacks to the
widespread use of the entropy maximizing methods is the limited availability of
the kinds of polling data that are needed to “drive” the method (Johnston and
Pattie, 2000). Berlin is a good test of
the respective values of the two ecological inference methods. We can compare the EzI estimates to the
survey estimates for the whole city (global estimates) and we can correlate and
compare the individual estimates of the two methods for the 78 Wahlkreise. We are thus making a double comparison - of
the similarity of the estimates of the two methods and of the methods to the
survey data. Of course, this latter
comparison is valid only if the survey data are accurate and this appears to be
the case as the close correspondence with the actual voting statistics
indicates (Bömermann, 2001).
The summary
ecological estimates are presented in Table 3, with the survey data also
presented for comparison. Since the
survey data are only available for 1999, the EMax estimates can only be
presented for that year and similarly, the EzI estimates for 2001 cannot be
verified by survey data. It is clear
that EMax provides a much better global estimate for the 1999 election than
EzI, as might be expected from the fact that the global survey results are used
to constrain the EMax values. The EzI
values are worrying, however, since the estimates are off by more than 10% for
the groups of interest. By contrast, the
EMax values are within 2-3% of the survey results. While no check can be made on the estimates
of the turnout of the PDS supporters from the survey data (it is impossible to
calculate a reliable value using the turnout estimate which are only available
for age groups, gender, and location in East or West Berlin), the PDS turnout
support is lower than the citywide average (65.6%). However, McKay (2000, 7) notes that there
tends to be a high correlation between strong support for the PDS and a higher
than average turnout.
Figure 3 – Comparisons of the initial
ecological inferences of the PDS support (%) among different demographic groups
in the 78 Wahlkreise (election
districts) of Berlin, using entropy maximizing (EMax) and King’s ecological
inference method (EzI).
To
try to understand why the EzI estimates are so far off the EMax values and the
survey results, we probed their distributional properties. Simple plots of the two sets of ecological
estimates (Figure 3) indicates the nature of the problem. In all four of the plots, while the EMax
estimates show a dramatic range (from 2% to over 50%) reflecting the actual
performance of the PDS in the 78 Wahlkreise
of Berlin, the EzI estimates are essentially invariant across the city, showing
only a few percentage points difference between East and West Berlin. This result is clearly inaccurate since the
maps in Figure 2 show a huge geographic spread of values, with the former Wall
Clearly,
EzI does not provide reliable estimates for either the global values or the
values for the electoral sub-units in Berlin.
To try to understand why this is the case, we examined the graphical
plots that are part of the optional output from the EzI program. Figure 4 displays the key indicators; a close
examination of the graphs shows both the inadequacies and benefits of King’s
ecological inference method. The basic
distribution of the data used to make the inferences is shown in Figure
4a. In this case, the aim is to estimate
the ratio of voters age 60 and over that supported the PDS. The size of the circle is proportional to the
number of voters in each Wahlkreis and
clearly, the difference between the districts in East and West Berlin is seen
in the two separate clusters of high and low PDS support. By contrast, the variation in the
distribution of the old population in the two parts of the city is not nearly
so great. The problem for the EzI
estimation can be seen in the lines on the graph, indicating the maximum likelihood
estimates (for the relationship of age over 60 and the PDS support). The middle line is the expected value of PDS
vote percentage given the ratio of the over 60 population and the other two
lines represent the 80% confidence intervals (King, 1997, 206). Only a small number of values fall within the
confidence bands as the maximum likelihood estimation procedure fails to pick
up the bi-modal (east and west Berlin) nature of the distribution.
Figure 4 – Distributional
Properties and Estimates of the Ecological Inferences of the PDS Vote in Berlin
1999. All the graphs are generated in
the EzI program (King, 1997).
In
Figure 4b, all possible combinations (tomography lines) of each pair of values
(for PDS and non-PDS support of the over 60 population) are plotted; the
posterior distribution of the quantity of interest is also shown. It is clear that aggregation bias is present
(the correlation between the district parameters and the independent variable, age
over 60). The bounds (where the tomography
lines cross the axes) are very wide, making accurate inferences difficult. Further evidence of the binary nature of the
Berlin data is provided in Figures 4c and 4d that show the density plots of the
estimated (inferred) district values for the PDS and the non-PDS votes. While the PDS distribution is peaked and
narrowly bounded around a mean of 5.4% (Figure 4c), the bi-modal nature of the
non-PDS vote shows that this distribution (Figure 4d) is not the antithesis of
the PDS vote, as it should be. The peaks
of this distribution at 5% and 55% show that this EzI modeling is flawed and
that alternative procedures are needed to generate reliable estimates for the
city as a whole and for its 78 constituent districts. King (1997, Chapters 9 and 16) offers
procedures to deal with such distributional issues, including non-parametric
estimation. In the case of Berlin,
however, it is obvious that the distributional problems are the result of the
bifurcated electorate in the city and it is evident that this key piece of
information should be used to derive the ecological inferences.
