Holger Lutz Kern & Jens Hainmueller - Opium For The Masses: How Foreign Media Can Stabilize Authoritarian Regimes

Advance Access publication July 30, 2009 Political Analysis (2009) 17:377–399
doi:10.1093/pan/mpp017
Opium for the Masses: How Foreign Media Can

Stabilize Authoritarian Regimes
Holger Lutz Kern

Institution for Social and Policy Studies, Yale University, New Haven, CT 06511
e-mail: [email protected] (corresponding author)
Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Jens Hainmueller
Department of Political Science, Massachusetts Institute of Technology, Harvard University,
Cambridge, MA 02139 e-mail: [email protected]
In this case study of the impact of West German television on public support for the East
German communist regime, we evaluate the conventional wisdom in the democratization
literature that foreign mass media undermine authoritarian rule. We exploit formerly classified
survey data and a natural experiment to identify the effect of foreign media exposure using
instrumental variable estimators. Contrary to conventional wisdom, East Germans exposed
to West German television were more satisfied with life in East Germany and more
supportive of the East German regime. To explain this surprising finding, we show that East
Germans used West German television primarily as a source of entertainment. Behavioral
data on regional patterns in exit visa applications and archival evidence on the reaction of
the East German regime to the availability of West German television corroborate this result.
1 Introduction
One of the most prominent subjects in comparative politics during the past few decades
has been the explanation of political regime change from authoritarian rule to democracy.
Beginning with the pathbreaking work of O’Donnell, Schmitter, and Whitehead (1986),
many authors have developed complex theoretical models of the determinants of such tran-
sitions. The early literature on democratization concluded that international factors played
no more than a secondary role in transitions to democracy, but more recent theoretical and
empirical work has put greater emphasis on the international context (Huntington 1991,
Pridham 1991, 1997; Starr 1991; Gasiorowski 1995; Linz and Stepan 1996; Whitehead
Authors’ note: This is one of several joint papers by the authors; the ordering of names reflects a principle of
rotation. Both authors contributed equally to this paper. Software to estimate local average response functions is
available upon request. We thank Alberto Abadie, Christopher Anderson, Jake Bowers, Daniel Butler, Alexis
Diamond, Andy Eggers, Justin Grimmer, Dominik Hangartner, Dan Hopkins, Guido Imbens, Lutz Kern, Jan
Lemnitzer, Walter Mebane, Na’ama Nagar, Beth Simmons, Hans-Jörg Stiehler, Susan Stokes, Christopher
Way, Robert Weiner, and seminar participants at Harvard, MIT, Dartmouth, Cornell, Northwestern, UC Davis,
and the University of Berne for very helpful comments. Brigitte Freudenberg at the Office of the Federal Com-
missioner for the Records of the National Security Service of the Former German Democratic Republic was of
great help in tracking down archival material. Peter Bischoff, Evelyn Brislinger, Kurt Starke, and especially
Hans-Jörg Stiehler patiently answered our many questions about the surveys conducted by the Central Institute
for Youth Research. The usual disclaimer applies.
Ó The Author 2009. Published by Oxford University Press on behalf of the Society for Political Methodology.
All rights reserved. For Permissions, please email: [email protected]
377
378 Holger Lutz Kern and Jens Hainmueller
1996; O’Loughlin et al. 1998; Kopstein and Reilly 2000; Starr and Lindborg 2003;
Doorenspleet 2004; Brinks and Coppedge 2006; Gleditsch and Ward 2006). In the wake
of this theoretical reorientation, foreign mass media have attracted increased attention as
a cause of democratization. The view that Western mass media made a significant contri-
bution to the fall of communism in Eastern Europe is widely shared among scholars and
policymakers (Giddens 2000; Nye 2004, 2008; Parta 2007). What makes the wide accep-
tance of this view puzzling is that empirical research on the impact of foreign mass media on
the stability of authoritarian regimes is almost nonexistent. Historical case studies of par-
ticular media sources such as Radio Liberty and Radio Free Europe (Lisann 1975; Nelson
1997; Puddington 2000) are suggestive but do not provide systematic evidence about their
impact. A few studies use survey data from authoritarian regimes to document a correlation

between exposure to foreign mass media and pro-democratic and pro-Western attitudes but
fall short of demonstrating that these correlations are indeed causal (Hesse 1988, 1990; Parta
2007). However, a clear distinction between correlation and causation is particularly crucial
in studies of media effects since it is well established that individuals tend to expose them-
selves to political messages they agree with (Bartels 1993; Stroud 2008).
In this paper, we address the question of the impact of foreign mass media on public
support for authoritarian regimes by exploiting a natural experiment in communist East
Germany. Because of East Germany’s topography, West German television broadcasts
could be received in most but not all parts of the country. We take advantage of this nat-
urally occurring variation to identify the causal effect of West German television exposure
on support for the East German regime. We rely on formerly classified survey data, be-
havioral data on exit visa applications, and archival evidence on the reaction of the East
German leadership to the availability of West German television. Our main result is that
exposure to West German television increased support for the East German regime. We
argue that the best explanation for this counterintuitive finding is that East Germans used
West German television primarily as a source of entertainment. In a society with a very
limited number of entertainment options, the ability to watch West German television made
a real difference in people’s daily lives. It offered them a vicarious escape from the scar-
cities, the queues, and the ideological indoctrination, making life under communism more
bearable and the East German regime more tolerable. We do not necessarily argue that
West German television’s political content did not undermine public support for the East
German regime at all. However, the evidence shows that the net effect of West German
television exposure was an increase in regime support. Behavioral data on regional patterns
in exit visa applications and archival evidence on the reaction of the East German leader-
ship to the availability of West German television corroborate these results.
2 The Effects of Foreign Media Exposure

During the Cold War, the United States attempted to reach audiences in communist
countries through radio stations such as Voice of America, Radio Liberty, and Radio Free
Europe. Some of these radio stations were explicitly set up as ‘‘surrogate’’ stations. They
specialized in emulating the domestic stations that countries behind the Iron Curtain would
have had if they had been free. U.S. international broadcasting policies were based on the
expectation that access to uncensored and trustworthy information would nurture a more
pro-democratic and pro-Western public opinion, restrain communist militarism and adven-
turism, and erode public support for communist rule in the long run (Presidential Study
Commission on International Radio Broadcasting 1973; Lisann 1975; Quester 1990;
Nelson 1997; Puddington 2000).
Opium for the Masses 379
After the end of the Cold War, both scholars and policymakers agreed that U.S. inter-
national broadcasting, together with radio stations sponsored by other Western countries
(e.g., Deutsche Welle and the BBC), had made a significant contribution to the demise
of communism in Eastern Europe (Giddens 2000; Nye 2004, 2008; Parta 2007). The de-
mocratization literature holds that Western broadcasts sapped the strength of communist
regimes by giving people behind the Iron Curtain hope and the assurance that the Free
World had not forgotten them. They provided information not available in the state-
controlled domestic media, thus allowing Eastern Europeans to compare communist
propaganda with credible information from abroad. Western broadcasts addressed issues
suppressed in the domestic media such as the existence of dissident movements, human
rights violations, and communist countries’ involvement in foreign wars. They enabled