Figures
4e and 4f present the results of the EzI modeling using separate models for the
two parts of Berlin. Of course, the
division of the city for this modeling reduces significantly the number of
cases available – 32 for East Berlin and 46 for West Berlin. The tomography plots with the contour lines
contain the points with the highest probability of containing the true
estimates for each part of the city. The
black diamond shows the locations of the most likely values. The respective estimates of the support of
the over 60 population for the PDS are 38.2% for East Berlin and 2.8% in West
Berlin, which are a lot closer to the “true values” (from the survey data of
39% and 2.4% respectively) than the initial pooled citywide estimates. The Berlin example has clearly shown the
importance of careful consideration of the distributional qualities of the data
and the requirement that the researcher using EzI be familiar with the characteristics
of the data, as King repeatedly stresses in his book.
A
final check on the validity of the EzI estimates is a plot of the values for
the 78 Wahlkreise against the estimates generated by the Emax procedure; the plot is presented
for East and West Berlin separately in Figure 5. Johnston and Pattie (2001) have shown how a
comparison of the two sets of estimates, one derived from a statistical
simulation and the other from mathematical assignment, can offer creditability
to both methods if they correspond closely.
As can be seen in Figure 5, the correlation is very close for West
Berlin (all values very close to the diagonal) with a r2 value of
.99. In East Berlin, also, with a wider
spread of values, the correlation is high (r2 = .81) though there is
more scatter around the diagonal. We
conclude that the bifurcation of the Berlin data set, suggested by the
distribution plots of the initial EzI model and the evident indications of
aggregation bias, was the correct modeling option since the second set of
estimates match closely both the global city estimates (from the survey data)
and the local estimates (from the EMax proportions). Like any estimate, there is clear variation
around the EzI numbers and though these are maximum likelihood estimates, they
are not infallible. As King (1997)
notes, it is always valuable to check the EzI estimates to “true” values if
they are available. While absolute truth
is an elusive quality in survey data, the size and representativeness of the
October 2001 mail sample in Berlin suggests a close fit to the actual
numbers. The recalculated EzI values
match the survey data very well.
Unfortunately, such survey results are not available for the smaller
geographic unit of the Wahlkreis and so, the only check available at
this scale is against the EMax values.
Figure 5 - The Relationship between the EzI and EMax
Estimates of the PDS Vote among Voters aged 60 and Over, Berlin 1999.
The
Geographic Distribution of the PDS Support in Berlin
Having demonstrated the reliability of
the EzI and the EMax estimates for Berlin in 1999, we turn now to the
distribution of these values across Berlin’s districts, paying close attention
to the effects of the division of the city along the Wall and to the clustering
of values in certain parts of the city.
While both sets of ecological estimates could be used in this exercise,
we used the EMax values, except in the case of the turnout estimates which
could only be derived using EzI.
In
examining the political geography of the PDS vote, we turn to three standard
tools of spatial analysis – cartographic displays, Morans I index and the 
statistic. Recourse to these tools is growing quickly
and with the integration of statistical models with computer mapping packages,
it is expected to grow even faster (O’Loughlin, 2002). The global clustering of the estimates are summarized by the
Morans I values and the associated Z-values in Table 4.[1] It is evident that
the PDS vote is strongly clustered in Berlin in both 1999 and 2001 and that
there is very little difference in the levels of concentration of the first and
second vote. The distributions of the
votes in Figure 2 anticipated this result, especially since the former Berlin
Wall still acts as a major dividing line in the political culture of the city.
Table 4. Row Standardized Moran's I for PDS Vote and Turnout
The
ancillary question to the clustering of the PDS vote is whether there is a
similar trend in the estimates of the ratios of the various age groups that
supported for the PDS. Based on the EMax
estimates, it is clear from the Morans I values in Table 4 that the level of
clustering is very similar or stated another way, the geographic clustering of
the age components of the PDS vote are consistent. It appears to make little difference which
socio-demographic component of the PDS support is examined. All show a strong concentration of high values
in the same region of East Berlin and low values in the west of the city. In this Berlin example, geography clearly
outshines any socio-demographic explanation for the variation in the support of
the PDS.
While Morans I provides an effective
summary of the overall level of clustering, it is not capable of distinguishing
the nature of clustering. Is the high
value of the index the result of one large concentration or multiple smaller
ones? In order to distinguish between
these possibilities, we use an index of local association, thestatistic.[2] Mapping these values for each of the 78 geographic units in Berlin clearly
shows the core of the PDS support and the changes that occurred between 1999
and 2001 (Figure 6). The most
significant positive values (associated with similar values in contiguous Wahlkreise)
are found in the same districts in 1999 and 2001, predominantly in the Bezirke
of Marzahn, Lichtenberg and
Hohenschönhausen. These districts were
predominantly working-class districts in the Communist years and have not yet seen
the kind of improvements that have characterized the central parts of the
former East German capital (Cochrane and Jonas, 1996; Cochrane and Passmore,
2001). Around this core of highest
support, most of the other voting districts in East Berlin also show highly
significant values. Only in the extreme
northern district of Pankow and in the neighborhoods along the former Wall are
there zones of non-significant values. It is in these
regions that gentrification and re-development is proceeding fast, transforming
not only the streetscape but also, changing the population composition of the
increasingly-richer zones (Krätke, 2001; MacDonagh, 1998; White and Gutting,
1998). In West Berlin, the significance values are defined by the similarity of low PDS
percentages to similar values in the surrounding districts. Between the two zones of significance is
central Berlin, the districts along the former Wall and the historic center of
the city that was built-up by the mid nineteenth-century. These are among the
most dynamic in terms of population change and mixed in demographic composition
(Cochrane and Passmore, 2001). If the
PDS is going to expand and grow beyond its East Berlin core, it is these
neighborhoods that offer the best possibilities where the PDS will be competing
for voters with the SPD and the Green party in future elections.