Eastern Europeans to compare their living standards with the (generally much higher)
standards of living in supposedly declining capitalist countries. Through their coverage
of domestic politics in democratic nations, they familiarized listeners with the functioning
of democracy and introduced them to freedoms and liberties unknown under communism.
In the long run, Western broadcasting nurtured pro-democratic attitudes and undermined
public support for communism (Rustow 1990; Diamond 1993a, 1993b; Dalton 1994;
Rohrschneider 1994, 1996; Whitehead 1996; Bennett 1998; Roberts 1999; Sükösd 2000).
The literature often cites the availability of West German television in East Germany as
a particularly vivid example of the democratizing force of foreign mass media. Rustow
(1990) for instance notes that West German television invited a constant comparison of
living standards between affluent West Germany and its much poorer East German neigh-
bor, thereby undermining the political legitimacy of the East German regime. According to
Quester (1990: 128), the availability of West German television posed a ‘‘constant threat of
destabilization and ferment.’’ Roberts (1999: 33) also draws a direct connection between
West German television and the fall of the East German regime, noting that ‘‘the avail-
ability of West German television in East Germany influenced attitudes over a long period
[. . .], compelling a reluctant regime to initiate change.’’ For Whitehead (1996: 6), ‘‘popular
attitudes in East Germany were so powerfully influenced by messages transmitted neutrally
from the West that democratization became unavoidable, whatever governments or polit-
ical leaders within or without might have wished or attempted.’’ Rohrschneider (1994: 935)
asserts that access to West German television ‘‘undoubtedly helped to erode the legitimacy
of the East German economic and political system.’’
West German television was indeed quite popular in East Germany; it was in fact much
more popular than East German television (Stiehler 2001).1 East and West Germans shared
the same language and very similar cultures, which should have maximized the potential
impact of West German broadcasts.2 West German television also devoted much attention
to politics in East Germany (e.g., political magazines such as Kontraste and Kennzeichen D).
There can be no doubt that because of the availability of West German television, most
East Germans had access to political information that the East German regime would have
preferred them not to have. All in all, the claim that West German television undermined
1
West German television refers to ARD and ZDF, the two primary West German public broadcasting stations.
Commercial television was introduced in West Germany in the mid-1980s but did not broadcast to East Germany
and could only be received in some areas near West Berlin. Since East Germany bordered Poland and Czecho-
slovakia, some East Germans could have watched television broadcasts originating in these countries. We feel
justified in ignoring this complication here since the communist orientation of these stations together with lan-
guage barriers severely limited their appeal.
2
Almost all East German households owned at least one television set in the 1980s (Staatliche Zentralverwaltung
für Statistik 1989).
public support for the East German regime is very plausible, but it has not been tested
empirically so far.
In its confidence in the power of foreign mass media to undermine authoritarian
stability, the democratization literature also fails to consider the possibility that foreign
media might inadvertently stabilize authoritarian regimes. West German television not
only informed its viewers about the grim reality of communist rule in Eastern Europe,
it also threw light on controversial aspects of capitalist societies such as crime, drug abuse,
mass unemployment, and the unequal distribution of wealth. German communications
scholars have therefore conjectured that West German television had a deterring effect
on East Germans, raising their awareness of the ‘‘dark sides’’ of capitalism and increasing
their appreciation for the advantages of ‘‘real existing socialism’’ (Hesse 1988: 118–22;

Meyen 2003b: 67–8). East Germans could not buy the consumer goods or travel to the
exotic places they encountered on West German television, but they also did not have
to worry about unemployment or homelessness. Confronted with social ills that did not
exist in East Germany, they might have come to the conclusion that the grass was not much
greener on the other side of the Berlin Wall.
Alternatively, people living under authoritarian regimes might simply see the mass me-
dia, both domestic and foreign, as a source of entertainment that offers a temporary escape
from the humdrum of daily life, the scarcities, the queues, and the ideological indoctri-
nation (Lazarsfeld and Merton 1948; Katz and Foulkes 1962). Especially Meyen
(2001, 2003b: 37–42) has stressed the escapism aspect of West German television. As
in other communist countries, life in East Germany was rather dull and uneventful. A scar-
city of restaurants, cinemas, theaters, and night and sports clubs contributed to domesticity
and boosted West German television’s importance as the primary source of entertainment.
As one German historian has put it, each night East Germans ‘‘collectively emigrated’’ to
West Germany in front of their television screens (Wolle 1998: 71). West German televi-
sion, and in particular its entertainment programming, may have offered East Germans
a vicarious escape that made life under communism more bearable and the East German
regime more tolerable.
Below, we test the conventional view that West German television exposure undermined
support for the East German regime together with the deterrence and escapism hypotheses.
The next section of the paper describes our research design and statistical methodology.
Section 4 presents the main results. Section V shows that access to West German television
was also related to the number of exit visa applications filed by East Germans, which serve
as a behavioral indicator of political dissatisfaction. Based on research in the archives of
the East German secret police and the ruling Socialist Unity Party (SED), Section VI shows
that the East German leadership itself was aware of the stabilizing effects of West German
television. The last section discusses our results.
3 Research Design
One challenge for research on media effects in authoritarian regimes is data availability.
Authoritarian regimes rarely permit independent survey research and when they them-
selves conduct public opinion surveys the results are often not made public. Thus, with
some exceptions (Geddes and Zaller 1989), reliable microlevel data on public opinion
in authoritarian regimes are not readily available. Past research on media effects in Eastern
Europe has attempted to address this problem by interviewing emigrees or visitors to
Western Europe. Hesse (1988, 1990), for example, interviewed refugees coming to West
Germany. Radio Liberty and Radio Free Europe interviewed Eastern Europeans visiting
the West to estimate the effects of exposure to Western radio programs on political attitudes
in Eastern Europe and the Soviet Union (Parta 2007). However, such surveys do not address
the problem of self-selection into media exposure. Soviet citizens who listened to Radio
Liberty almost certainly had different political attitudes than nonlisteners to begin with.
To attribute differences in political attitudes between these two groups to Radio Liberty
exposure is therefore problematic.
We rely on formerly classified survey data collected by the Zentralinstitut für
Jugendforschung (Central Institute for Youth Research) that have become available to
researchers after German reunification. The Central Institute was founded in 1966 to offer
scientific guidance to East German authorities on questions of youth policy. Its critical
reports aroused the hostility of parts of the East German bureaucracy and during most

of its existence, researchers affiliated with the Central Institute were not allowed to publish
any of their research. Between 1966 and 1990, the Central Institute conducted several hun-
dred surveys of East German high school and college students, apprentices, and young
workers. For each survey, questionnaires had to be approved by SED party officials, several
government departments, and the Central Office for Statistics. Surveys representative of
East Germany as a whole could not be conducted for political reasons; the only stratum of
the East German population the Central Institute was allowed to survey were teenagers and
young adults. The range of permissible survey questions was likewise restricted. Given
these constraints, the Central Institute relied on cluster sampling of teenagers and young
adults in schools, universities, and firms (Friedrich, Förster, and Starke 1999).
Researchers at the Central Institute were aware that people living under an authoritarian
regime might be reluctant to participate in public opinion surveys and answer politically
sensitive questions. To convince participants that survey responses would be completely
anonymous the researchers eschewed face-to-face interviews. They always distributed
questionnaires in group settings (e.g., high school or college lecture classes), thereby
giving respondents the opportunity to see for themselves that the questionnaires were un-
marked and could not be traced back to them. Completed questionnaires were collected in
a sealed urn (Friedrich 1990). Ex post, it is impossible to determine with certainty to what
extent these procedures eliminated preference falsification (Kuran 1991). Nonetheless, as
we will show below, survey responses were fairly critical of the East German regime.
Moreover, even if the existence of preference falsification cannot be completely ruled
out, it almost certainly does not create a problem for our research design. We are not pri-
marily concerned with identifying the true level of East Germans’ regime support. Instead,
we estimate the difference in regime support between East Germans who were exposed to
West German television and East Germans who were not exposed to West German tele-
vision. Unless levels of preference falsification varied systematically between these two
groups, our media effect estimates will be unbiased.
Individuals tend to expose themselves to political messages they agree with, making it
difficult to distinguish between association and causation in studies of media effects
(Bartels 1993; Stroud 2008). If exposure to West German television is associated with
critical attitudes toward the East German regime, it could be that West German television
affected viewers’ attitudes but it is equally plausible that East Germans dissatisfied with the
regime were more likely to watch West German television in the first place. To address this
self-selection problem we take advantage of a natural experiment. Natural experiments are
observational studies in which some exogenous process (‘‘nature’’) assigns units to dif-
ferent types of treatments in a haphazard fashion that is ‘‘as good as random’’ (Freedman
2005; Morgan and Winship 2007; Dunning 2008). We identify the effects of exposure to
West German television by exploiting a natural experiment generated by East Germany’s