Figure 6 - Distribution of the Indexes of the PDS
Vote Percentages by Wahlkreis in Berlin, 1999 and 2001.
The composition of the PDS vote
can be examined using the EMax estimates of the level of support by age
group. The survey results in Berlin
showed that the PDS gained more support from older voters in East Berlin (31.6%
of those aged 25 to 34 compared to 39% for those age 60 and more) and from
younger voters in West Berlin (6.0% for 25 to 34 year olds and 2.4% for those
60 and older) (Bömermann, 2001). Gender
differences were tiny (they are not examined further) and were dwarfed by the
geographic difference across the former Wall.
In the East, younger women were more supportive of the PDS but after age
45, men gave slightly more support to the party. As in most democracies, turnout increased
with age for both genders in both parts of the divided Berlin.
Mapping both the estimates and the statistics for the estimates allows a consideration of the
differential character of the PDS appeal across the city. While Figure 7 shows the estimates for the
35-59 age group (the largest age-defined voting block in the city), the maps
for the other age groups are very similar and are not shown here. In approximately half of the Wahlkreise
in East Berlin in the 1999 election, the PDS received more than 39.7% of the
vote of the middle-aged population, significantly higher than their overall
ratio in the city (17.7%). It is clear
from the maps in Figure 7 that the more economically-deprived regions of the
East are where the party does best. Not
only is the PDS a regional-protest party, claiming to represent all Easterners,
but it also gains added support from the poorer segments of the East German
population, attracted by its leftist ideology and its historical claim to voice
the interests of the working-class (McKay, 2000; Segert, 1998). In contrast to the core support in the East
is the very weak support in the richest parts of West Berlin where the party
received less than 5% of the vote in many voting districts. The map in Figure 7 is very similar to those of Figure 2 and
indicates again that the PDS vote does not vary as much within the
socio-demographic categories (by age group, by gender, by education, by income)
as it does across the former Berlin Wall.
The dramatic importance of
the geographic divide in Berlin compared to the small differences between the
usual socio-economic categories is further illustrated in Figure 8. Unfortunately, socio-economic data are not
available for the 78 Wahlkreise and we had to turn to the income data
from the micro-census that was conducted in Berlin in 1995 (Statistisches Landesamt Berlin, 1996)
and social welfare data collected yearly by
Figure 7 - Distribution of the EMax Estimates of PDS
Support among 35-59 Year Old Voters and Indexes by Wahlkreis in Berlin, 1999.
the Statistisches
Landesamt. These data were
reported for the 23 Bezirke of the city (before the municipal reform)
and by matching the Wahlkreise EMax estimates of the young voters (18-34
years old) to the Bezirke , e were able to develop the trellis plots of
Figure 8. (Trellis graphs are
conditioning plots with one or both axes further divided to illustrate key data
relationships; see Chambers et al.,
1983). On these graphs, the small
circles indicate the respective values for the 23 Bezirke (see Figure 1 for
their locations and boundaries) and the line indicates the loess curve. (Loess is a technique that uses
non-parametric regression procedures and is widely used in graphics
visualizations. It fits a function of the predictor variables in a moving
fashion that is analogous to a moving average in a time series. We used the S-Plus statistical package for
the modeling; see Cleveland, 1993 and Venables and Ripley, 1997). Loess modeling has the advantage of
summarizing non-linear relationships to highlight general trends. The income and welfare categories chosen here
are for illustration and do not cover all possibilities but their display
clarifies the divisions of the city.
In the highest
income category (monthly household net income of more than 3000
Figure 8 – Trellis Plots of the EMax Estimates of PDS Support among 18-35
Year Olds by Bezirk in Berlin, 1999.
These trellis
plots for Berlin are highly unusual for a Western democratic election. They demonstrate clearly that party support
is not related to neighborhood income levels but, instead, is determined by the
location of the household, either west or east of the former Wall. Though geographers have claimed for decades
that a voter’s location in the urban fabric is a small but important part of
the explanation of the electoral choice, the example of contemporary Berlin
clearly shows that it can be the most significant factor - far exceeding any of
the usual socio-demographic explanations.
While we are not arguing that Berlin represents any kind of typical
scenario for the future, it is perfectly clear that the legacy in the city
emanating from the Cold War division is still dominant in the electoral
landscape of the PDS, and by extension, to the other parties. As German electoral politics adapts to the
conditions of the post-Wall world, the speed with which this electoral canyon
is filled will be a good indicator of the resumption of normal Western-style
electoral politics.
The final
analysis reports the result of the EzI estimation of the turnout of the PDS
voters. In this modeling, the T variable
(the ratio we wish to calculate) is turnout while the X variable (used to make
the calculation) is the PDS 1999 and 2001 second vote. We had to resort to EzI because no citywide
estimates are available from census data for the relative turnout of PDS
supporters and non-PDS voters. For the
city as a whole, we estimated that the turnout of PDS supporters was 54.1% in
1999 and 55.5% in 2001, compared to citywide averages of 65.6% and 68.2% respectively. Turnout since the unification of Berlin has
been consistently lower in the eastern half of the city, the location of the
PDS core supporters. The lower turnout
partly reflects a disillusionment in the former GDR with the nature of the
German political system (Segert, 19989).
It has the effect of reducing the potential PDS membership of the state
parliaments and the Bundestag and suggests that a larger pool of PDS voters
might be mobilized to go to the polls.
The geographic
distribution of the turnout of PDS voters in Berlin is shown for 1999 and 2001
in Figure 9. Some voters in core PDS
areas will not vote because they think that the party is mobilizing its voters;
in areas with few PDS voters, their supporters might be more mobilized to vote
than the average voter in these neighborhoods.
In Figure 9, parts of East Berlin showed more than 60% turnout in 1999
but other Wahlbezirke had PDS voter
turnout as low as 50%. By 2001, the
turnout of PDS supporters increased in most districts and nine Wahlkreise in West Berlin showed significantly
more than average PDS turnout. It is in
the older and poorer parts of East Berlin and the richest parts of West Berlin
that the PDS voter turnout remains lowest.
Like most turnout displays, the maps in Figure 9 correspond to income
for East Berlin (lower income neighborhoods have lower turnout) but the
relationship becomes more complex for West Berlin where the special character
of the PDS party supporter defies the usual expectations.
Figure 9 - Distribution
of EzI Estimates of the Turnout of the PDS Supporters by Wahlkreis in
Berlin, 1999 and 2001.
Data limitations
prevent a more complete analysis of the socio-demographic components of the PDS
vote in Berlin. The lack of adequate
census data for the small areal units is especially discouraging since it
precludes the fitting of ecological inferential models for groups defined by
educational attainment, housing status, and occupation. Nevertheless, within the limitations of
existing data, we have shown clearly that the division of the city into its
eastern and western sectors along the line of the former Berlin Wall far
surpasses any compositional explanation of the vote for the PDS. Depending on which side of the geographic
divide they sit, different groups come together to vote in approximately the
same proportion for the party. Old and
young voters in West Berlin are much more alike than their age counterparts on
the other side of the former Wall. After
a decade of elections to state and federal parliaments, this divide is not
easing and is strengthening in some ways as the dominance of the PDS in the
eastern half of the city grows.
Conclusions
This geography of the PDS vote in Berlin
shows that “die Mauer in den Köpfen” is still a reality in German
elections. In his 1991 poem, “Die Mauer”
(The Wall), Reiner Kunze (1998) anticipated this development - “Als wir sie
schleiften, ahnten wir nicht, wie hoch
sie ist in uns” (“When we tore it down, we did not guess, how high it is in
us”). At the national level, the growing
support of the PDS is challenging the two main parties, the Christian Democrats
and the Social Democrats, to try to hold onto the eastern support that they
worked assiduously to cultivate in the aftermath of the unification in
1990. If either party loses its relative
standing in the east to the PDS, its claim to be a national party will be
hollow. After a decade of unification,
there is little sign that the large electoral and ideological differences
between East and West are ebbing. In
fact, one could argue that they are magnifying as the PDS goes from strength to
strength on the basis of an appeal that remains partly ideological (left-wing)
and partly regional identification. In
its September 2002 electoral campaign, the party is using slogans such as “That
no one loses is the aim”, “New jobs – the basis for social equality and
readiness for the future”, and “ From Germany’s edge to Europe’s middle – East
Germany needs a new beginning”, mixing a left-wing appeal with a regional
advocacy. It is running candidates in 13
single-member constituencies in the former West Germany, as well as 38 in the
former GDR; the party is expected to get party-list seats by exceeding the 5%
national hurdle and will almost certainly increase its current representation
of 37 in the Bundestag. On the basis of
public opinion polls, the PDS is predicted to gain about 6% of the German-wide
vote in the September 2002 federal elections (Die Welt, 26 June, 2002).