Fig. 1 Over-the-air signal strength of West German television broadcasts. Source: Adapted from
Die Welt, October 27, 1980.
topography. As it turns out, West German over-the-air television broadcasts could not be
received in all parts of East Germany. Especially the Dresden district in the Southeast was
largely cut off from West German television due to topographical features and its distance
from West German broadcasting towers (Fig. 1).
3.1 Survey Data

The survey data were collected between November 1988 and February 1989, less than
1 year before the fall of the Berlin Wall in November 1989.3 It is important to note that
at that time no one expected the fall of the East German regime, which appeared to be one
of the most stable communist regimes in Eastern Europe (Zelikow and Rice 1995). The
sample contains mostly teenagers and young adults (the mean age is 23 years; the lower
and upper quartiles are 18 and 26 years) from eight East German districts (Dresden, East
3
Zentralarchiv für Empirische Sozialforschung ZA 6008: Politisches Klima und gesellschaftliche Bedingungen in
der DDR 1988/89.
Berlin, Magdeburg, Cottbus, Leipzig, Erfurt, Karl-Marx-Stadt, and Schwerin). The total
number of respondents is N 5 3564.
3.2 Causal Inference Using Instrumental Variables

Our main quantity of interest is the effect of West German television exposure on expressed
support for the East German regime. We code a binary variable D that takes the value 1 for
respondents who watch West German television daily, multiple times each week, once per
week, or less than once per week, and 0 for respondents who never watch West German
television.4 Borrowing from the literature on causal inference in statistics (Rubin 1974,
1978; Holland 1986; Rosenbaum 2002), we refer to D as ‘‘treatment indicator,’’ the set

of respondents exposed to West German television as ‘‘treatment group,’’ and the set
of respondents not exposed to West German television as ‘‘control group.’’ We define
Y1 and Y0 as the potential outcomes under treatment and control, that is, the levels of ex-
pressed regime support that a respondent would have had with and without exposure to
West German television. For each respondent, the effect of West German television ex-
posure is defined as the difference between these two potential outcomes (Y1 2 Y0).
The ‘‘fundamental problem of causal inference’’ (Holland 1986) is that for each respondent
we never get to observe both potential outcomes but only the realized outcome Y 5 DY1 1
(1 2 D)Y0. In other words, for a respondent exposed to West German television, we never
get to observe the counterfactual level of regime support that she would have had in the
absence of exposure to West German television (and vice versa).
To obtain unbiased treatment effect estimates, we need to find a suitable control group of
respondents that is sufficiently similar to the treatment group in all relevant characteristics
except that it was not exposed to West German television. Finding a good control group is
difficult in observational studies because selection into treatment is usually associated with
the potential outcomes. It is for this reason that the naive comparison of the regime support
of East Germans exposed to West German television with the regime support of East
Germans not exposed to West German television would not be very informative about
the causal effect of West German television exposure. We could attempt to make this com-
parison more credible by controlling for observable characteristics known to affect both
selection into West German television exposure and political attitudes using some method
of covariate adjustment (e.g., regression or matching). Such an approach would still be
problematic, however, since we cannot rule out the possibility that East Germans selected
into West German television exposure based on unobservable characteristics.
Instrumental variables (IV) (Imbens and Angrist 1994; Angrist, Imbens, and Rubin
1996; Abadie 2003; Imbens and Wooldridge 2008) offer a more credible identification
strategy by exploiting the fact that not all parts of East Germany had access to West
German television. We code a binary instrument Z based on information about respond-
ents’ place of work. It takes the value 0 for respondents living in the Dresden district and
the value 1 for respondents living in districts with access to West German television
4
Respondents were asked ‘‘How often do you watch West German television? ’’ Responses are coded on a five-
category scale ranging from daily to never. We dichotomize this variable to facilitate the interpretation of our
results. One could also think of West German television exposure as a multivalued treatment, but that would make
identification much more complicated (see Imbens and Wooldridge 2008 for details). Moreover, note that East
Germany’s topography induced variation in whether East Germans watched West German television but not
necessarily in the specific amount they watched. In other words, West German television exposure might only
be ‘‘as good as random’’ when dichotomized. Results are substantively identical when respondents watching
West German television less than once per week are also coded as 0.
(East Berlin, Magdeburg, Cottbus, Leipzig, Erfurt, Karl-Marx-Stadt, and Schwerin). One
can think of this instrument as inducing exogenous variation in the treatment.5
We follow Imbens and Angrist (1994) in conceptualizing IV identification in terms of
potential treatment indicators. Let Dz represent potential treatment status given Z 5 z. For
example, D0 5 0 and D1 5 1 means that a respondent would not watch West German
television if she lived in the Dresden district but that she would watch West German tele-
vision if she did not live in the Dresden district. The treatment status indicator can then be
expressed as D 5 ZD1 1 (1 2 Z)D0. Similar to the missing data problem for potential
outcomes, we only get to observe Z and D (and therefore Dz for individuals with Z 5 z) but
never both potential treatment indicators for the same individual. Following the terminol-
ogy in Angrist, Imbens, and Rubin (1996), we can distinguish four groups of respondents:

Compliers: D0 5 0 and D1 5 1. Compliers are respondents who watch West German
television if they do not live in Dresden but who do not watch West German television if
they live in Dresden.
Always-takers: D1 5 D0 5 1. These are respondents who always watch West German
television, no matter where they live.
Never-takers: D1 5 D0 5 0. Similarly, these are respondents who never watch West
German television, no matter where they live.
Defiers: D0 5 1 and D1 5 0. These are respondents who watch West German television
if they live in Dresden but who do not watch West German television if they do not live in
Dresden.
Since we only observe one of the potential treatment indicators (D0, D1), we cannot
directly identify the group to which any particular respondent belongs. However, it turns
out that under certain assumptions that we will discuss next, IVallow us to estimate average
treatment effects for the subgroup of compliers (62% of our respondents) even though we
cannot individually identify compliers in our sample.
3.2.1 Identification assumptions

Let Yzd represent the potential outcome if Z 5 z and D 5 d. For example, Y00 denotes the
regime support of a respondent who lives in the Dresden district (Z 5 0) and does not watch
West German television (D 5 0). Following Abadie (2003), this leads to the following
nonparametric assumptions under which IV can be used to identify causal effects. X
represents a vector of predetermined covariates.
(i) Ignorability of the instrument: conditional on X, the random vector (Y00, Y01, Y10,
Y11, D0, D1) is independent of Z.
(ii) Exclusion of the instrument: P(Y1d 5 Y0d|X) 5 1 for D 2 {0, 1}.
(iii) First stage: 0 < P(Z 5 1|X) < 1 and P(D1 5 1|X) > P(D0 5 1|X).
(iv) Monotonicity: P(D1 > D0|X) 5 1.
Are these identification assumptions plausible? Assumption (iv) rules out the existence of
defiers. It is highly unlikely that there were East Germans who would have watched West
German television if they had lived in Dresden but who would not have watched West
German television if they had not lived in Dresden. Assumption (iii) is also innocuous.
5
Could we have used a regression-discontinuity design (RDD) instead? RDDs are sometimes used with spatial
treatment discontinuities (Black 1999). However, our survey data do not allow us to implement a spatial RDD
because doing so would require information about respondents’ place of residence at a much more disaggregate
level.
Table 1 West German television exposure and place of residence
Living in
How often do you watch Dresden district Other districts

West German television? Count (%) Count (%)
Daily 42 (5) 1820 (65)

Multiple times each week 69 (9) 732 (26)
Once per week 17 (2) 47 (2)
Less than once per week 108 (14) 101 (4)
Never 498 (64) 46 (2)
Missing 50 (6) 33 (1)

Total 784 (100) 2779 (100)
It guarantees that Z and D are correlated conditional on X. Given that most people in the
Dresden district were cut off from West German television broadcasts, not living in the
Dresden district is highly correlated with exposure to West German television. Table 1
displays the frequency of West German television consumption in our sample. As one
would expect, respondents living in the Dresden district were much less likely to watch
West German television than respondents living in other districts. The sample correlation
between living in Dresden (Z) and consumption of West German television (D) is 0.74.
When we regress Z on D while controlling for our extensive set of covariates (see below),
the t-statistic on the treatment indicator is about 60.
Assumptions (i) and (ii) need more justification. Assumption (i) states that place of
residence (Z) is ‘‘as good as randomly assigned’’ once we condition on X. Assumption
(ii) implies that variation in the instrument does not change the potential outcomes other
than through its effect on D. It allows us to define potential outcomes in terms of D alone, so
that we have Y0 5 Y00 5 Y10 and Y1 5 Y01 5 Y11. Taken together, these two assumptions
guarantee that conditional on X the instrument only has an effect on the outcome through
the variation it induces in the treatment. Conditional on covariates, living in Dresden as
such does not directly affect respondents’ political attitudes.
Assumptions (i) and (ii) might only hold after we condition on a set of covariates X. Our
research design enables us to control for confounding factors at two levels. Below, we show
that the Dresden district was very similar to the other East German districts in our sample.
In this sense, aggregate-level differences are controlled for by design. In addition, the sur-
vey contains a number of individual-level characteristics that we can directly adjust for by
including them in the estimations.
3.2.2 Threats to the exclusion restriction

The exclusion restriction would be called into question if the Dresden district was funda-
mentally different from other districts in our sample. Living conditions in Dresden for
example might have been worse than in other parts of East Germany, making Dresden
respondents more likely to express dissatisfaction with the East German regime.
Figure 2 shows that this possibility is rejected by the data. It displays characteristics
for the eight districts in our sample for the year 1988 (Staatliche Zentralverwaltung für
Statistik 1989; Grundmann 1997). The x axis shows standard deviations from the medians;
the y axis lists the district characteristics. Red solid circles display values for the Dresden
district; empty circles display values for the other districts. The first 24 rows in Fig. 2 list
% industry
% crafts and construction
% agriculture
% services and transportation
2
population density (pop./km )
population movement
% working age
% female
% college
% skilled
% unskilled
2
housing space (m )
% bathroom
% interior toilet
% modern heating
telephone extensions
divorce rate
crime rate
consumption expenditures

people per medical doctor
people per dentist
2
nitrogen oxides (tons/km )
2
sulfur dioxide (tons/km )
particulate matter (tons/km2)
vote share LDP
vote share CDU
vote share SED
turnout
−3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3

Standard deviations from the median
Fig. 2 District characteristics. The graph shows balance in district characteristics for the eight
districts in our sample. The x axis shows standard deviations from the medians; the y axis lists the
district characteristics. Red solid circles display values for the Dresden district; empty circles display
values for the other districts. Note that for several district characteristics only seven values are shown
as data for East Berlin are not always available. Data are taken from Staatliche Zentralverwaltung für
Statistik (1989) and Grundmann (1997).
socioeconomic characteristics; they do not provide any indication that the Dresden district
was fundamentally different from the other districts.
What about differences in political attitudes before West German television ever be-
came available in East Germany? The last four rows of Fig. 2 show turnout and vote shares
for the Christian Conservatives (CDU), Liberals (LDP), and the SED in the 1946 state
elections held in the German territories then occupied by the Soviet Union.6 These elec-
tions were the first state elections after the collapse of the Third Reich. Although the Soviet
occupation forces supported the SED and discriminated against the CDU and LDP, the
elections were still reasonably free (Schmitt 1993; Hajna 2000). As we can see in the sec-
ond to last row in Fig. 2, the SED’s vote share in the Dresden district was close to average.
The same is true for turnout and support for the CDU and LDP. There is no evidence that
long-standing regional differences in political culture predating the introduction of West
German television invalidate the exclusion restriction.
3.2.3 Threats to the ignorability assumption

East Germans who desired to watch West German television might have moved away
from the Dresden district. If interest in West German television was correlated with re-
gime support, which seems likely, this kind of sorting behavior would invalidate our
6
The data are taken from Broszat and Weber (1993), adjusted for redistricting, and weighted by population size.
East Berlin has been excluded since West German parties could run their own candidates in East Berlin at that
time.
instrument. Residential mobility however was exceptionally low in East Germany. On

average, East Germans moved across county borders once every 64 years, that is, ap-
proximately once per lifetime (Grundmann 1998).7 The lack of functioning labor and
housing markets accounts for this extremely low level of residential mobility. In East
Germany’s centrally planned economy, factor allocation was controlled by the state.
Mobility of labor between occupations and firms was not desired since it interfered with
central planning and increased costs, and low labor mobility led to low residential mobility.
Second, a dramatic shortage of housing depressed residential mobility further (Grünert
1996; Grünert, Bernien, and Lutz 1997; Grundmann 1998; Wolle 1998: 182–88; Uunk,
Mach, and Mayer 2005). We therefore think it highly unlikely that the ignorability assump-
tion is violated by spatial sorting.