Berlin
is a city undergoing dramatic change, in its economic profile, in its streetscapes,
in its international linkages, and as a new national center of globalization in
Germany. Unlike other east European
countries, where former Communist parties were devastatingly tainted by their
behavior during the Cold War division of the continent, the PDS has managed to
re-create itself as a protector of regional interests and as the voice of those
who have been negatively affected by the incorporation of the socialist East
into the capitalist West. No other
former Communist country disappeared like the GDR so there cannot be any
certainty that a similar party could not have arisen elsewhere. Perhaps the closest parallel to the PDS is
the Lega Nord in Italy which has advocated the economic interests of the North
of the country in the aftermath of the shake-up of Italian politics in
1992. But Italy has not undergone the
massive economic and political shifts that have characterized the former
GDR. In Berlin, the post-Communist
changes have been dramatic and the contrasts greater than elsewhere because of
the proximity of the Western part of the city and the huge redevelopment
projects in the center.
Using
new methodologies of inferring individual behavior from aggregate data, we have
shown in this paper that the usual compositional factors that political
scientists rely on to understand electoral behavior are severely limited for
the PDS experience in contemporary Berlin.
The party’s appeal is highly focused on one part of the city and cuts
across all demographic groups in that region.
By contrast, its appeal in the West is tiny, though it is showing signs
of penetrating previously-strong areas of support for the Greens and for the
Social Democrats in the older and poorer parts of the West. To break out of its eastern bailiwicks, the
PDS will have to make further inroads into the poorer Western neighborhoods
where unemployment remains high. To accomplish this, it will have to take votes
away from its partner in the Berlin coalition government, the SPD. Much depends on the perception of the PDS’
performance while in power in the city.
Berlin is thus a test case of the future profile of the German electoral
scene and an intriguing example of the ability of a party to move from a
regional to the national stage.
__________________________________________________________________________________
Bibliography
Barker, Peter, ed. The
Party of Democratic Socialism in Germany. Atlanta, GA: Rodopi, 1998.
Bauer-Kaase,
Petra, “German
Unification: The Challenge of Coping With Unification,” in W. Donald Hancock
and Helga Walsh, eds., German
Unification: Processes and Outcomes. Boulder:
Westview Press, 1994, 285-313.
Bauer-Kaase,
Petra and Max Kaase,
“Five Years of Unification: The Germans on the Path to Inner Unity?” German Politics 5, 1: 1-25, April 1996.
Baylis,
Thomas A.,
“Institutional Destruction and Reconstruction in Eastern Germany,” in Patricia
Jo Smith, ed., After the Wall: Eastern
Germany Since 1989. Boulder, CO: Westview Press, 1998, 15-33.
Becker-Cantarino, Barbara, “Reflections on a Changing Berlin,” in Barbara
Becker- Cantarino, ed., Berlin in Focus: Cultural Transformations in
Germany. Westport, Connecticut; Praeger, 1996, 1-35.
Bolzen,
Stefanie, “Der
Antikriegkurs brachte der PDS die meisten Stimmen (The Anti-War Course Brought
Most Votes to the PDS).” Die Welt 2, 23
October 2001.
Bömermann, Hartmut, “Die Wahl zum Abgeordnetenhaus
von Berlin am 10. Oktober 1999: Ergebnisse der repräsentativen Wahlstatistik
(The Election of the Berlin House of Parliament on 10 October 1999: Results of
Representative Election Statistics)” Berliner
Statistik Monatschrift 1/01, 14-23, 2001.
Borneman,
John, Belonging
in the Two Berlins. Kin, State, Nation. New York: Cambridge University
Press, 1992.
Brie,
Michael, “Das
Politische Projekt PDS – eine unmögliche Möglichkeit (The PDS Political Project
– An Impossible Possibility),” in Michael Brie, Martin Herzig and Thomas Koch,
eds., Die PDS: Empirische Befunde und
kontroverse Analysen (The PDS:
Empirical Results and Controversial Analyses). Cologne: PapyRossa Verlag, 1995, 9-38.
Chambers, John M., William S. Cleveland, Beat Kleiner, and Paul
A. Tukey, Graphical Methods for Data Analysis. Belmont, CA: Duxbury Press, 1983.
Cleveland
William S., Visualizing Data. Summit, NJ: Hobart
Press, 1993.
Cochrane,
Allan and Andrew Jonas, “Re-imagining Berlin: World City, National Capital or Ordinary
Place?” European Urban and Regional
Studies 6, 2:145-164, April
1999.
Cochrane,
Allan and Adrian Passmore, “Building a National Capital in an Age of Globalization: The Case of
Berlin,” Area 33, 4: 341-352, October 2001.
Conradt,
Cooper,
Belinda, “The
Changing Face of Berlin,” World Policy
Journal 15, 3: 57-68, Fall 1998.
Dalton,
Russell J. and W. Bürklin, “The Two German Electorates,” in Russell J Dalton, ed., Germans Divided: The 1994 Bundestag
Election. Washington D.C.: Berg,
1996, 183-209.
Davies-Withers,
Suzanne, “Quantitative Methods: Advancement in Ecological
Inference,” Progress in Human Geography 25,
1:87-96, January 2001.
Deggerich, Markus, “Das PDS
–Problem der SPD (The PDS – Problem for the SPD),” Der Spiegel 21
October, 2001,
Doud,
Duke
Vic and Keith Grime,
“Inequality in Post-Communism,” Regional
Studies 31, 9: 883-890, December
1997.