3.3 Covariates
Even though the use of a natural experiment reduces the danger of confounding, some
imbalances between the treatment and control groups might exist. Fortunately, the survey
contains a relatively rich set of control variables that allows us to relax the ignorability
assumption for the instrument. Here, we face the usual tradeoff involved in the choice
of covariates (Rosenbaum 2002: 76). We want to avoid posttreatment bias, which is caused
by adjusting for variables that are themselves affected by the instrument or the treatment
(Rosenbaum 1984). Since we only have cross-sectional data at our disposal, it is not im-
mediately obvious which variables are pretreatment in this sense. Age, gender, and father’s
and mother’s occupational classification are the only variables in the survey that are caus-
ally prior to television exposure. We call this set of variables the limited covariate set. But
we also want to maximize the credibility of the ignorability assumption. We therefore also
show results for an extensive set of covariates. This set adds marriage status, living sit-
uation (whether the respondent is living alone or with a partner), number of children, high-
est educational attainment, occupational classification, net monthly income, and
employment status to the limited covariate set. For some of these variables, it is not entirely
inconceivable that they might be affected by the instrument or the treatment. As we will
see, however, our results are virtually identical no matter which covariate set we use. All
covariates are fully factorized to avoid functional form assumptions. Missing data, which
affect less than 1% of our sample, are treated as additional categories.
3.4 Estimators
In the absence of covariates, the so-called Wald estimator identifies average treatment
effects for compliers, also called local average treatment effects (LATE) (Imbens and
Angrist 1994):
covðY; ZÞ E½YjZ 5 12E½YjZ 5 0

aLATE 5 5 5 E½Y1 2Y0 jD1 >D0 : ð1Þ
covðD; ZÞ E½DjZ 5 12E½DjZ 5 0
Recall that compliers are individuals for which treatment status is exogenously manip-
ulated by the instrument. In the absence of covariates, the popular two-stage least squares
estimator (2SLS) reduces to the Wald estimator and thus also identifies LATE. Once we
7
East Germany was divided into 15 districts (including East Berlin), which in turn were divided into counties. The
total number of counties was 217.
Table 2 Outcome variables
Fully Largely Largely Fully

disagree, disagree, agree, agree,
n (%) n (%) n (%) n (%) Missing, Total,
1 2 3 4 n (%) n (%)
Convinced of Leninist/Marxist 399 (11) 850 (24) 1697 (48) 571 (16) 47 (1) 3564 (100)
worldview
Feel closely attached to East 154 (4) 600 (17) 1963 (55) 809 (23) 38 (1) 3564 (100)
Germany
Political power is exercised in 301 (8) 874 (25) 1969 (55) 373 (10) 47 (1) 3564 (100)
ways consistent with my

views
condition on covariates, however, 2SLS no longer identifies LATE unless treatment effects
are constant within strata of X (Abadie 2003; Morgan and Winship 2007). This constant
treatment effects assumption is often implausible since it requires that the treatment has the
same effect for individuals with the same covariate values. In particular, the treatment
effect for compliers is not allowed to differ from the treatment effect for noncompliers.
To allow for heterogeneous treatment effects, we also implement a new class of IV esti-
mators called local average response functions (LARF) recently proposed by Abadie
(2003). The semiparametric LARF estimator allows for the identification of LATE con-
ditional on covariates even without the constant treatment effects assumption.8
3.5 Outcome Variables

The survey includes several questions that measure support for the East German regime.
Here, we focus on the three questions that seem most relevant:9
To what extend do you agree with the following statements:
‘‘I am convinced of the Leninist/Marxist worldview.’’
‘‘I feel closely attached to East Germany.’’
‘‘In East Germany, political power is exercised in ways consistent with my views.’’
Response categories are fully disagree, largely disagree, largely agree, and fully agree. The
distribution of the outcome variables is shown in Table 2.
4 Results
Table 3 presents the first set of results. The first column shows simple differences in means
between exposed and unexposed respondents. Results for all three outcome variables sug-
gest that watching West German television had only a small and statistically insignificant
effect on regime support. However, there is every reason to believe that these estimates are
confounded by self-selection.
8
Standard errors are adjusted for clustering within the 169 sampling units (e.g., college lecture classes). We use the
Eicker-Huber-White sandwich estimator for the 2SLS models and a pairs cluster bootstrap for the LARF spec-
ifications (Davison and Hinkley 1997: 101–103).
9
Results are substantively identical when other questions are used instead.
Table 3 Effect of West German television exposure on regime support
Estimator Diff. LATE 2SLS LARF 2SLS LARF

Covariate set — — Limited Limited Extensive Extensive
Convinced of Leninist/Marxist worldview

West German TV 20.079 0.147 0.205 0.204 0.198 0.204
(0.053) (0.083) (0.084) (0.084) (0.087) (0.108)
Feel closely attached to East Germany
West German TV 20.013 0.217 0.258 0.255 0.256 0.251
(0.044) (0.067) (0.072) (0.075) (0.073) (0.090)
Political power exercised in ways consistent with my views
West German TV 20.014 0.158 0.193 0.191 0.186 0.185

(0.047) (0.078) (0.082) (0.083) (0.081) (0.106)
Note. N 5 3441. The table shows treatment effect estimates for each specification with cluster-adjusted standard
errors in parentheses. Diff. denotes the difference in means between exposed and unexposed respondents. LATE is
the unconditional average treatment effect for compliers. 2SLS is the two-stage least squares estimator. LARF is
the local average response function estimator. The limited covariate set includes age, gender, and father’s and
mother’s occupational classification. The extensive covariate set adds marriage status, living situation, number of
children, highest educational attainment, occupational classification, net monthly income, and employment
status to the limited set. Response categories for the outcome variables are coded as fully disagree 5 1, largely
disagree 5 2, largely agree 5 3, and fully agree 5 4.
Column 2 displays unconditional LATE estimates. According to these specifications,

exposure to West German television increased regime support by 0.15–0.22 on the four-
point response scale. These treatment effect estimates are statistically significant at the
.10 level or better.
The unconditional LATE estimates are confirmed when we introduce respondent-level
covariates in columns 3–6. Exposure to West German television continues to have a positive
effect on regime support; effect sizes vary between .19 and .26 depending on the exact
specification. All estimates except one are statistically significant at the .05 level. Estimates
for the subgroup of compliers (LARF, columns 4 and 6) and the whole sample (2SLS,
columns 3 and 5) are virtually identical, which suggests that the constant treatment effects
assumption is not critical in our application. The limited and extensive covariate sets also
lead to very similar results. Overall, we find that exposure to West German television had
a substantial positive effect on regime support. Evaluated at the means of the three outcome
variables, the average increase in regime support due to West German television exposure
was between 11% and 15%.10
How can we explain this surprising positive effect of West German television exposure?
The survey also contains several questions that allow us to test the deterrence and escapism
hypotheses. Recall that both hypotheses predict that exposure to West German television
would increase support for the East German regime. The deterrence hypothesis posits that
West German television exposed East Germans to negative aspects of West German society
and thereby convinced them of the dangers of capitalism and the merits of communist rule.
The escapism hypothesis in contrast predicts that West German television, and in particular
its entertainment programming, allowed East Germans to vicariously escape the dull life
under communism at least for a couple of hours each night, making their lives more bear-
able and the East German regime more tolerable.
10
LARF ordered probit results, shown in the online supplement, are substantively similar to the results from these
linear models.
Table 4 Deterrence and escapism
Covariate set Limited Extensive
Do you feel attached to West Germany? (mean 5 2.28; SD 5 1.08)

West German TV 20.193 (0.111) 20.172 (0.113)
N 1141 1141
How satisfied are you with the available options for recreational activities? (mean 5 3.05; SD 5 1.10)
West German TV 0.395 (0.117) 0.405 (0.110)
N 1148 1148
Note. Results are 2SLS treatment effect estimates with cluster-adjusted standard errors in parentheses. The limited

covariate set includes age, gender, and father’s and mother’s occupational classification. The extensive covariate
set adds marriage status, living situation, number of children, highest educational attainment, occupational
classification, net monthly income, and employment status to the limited set. The 5-point response categories
range from not at all 5 1 to fully 5 5.
Respondents were asked to what extent they felt attached to West Germany (upper half
of Table 4).11 If the deterrence hypothesis is correct and West German television exposed
East Germans to the ‘‘dark side’’ of West German society, we would expect to see a negative
effect of West German television exposure on attachment to West Germany. And this is
indeed what we find, although the effect estimate is not all that large and does not quite
reach conventional levels of statistical significance. Exposure to West German television
on average decreased attachment to West Germany by about .17 on the 5-point response
scale. This result is consistent with the deterrence hypothesis, although we are the first to
admit that it would be desirable to have survey items on specific aspects of West German
society instead of this more general question. Political constraints prevented the Central
Institute from adding detailed questions about West Germany to their questionnaires.
The escapism hypothesis receives much stronger support from the data. Respondents
were also asked about their satisfaction with options for recreational activities. If West
German television was mainly a means of vicarious escape from ‘‘real existing socialism,’’
exposure to West German television should have increased respondents’ satisfaction with
how they could spend their free time. As one can see from the lower half of Table 4, ex-
posure to West German television indeed had such an effect. On average, it increased sat-
isfaction by about .40 on the 5-point response scale. This effect is more than twice as large
as the deterrence effect; it is also highly statistically significant. Even if deterrence con-
tributed to the positive effect of West German television exposure on regime support, the
primary causal channel seems to be escapism.
These results are consistent with previous findings on the use of television in East
Germany. We know from detailed survey data on East Germans’ television viewing habits
that they watched television primarily for its entertainment value.12 When asked what
changes they would like to see in East German television programming, they voted for
more entertainment and less politics (Meyen 2001, 2003a). Former East Germans’ tele-
vision viewing habits still differ from those of West Germans although all Germans now
have access to the same channels. Former East Germans have a stronger preference for
11
Note that only a subset of survey participants was asked these questions. The sample sizes in Table 4 are there-
fore smaller than in Table 3.
12
Deutsches Rundfunkarchiv (German Broadcasting Archive), Ergebnisse der Programmarbeit im Zeitraum 1.
Juli 1988 bis 8. Januar 1989, Programmredaktion Analyse II/1988.
entertainment programming than West Germans. They also favor commercial stations,
which put greater emphasis on infotainment and soap operas, over public broadcasting
stations (Darschin and Zubayr 2000; Früh and Stiehler 2002).
We have conducted extensive robustness checks, all of which are fully documented in
the online supplement. We have replicated our analysis for regional subsamples, each com-
posed of respondents living in Dresden and in one of the seven districts with access to West
German television. If there was regional variation in unobserved confounders such as
respondents’ levels of preference falsification, there should be heterogeneity in treatment
effect estimates across these subsamples. There is none. We have also replicated our anal-
ysis using data from an earlier survey conducted in 1984. In addition to respondents from
Dresden, it contains a sample of young adults from several smaller East German cities

(Halle, Schwerin, Dessau, Magdeburg, and Bitterfeld). Across all specifications, West
German television exposure has a positive effect on East Germans’ regime support, with
effect sizes similar to the ones reported here. The survey also contains 167 respondents
from Greifswald, a town in the Northeast of East Germany that for the most part also had
no access to West German television (Fig. 1). We reestimated our models with these
Greifswald respondents (instead of Dresden respondents) and again found a positive effect
of West German television exposure. Taken together, these additional survey data results
demonstrate that the findings reported here are not driven by biases particular to the
1988–89 survey or peculiarities of the Dresden district.
5 Exit Visa Applications

Is it possible to corroborate our survey data results with other indicators of regime support?
A very direct measure of political (dis-)satisfaction is the decision to emigrate. Even
though East Germans had no legal right to emigrate to another country, some East Germans
insisted on moving to West Germany. Applying for an exit visa often had severe conse-
quences. Applicants faced numerous punishments such as getting fired from their jobs,
informal discrimination at work, confiscation of their passports, and criminal persecution
(Mayer 2002: 177–181; Pfaff 2006).
Table 5 shows the number of new exit visa applications (per 1000 residents) filed in the
districts in our sample in 1988. In the Dresden district, the rate at which such applications
were filed was more than twice as high as the average.13
Table 5 is suggestive, but it is preferable to examine the relationship between exit visa
applications and access to West German television at a more disaggregate level. In the mid-
1980s, the Department of Postal Services, responsible for radio and television broadcasting
in East Germany, conducted an investigation into West German television’s over-the-air
signal strength in the 17 counties that made up the Dresden district.14 The resulting map,
submitted to the Politburo as part of a larger collection of material on West German tele-
vision in East Germany, is shown as Fig. 3. It shows counties without access to West
German television (white), counties with partial access to West German television, depend-
ing on the weather and exact location (light gray), and counties with full access to West
German television (dark gray).
13
The same pattern existed throughout the 1980s. Official statistics can be found in Office of the Federal Com-
missioner for the Records of the National Security Service of the Former German Democratic Republic (here-
after BStU) MfS-ZKG 10734.
14
Federal Archive, Foundation for the Archive of the Parties and Mass Organizations of the GDR (hereafter BArch
SAPMO) DY 30/J IV2/2/2317.
Table 5 Exit visa application rates in 1988
District Rate
Dresden 16.4
East Berlin 11.8
Karl-Marx-Stadt 9.8
Leipzig 7.9
Erfurt 5.0
Cottbus 3.5
Magdeburg 2.8
Schwerin 2.4
Mean 7.45

Note. The table shows the number of new exit visa
applications (per 1000 residents) filed in 1988.
Source: BStU MfS-ZKG 10734.
We were also able to obtain county-level exit visa application data for 1985–88.15 If
exposure to West German television had a positive effect on regime support, we would
expect the rate at which new exit visa applications were filed to be negatively associated
with the extent to which residents of these counties had access to West German television.
Fig. 3 Over-the-air signal strength of West German television broadcasts within the Dresden district.
The map shows spatial patterns in access to West German television across the 17 counties in the
Dresden district. White areas had no access, areas shaded in light gray had partial access, and areas
shaded in dark gray had full access to West German television. Source: Adapted from original map in
BArch SAPMO DY 30/J IV2/2/2317: 75.
15
BStU MfS BV Dresden Stellvert. OP 21; BStU MfS BV Dresden BKG-8028/8029.
1985
full
partial
Access to West German television