Eith,
Ulrich, “New
Patterns in the East? Differences in
Voting Behavior and Consequences for Party Politics in Germany,” German
Politics and Society 18,
3:119-136, Fall 2000.
Elkins,
Thomas H. and Burkhard Hofmeister, Berlin: The Spatial Structure of a Divided
City.
London:
Methuen, 1988.
Ertman,
Thomas, “Eastern
Germany: Ten Years after Unification,” German Politics and Society 18, 3: 1-12, Fall 2000.
Evans,
Geoffrey and Stephen Whitefield, “Economic Ideology and Political Success:
Communist-Successor Parties in the Czech Republic, Slovakia and Hungary
Compared,” Party Politics 1, 4:565-578, October 1995.
Fotheringham,
A. Stewart, “A Bluffer’s Guide to ‘A Solution to the Ecological
Inference Problem’,” Annals, Association
of American Geographers 90,
3:582-586, September 2000.
Gibowski,
Wolfgang G.,
“Germany’s General Election in 1994: Who Voted for Whom,” in
Higley,
John, Judith Kullberg, and Jan Pakulski, “The Persistence of Post-Communist Elites,” Journal of Democracy 7,
2:133-147, April 1996.
Hough,
Daniel, “’Made
in Eastern Germany’: The PDS and the Articulation of Eastern German Interests,”
German Politics 9, 2: 125-148,
August 2000a.
Hough,
Daniel,
“Societal Transformation and the Creation of a Regional Party: The PDS as a
Regional Actor in Eastern Germany,” Space and Polity 4, 1: 57-75, May 2000b.
Howard,
Marc M., “An
East German Ethnicity? Understanding the New Division of Unified Germany,” German Politics and Society 13, 4: 49-70, Winter 1995.
Ishiyama,
John T.,
“Discussion and Conclusions,” in John T. Ishiyama, ed., Communist Successor Parties in Post-Communist Politics. New York:
Nova Scotia Publisher Inc, 1999a, 299-334.
Ishiyama,
John T.,
“Introduction and Theoretical Framework,” in John T. Ishiyama, ed., Communist Successor Parties in
Post-Communist Politics. New York: Nova Scotia Publisher Inc, 1999b, 1-12.
Ishiyama,
John T., “What
Kinds of Parties are Emerging? Patterns of Successor Parties Organizational
Development,” in John T. Ishiyama, ed., Communist
Successor Parties in Post-Communist Politics. New York: Nova Scotia
Publisher Inc, 1999c, 131-151.
Johnston,
Ron, The Geography of English
Politics- The 1983 General Election. London: Croom Helm, 1985.
Johnston, Ron, Review of Gary King, A Solution to the Ecological
Inference Problem: Reconstructing Individual Behavior from Aggregate Data. Environment
and Planning A 30, 7: 1323-24, July 1998.
Johnston, Ron and Charles J. Pattie, “Ecological Inference and
Entropy-Maximizing: An Alternative Estimation Procedure for Split-Ticket
Voting,” Political Analysis 8, 4:333-45, Fall 2000.
Johnston, Ron and Charles J.
Pattie, “On
Estimates of Split-Ticket Voting” (Available from http://www.people.fas.harvard.edu/~burden/papers.html),
2001.
King, Gary, A Solution to the
Ecological Inference Problem: Reconstructing Individual Behavior from Aggregate
Data. Princeton, NJ: Princeton
University Press, 1997.
Kitschelt, Herbert, “The Formation of Party Systems
in East Central Europe,” Politics and
Society 20, 1:7-50, June 1992.
Kitschelt,
Herbert, The Transformation of European Social
Democracy. Cambridge University Press: Cambridge, 1994.
Kitschelt,
Herbert,
“Formation of Party Cleavages in Post-Communist Democracies,” Party Politics 1, 4: 447-472, October 1995.
Kitschelt, Herbert, Dmitri Dimitrov, and Kanev, A., “The Structuring of the Vote in Post-Communist
Party Systems: The Bulgarian Example,”
European Journal of Political Research 27,
2:143-160, February 1995.
Kleinfeld,
Gerald R., “The
Return of the PDS,” in
Kleinfeld,
Gerald R., “The
PDS: An Unexpected Intruder from the East,” in Peter H. Merkl, ed., The Federal Republic of Germany at Fifty. New York, NY: New York University Press,
1999, 160-173.
Krätke,
Stefan, “Berlin:
Towards a Global City?” Urban Studies 38, 10:1777-1799, September 2001.
Krisch,
Henry, “The
Party of Democratic Socialism: Left and East,” in Russell J. Dalton ed., Germans Divided: The 1994 Bundestag Election.
Washington D.C.:Berg, 1996, 109-133.
Kulick, Holger, “Gysi feiert 22.6 Prozent –
‘Das ist ein klarer Mitregierungsauftrag’ (Gysi Celebrates 22.6Percent – ‘That
is a Clear Direction for Government,’” Der Spiegel, October 21, 2001.