no
1986
full
partial
no
1987
full

partial
no
1988
full
partial
no
4 8 12 16
Exit visa applications per 1,000 residents
Fig. 4 Exit visa application rates and access to West German television for counties in the Dresden
district. Each of the four panels shows three box-and-whisker plots for groups of counties with
different levels of access to West German television. Görlitz-Stadt and Görlitz-Land have been
merged since only the total number of exit visa applications is available for these two counties, so the
sample size is 16 counties. Three of these counties had full access, six partial access, and seven no
access to West German television. Source: BStU MfS BV Dresden Stellvert. OP 21; BStU MfS BV
Dresden BKG-8028/8029.
For each year, Fig. 4 plots the distribution of exit visa application rates for counties with
full access, partial access, and no access to West German television. We find a clear dose-
response relationship: In all 4 years, the rate at which exit visa applications were filed was
highest in the group of counties without access to West German television and lowest in the
group of counties with access to West German television. Rates in the group of counties
with partial access were always in-between. These differences are large in substantive
terms. Averaging over all four cross-sections, the mean rate at which applications were
filed in counties without access to West German television was more than twice as high
as in counties with (partial or full) access to West German television (6.03 versus 2.82
applications per 1000 residents; p 5 .02).
6 Historical Evidence
A third type of empirical evidence that we can bring to bear is archival material on the
reaction of the East German regime to the availability of West German television. We ex-
pect that if West German television helped to stabilize the East German regime, this sur-
prising fact would not have gone unnoticed by East Germany’s massive state security
apparatus and the East German leadership.
Already in the 1960s, a large majority of East Germans, including many SED members,
watched West German television. Although it was never illegal to watch West German
television (except for members of the armed forces and the police), in the 1960s and
1970s East German authorities attempted to suppress its popularity through propaganda
and harassment campaigns. The East German regime considered West German television
a source of ideological destabilization and feared its effects on public opinion. It therefore
attempted to limit its availability as much as possible.16 After the construction of the Berlin
Wall in 1961, for example, SED youth squads tore down hundreds of roof antennas directed
toward West Germany. The extreme unpopularity of such campaigns however meant that
they were soon abandoned (Wolle 1998: 69–71; Stiehler 2001: 13–16; Holzweissig 2002:
49–65).
The attitude of the East German regime eventually changed. It had no choice but
to accept that a large majority (85%–90%) of East Germans regularly tuned in to West
German television. By 1984, 40% of East German households were connected to commu-

nity antennas, large antenna systems linked to individual households via cable. Community
antennas often carried West German channels.17 East Germans not connected to commu-
nity antennas relied on smaller roof antennas. In some areas, however, reception of West
German broadcasts was poor or even impossible and residents increasingly complained to
state authorities about their inability to watch ‘‘international programs,’’ as West German
television was euphemistically called.18 State-owned housing companies sometimes paid
for the installation of community antennas, thus providing their tenants with access to West
German television.19 In other cases, East Germans joined forces in grassroot efforts and
paid for the installation of community antennas themselves. Although East German au-
thorities were suspicious of such grassroot activities and kept them under close surveil-
lance, they did nothing to prevent them (Stiehler 2001: 95–96).
Since the mid-1980s, East German authorities were increasingly confronted with
requests from individuals and community initiatives to permit the installation of satellite
dishes. Such requests came especially from parts of East Germany where even with com-
munity antennas West German television reception was poor (such as parts of the Dresden
district). The private use of satellite dishes was prohibited in East Germany but some East
Germans nonetheless installed satellite dishes they had illegally imported from West
Germany or Hungary.20 When the Department of Postal Services, which was responsible
for the enforcement of telecommunications regulations, demanded the removal of such
satellite dishes and imposed fines their owners would simply remove them only to reinstall
them later.21
In a March 1988 report to the Politburo, the Department conveyed a dramatic picture of
the prevailing public mood:
Demands from grass-roots initiatives and individual citizens to allow the reception of West German
television [via satellite dishes] are becoming more and more brazen. There are constant disputes.
The whole topic is increasingly becoming a political issue, especially since these grass-roots ini-
tiatives and individual citizens attempt to gain access to satellite television at all costs and fines
imposed for violations of legal regulations do not have the desired effect.22
Several months earlier, the Department had insisted on the removal of a satellite dish in
Marienberg that provided 4000 households with access to West German television. This
16
BArch SAPMO 30/IVA2/902/68.
17
BArch SAPMO DY 30/IV2/2.039/276: 4–12.
18
BStU 3189/87 IV: 2–5; BStU 3189/87 III: 58–61, 156–7.
19
BStU 3189/87 IV: 6, 62.
20
BArch SAPMO DY 30/IV2/2.039/276: 36–40, 47–49.
21
BArch SAPMO DY 30/IV2/2.039/276: 40–43.
22
BArch SAPMO DY 30/IV2/2.039/276: 37.
ruling led to massive protests and even local SED party officials recommended that the
satellite dish should be allowed to stay since its removal would ‘‘lead to serious confron-
tations with thousands of citizens.’’23 A similar case happened in the town of Weissenberg.
Residents of Weissenberg too had illegally installed a satellite dish. Pleading with state
authorities who insisted on its removal, local SED party officials as well as the mayor
pointed out that members of their community were ‘‘much more content’’ since the in-
troduction of West German television. Their attitudes toward the East German regime
had become ‘‘more positive’’ and all applications for exit visas had been withdrawn.
The dismantling of the satellite dish would cause ‘‘enormous political problems’’ since
it would contradict the wishes of the entire town.24
The Politburo decided in August 1988 on the basis of such reports to almost completely

liberalize the private use of satellite dishes (Holzweissig 2002; Stiehler 2001: 95).25 Six
months later, in February 1989, 203 satellite dishes were officially registered in the Dres-
den district, with 250 further applications pending.26 The archival evidence we have sum-
marized here illustrates a dramatic change in the regime’s attitude toward West German
television. In the 1960s and 1970s, West German television was seen as undermining pub-
lic support for communist rule. But during the 1980s, the East German regime began to
realize that West German television, not unlike ancient Rome’s ‘‘bread and circuses,’’
helped to keep its citizens content and docile. In 1988, the East German leadership appar-
ently even contemplated building broadcast relay stations that would have amplified the
West German television signal so as to extend its range to the Dresden district (Wolle 1998:
69–71).
7 Discussion
In this paper, we have taken advantage of a natural experiment in East Germany to test the
commonly held view that foreign mass media undermine the stability of authoritarian
regimes. We have found that exposure to West German television increased support for
the communist regime among East German teenagers and young adults. Our analysis sug-
gests that the best explanation for this counterintuitive finding is escapism: West German
television, and especially its entertainment programming, allowed East Germans to vicar-
iously escape life under communism at least for a couple of hours each night, making their
lives more bearable and the East German regime more tolerable. West German television
also broadcast high-quality hard news programs, some of which offered in-depth coverage
of politics in East Germany. There is no doubt that this political content could have under-
mined public support for the East German regime. What we have found, however, is that
West German television exposure resulted in a net increase in regime support.
Behavioral data on exit visa applications and archival material on the reaction of the East
German leadership to the availability of West German television corroborate our results.
We have found that in the Dresden district, where levels of access to over-the-air West
German television broadcasts varied from county to county, exit visa application rates were
systematically higher in counties without West German television. And the archival evi-
dence clearly shows that by the late 1980s, East German authorities had come to realize that
23
BArch SAPMO DY 30/IV2/2.039/276: 41.
24
BArch SAPMO DY 30/IV2/2.039/276: 45.
25
BStU Mf S BV Dresden 11147.
26
BArch SAPMO DY 30/J IV2/2/2317: 72–77. Note that satellite dishes were not sold in East Germany and had to
be imported from abroad, which prevented most East Germans from obtaining one.
West German television was contributing to the stability of the regime, not undermining it.
In an ironic twist for Marxism, capitalist television seems to have performed the same nar-
cotizing function in communist East Germany that Karl Marx had attributed to religious
beliefs in capitalist societies when he condemned religion as ‘‘opium of the people.’’
Our survey only includes teenagers and young adults, which raises questions about
the generalizability of our findings to other strata of the East German population. The fact
that the behavioral and archival evidence agrees with our survey data results suggests that
our inferences are not necessarily limited in their generalizability. But given the lack of
public opinion data representative of East Germany as a whole, we cannot prove that West
German television exposure had the same effect on other strata of the East German
population.