Kunze, Reiner, “Die
Mauer (The Wall),” in Reiner Kunze, Ein
Tag auf dieser Erde: Gedichte (A Day on this Earth: Poems),
Frankfurt-am-Main: Fischer Verlag, 1998.
Langguth,
Gerd, In Search of Security: A Socio-Psychological
Portrait of Today’s Germany. Westport,
CT: Praeger, 1995.
Lang, Joachim, Patrick Moreau and Viola Neu, Auferstanden aus Ruinen? Die PDS nach dem Super-Wahljahr 1994 (Resurrection from the Ruins: The
PDS after the Super Election Year 1994), Bonn: Konrad Adenauer Stiftung,
1995.
Larres,
Klaus “Germany
1989: The Development of a Revolution,” in Klaus Larres, ed., Germany Since
Unification, New York, NY: MacMillan, 1998, 33-62.
Lawson,
Sue, “Berlin:
The New Capital,” in Peter James, ed., Modern
Germany: Politics, Society and Culture.
Routledge: London, 1998, 13-30.
Leiby,
Richard A. The
Unification of Germany1989-1990. Westport, CT:
Greenwood Press, 1999.
Mahr,
Alison and John D. Nagle, “Resurrection of the Successor Parties and Democratization in
East-Central Europe,” Communist and
Post-Communist Studies 28,
4:393-409, December 1995.
MacDonagh, Giles, Berlin. New York, NY: St. Martin’s Press, 1998.
McFalls,
Laurence,
“Eastern Germany Transformed: From Postcommunist to Late Capitalist Political
Culture,” German Politics and Society
17, 2:1-25, Summer 1999.
McKay,
Joanna, “The
Wall in the Ballot Box: The Berlin Election of 1995,” German Politics 5,
2: 276-291, August 1996.
McKay,
Joanna, “Keeping
the Red Flag Flying: The Electoral success of the PDS in East Berlin,” Journal
of Communist Studies and Transition Politics 16, 3:1-20, September 2000.
Moreau, Patrick, Die PDS: Profil einer
antidemoktratischen Partei. Munich: Hanns-Seidel-Siftung, 1998.
Mushaben,
Joyce Marie,
“Auferstanden aus Ruinen: Social Capital and Democratic Identity in the New
Länder,” German Politics and Society 15, 4: 43-58, Winter 1997.
Neugebauer, Gero and Richard Stöss,
Die PDS: Geschichte. Organisation.
Wähler. Konkurrenten. Opladen: Leske
und Budrich, 1996.
O’Loughlin,
John “Can King’s Ecological Inference Method Answer a Social Scientific
Puzzle: Who Voted for the Nazi Party in Weimar Germany?” Annals, Association of American Geographers 90, 3: 592-601, September 2000.
O’Loughlin,
John “The Regional Factor in Contemporary Ukrainian Politics: Scale, Place,
Space, or Bogus Effect,” Post-Soviet
Geography and Economics 42, 1:
1-33, January-February 2001.
O’Loughlin, John. “The Electoral Geography of Weimar Germany: Exploratory Spatial Data Analyses (ESDA) of Protestant Support for the Nazi Party” Political Analysis 10, 3: Summer 2002 (forthcoming).
O’Loughlin,
John,
Oswald,
Franz, “The
Party of Democratic Socialism: Ex-Communists Entrenched as East German Regional
Protest Party,” Journal of Communist
Studies and Transition Politics 12,
2:73-195, June 1996.
Pammett
Jon H. and Joan DeBardeleben, “The Meaning of Elections in Transitional Democracies:
Evidence from Russia and Ukraine,” Electoral Studies 15, 3: 363-381, September 1996.
Parkes,
Stuart, Understanding Contemporary Germany.
London: Routledge, 1997.
Patton,
Phillips,
Ann L., The Third Victor: the PDS after the
Elections. Washington, DC: American
University, AICGS Ford Foundation Fellow in GDR Studies, 1988-1989, 1999 ( http://www.aicgs.org/wahlen/elect98.html).
Rattinger, Hans, Parteiidentifikation in Ost- und West-Deutschland (Party Identification in East and
West Germany). Berlin: Akademie Verlag, 1994.
Rattinger,
Hans, and Joel Kraemer, “Economic Conditions and Voting Preferences in East and West
Germany, 1989-1994,” in Carsten Zelle and Christopher Anderson, eds., Stability and Change in German Elections:
How Electorates Merge, Converge or Collide.
Westport, Ct: Praeger, 1998, 99-121.
Read,
Anthony, and
Roesler,
Jörg, “The Party
of Democratic Socialism in the German Unification Process,” in Patricia Jo
Smith, ed., After the Wall: Eastern
Germany Since 1989, Boulder, CO: Westview Press, 1998, 51-75.
Roth,
Dieter, Examining the Vote: The Results of Germany’s
National Election- Germany in the
Year of Decision. Baltimore, MD: American Institute for Contemporary
Germany Studies Seminar at John Hopkins University, 1999.
Schmitt, K., “Im Osten
Nichts Neues? Das Kernland der Deutschen Arbeiterbewegung und did Zukunft der
Politischen Linken (Nothing New in the East? The Core Region of the German Workers’ Movement
and the Future of the Political Left),” in Wilhelm Bürklin and Dieter Roth,
eds., Das Superwahljahr, Cologne:
Bund Verlag, 1995.
Segert,
Astrid,
“Problematic Normalization: East German Workers Eight Years after Unification,”
German Politics and Society 16, 3:105-123, Fall 1998.
Staab,
Andreas, National Identity in Eastern Germany: Inner
Unification or Continued Separation?
Westport, CT: Praeger, 1998.
Statistiches Landesamt Berlin, Statistisches
Jahrbuch 1996, Berlin:
Statistisches Landesamt, 1996.
Statistisches Landesamt Berlin, “Wahl zum
Abgeordnetenhaus von Berlin am 10. Oktober 1999: Ergebnisse der
repräsentativen Wahlstatistik (The Election of the Berlin House of Parliament
on 10 October 1999: Results of Representative Election Statistics)” Statistischer
Bericht B VII 2-99, 1999.
Strom,
Elizabeth and Margit Mayer, “The New Berlin,” German
Politics and Society 16,
4:122-139, Winter 1998.
Venables,
William N. and Brian D. Ripley, Modern Applied Statistics with
S-Plus, 2nd
edition. Berlin:
Springer Verlag, 1997.
Von Ditfurth, C. Ostalgie oder linke Alternative: Meine Reise durch die PDS (Ostalgia or
Left Alternative: My Journey through the PDS). Cologne: Kiepenheuer &
Witsch, 1998.
Waller,
Michael,
“Adaptation of the Former Communist Parties of East Central Europe: A Case for
Democratization,” Party Politics 1, 4: 373-390, October 1995.
Weber,
Torsten.
“Unternehmer warnen vor Rot-Rot: Wirtschaft zieht Ampelkoalition vor
(Entrepreneurs Warn about a Red –Red Coalition: Business Prefers a Traffic
Light Coalition),” Die Welt, October 23, 2001.
Welt, Die, “Welt-Umfrage: Union legt zu,
SPD verliert, (Welt-Poll: Union is rising, SPD is losing)” 26 June 2002.
White,
Paul and Daniel Gutting, “Berlin: Social Convergences and Contrasts in the Reunited City,” Geography 83,
3: 214-226, July 1998.
Wiesenthal,
Helmut, “East
Germany as a Unique Case of Societal Transformation: Main Characteristics and
Emergent Misconceptions,” in Andreas Pickel and Helmut Wiesenthal, eds., The Grand Experiment: Debating Shock
Therapy, Transition Theory, and the East German Experience, Boulder, CO: Westview Press, 1997, 45-72.
Wittich,
Dietmar,
“Mitglieder und Wähler der PDS (Members and Voters of the PDS)” in Michael
Brie, Martin Herzig und Thomas Koch, eds.,
Die PDS: Empirische Befunde und kontroverse Analysen (The PDS: Empirical Results and
Controversial Analyses). Cologne: PapyRossa Verlag, 1995
Zelle,
Carsten, “Socialist Heritage or Current Unemployment:
Why Evaluations of Democracy and Socialism differ Between the Germans in the
East and West?” Paper presented at the Conference of the German Studies
Association, Washington DC, 1997.
Zelle,
Carsten,
“Factors Explaining the Increase in PDS Support After Unification,” in Carsten
Zelle and Christopher Anderson eds., Stability
and Change in German Elections: How Electorates Merge, Converge or Collide, Westport, CT:
Praeger, 1998, 223-247.
Ziblatt,
Daniel F.,
“Putting Humpty Dumpty Back Together Again: Communism’s Collapse and the
Reconstruction of the East German Ex-Communist Party,” German Politics and
Society, 16, 1: 1-29, Fall 1998.
Zimmer,
Matthias,
“Germany Divided, Germany Unified: The National Question, 1945-1990,” in
Matthais Zimmer , ed., Germany: Phoenix
in Trouble?, Edmonton, Alberta: University of Alberta Press, 1997, 57-83.
[1] The formula for Morans I is Morans I is
derived from I = (N/So)Si Sj wij xi xj
/ Si xi2 where wij is an element of a spatial weights
matrix W that indicates whether or not i and j are contiguous; the spatial
weights matrix is row-standardized such that its elements sum to 1; xi is
an observation at location i (expressed as the deviations from the observation
mean); and So is a normalizing factor equal to the sum of all
weights (Si Sj wij). The significance of Moran’s I is assessed by
a standardized z-score that follows a normal distribution and is computed by
subtracting the theoretical mean from I and dividing the remainder by the
standard deviation.
[2] The statistic is derived from = Si wij(d) yj / Sj yj where wij(d) is an element in a
binary contiguity matrix (not row-standardized) and yj is an observation at location j. The statistics should be
interpreted as a measure of like values around a particular observation. The
significance of the index can be assessed by standard z-scores. A positive
z-value for the statistic at a particular location implies spatial clustering
of high values around that location; a negative value indicates a spatial
grouping of low values.