We are the first to admit that our results do not provide a comprehensive explanation for
why some authoritarian regimes are able to stay in power while others are replaced by
democratic governments. The study of individuals’ political preferences, including how
they are shaped by exposure to foreign mass media, is only one step toward a better
understanding of authoritarian stability and decline (Welzel 2006; Welzel and Inglehart
2008). Work complementary to ours has analyzed the conditions under which individuals’
political preferences lead to collective revolutionary action (Granovetter 1978;
Bikhchandani, Hirshleifer, and Welch 1988; Kuran 1991; Lohmann 1994) and how the
international opportunity structure affects the emergence and success of popular move-
ments against authoritarian rule (Pridham 1991, 1997; Whitehead 1996; Linz and Stepan
1996). Even though our results are context specific (East Germany in the middle to late
1980s) and limited to the impact of one specific foreign media source, we believe they
usefully complement the existing literatures on transitions to democracy and political
legitimacy in authoritarian regimes.
We believe our results also have relevance for contemporary U.S. international broad-
casting efforts. After the attacks of September 11, Congress substantially increased funding
for public diplomacy activities, which had been significantly reduced after the end of the
Cold War (Nye 2004, 2008). In 2008 alone, the United States has spent almost $700 million
on radio and television broadcasts to the Middle East (Radio Sawa and Alhurra TV),
Eastern Europe (Radio Free Europe/Radio Liberty), Asia (Radio Free Asia), Cuba
(Radio/TV Martı´), and the Voice of America. As during the Cold War, the goal of contem-
porary U.S. international broadcasting is to win the hearts and minds of foreign audiences,
to increase understanding of and support for U.S. foreign policies, and to shift public opin-
ion in non-democratic regimes in a more pro-democratic direction. But if history is any
guide, foreign audiences might be more interested in the entertainment value of U.S. broad-
casts than their political content.
Finally, our results also offer a caveat to optimistic predictions about the democratizing
force of modern information technologies such as the Internet. Authoritarian regimes seem
to be well aware of the fact that modern information technology not only breaks down
national barriers to the transmission of information but also provides entertainment. China,
for instance, successfully pursues a strategy of access and control, embracing the economic
and entertainment uses of the internet to promote development and sustain public support
while strictly controlling political content (Kalathil and Boas 2003; Shirk 2007).
References
Abadie, Alberto. 2003. Semiparametric instrumental variable estimation of treatment response models. Journal of
Econometrics 113:231–63.
Angrist, Joshua D., Guido W. Imbens, and Donald B. Rubin. 1996. Identification of causal effects using instru-
mental variables. Journal of the American Statistical Association 91:444–72.
Bartels, Larry M. 1993. Messages received: The political impact of media exposure. American Political Science
Review 87:267–85.
Bennett, W. Lance. 1998. The media and democratic development: The social basis of political communication. In
Communicating democracy: The media and political transitions, ed. Patrick H. O’Neil, 195–207. Boulder:
Lynne Rienner.
Bikhchandani, Sushil, David Hirshleifer, and Ivo Welch. 1998. Learning from the behavior of others: Conformity,
fads, and informational cascades. Journal of Economic Perspectives 12(3):151–70.
Black, Sandra E. 1999. Do better schools matter? Parental valuation of elementary education. Quarterly Journal of
Economics 114:577–99.
Brinks, Daniel, and Michael Coppedge. 2006. Diffusion is no illusion: Neighbor emulation in the third wave of
democracy. Comparative Political Studies 39:463–89.

Broszat, Martin, and Hermann Weber. 1993. SBZ-Handbuch: Staatliche Verwaltungen, Parteien, gesellschaftliche
Organisationen und ihre Führungskräfte in der Sowjetischen Besatzungszone Deutschlands 1945–1949.
2nd ed. Munich: Oldenbourg.
Dalton, Russell J. 1994. Communists and democrats: Democratic attitudes in the two Germanies. British Journal
of Political Science 24:469–93.
Darschin, Wolfgang, and Camille Zubayr. 2000. Warum sehen die Ostdeutschen anders fern als die Westdeut-
schen? Media Perspektiven 6:249–57.
Davison, A. C., and D. V. Hinkley. 1997. Bootstrap methods and their application. Cambridge and New York:
Cambridge University Press.
Diamond, Larry, ed. 1993a. Political culture and democracy in developing countries. Boulder, CO: Lynne Rienner.
———. 1993b. The globalization of democracy. In Global transformation and the Third World, eds. Robert O.
Slater, Barry M. Schutz, and Steven R. Dorr, 31–69. Boulder, CO: Lynne Rienner.
Doorenspleet, Renske. 2004. The structural context of recent transitions to democracy. European Journal of
Political Research 43:309–35.
Dunning, Thad. 2008. Improving causal inference: Strengths and limitations of natural experiments. Political
Research Quarterly 61:282–93.
Freedman, David A. 2005. Statistical models: Theory and practice. Cambridge and New York: Cambridge
University Press.
Friedrich, Walter. 1990. Mentalitätswandlungen der Jugend in der DDR. Aus Politik und Zeitgeschichte. Beilage
zur Wochenzeitung Das Parlament 16–17:25–37.
Friedrich, Walter, Peter Förster, and Kurt Starke. 1999. Das Zentralinstitut für Jugendforschung Leipzig
1966–1990. Berlin: edition ost.
Früh, Werner, and Hans-Jürg Stiehler. 2002. Fernsehen in Ostdeutschland. Berlin: Vistas.
Gasiorowski, Mark J. 1995. Economic crisis and political regime change: An event history analysis. American
Political Science Review 89:882–97.
Geddes, Barbara, and John Zaller. 1989. Sources of popular support for authoritarian regimes. American Journal
of Political Science 33:319–47.
Giddens, Anthony. 2000. Runaway world. New York: Routledge.
Gleditsch, Kristian Skrede, and Michael D. Ward. 2006. Diffusion and the international context of democrati-
zation. International Organization 60:911–33.
Granovetter, Mark. 1978. Threshold models of collective behavior. American Journal of Sociology 83:1420–43.
Grundmann, Siegfried. 1997. Territorialplanung in der DDR: Indikatoren zur Analyse regionaler Disparitäten —
Die sozial-räumliche Struktur der DDR in den 80er Jahren. In Regionale Strukturen im Wandel, ed. Annette
Becker, 7–44. Opladen: Leske1Budrich.
———. 1998. Bevölkerungsentwicklung in Ostdeutschland. Opladen: Leske1Budrich.
Grünert, Holle. 1996. Das Beschäftigungssystem der DDR. In Arbeit, Arbeitsmarkt und Betriebe, eds. Burkart
Lutz, Hildegard M. Nickel, Rudi Schmidt, and Arndt Sorge. Opladen: Leske1Budrich.
Grünert, Holle, Maritta Bernien, and Burkart Lutz. 1997. Das Beschäftigungssystem der DDR: Funktionsweise,
Entwicklungstendenzen und Folgewirkungen. In Der ostdeutsche Arbeitsmarkt in Gesamtdeutschland: Anglei-
chung oder Auseinanderdriften? eds. Doris Beer, Christian Brinkmann, Axel Deeke, and Sabine Schenk,
127–49. Opladen: Leske1Budrich.
Hajna, Karl-Heinz. 2000. Die Landtagswahlen 1946 in der SBZ. Frankfurt am Main: Peter Lang.
Hesse, Kurt R. 1988. Westmedien in der DDR: Nutzung, Image und Auswirkungen bundesrepublikanischen
Hörfunks und Fernsehens. Köln: Verlag Wissenschaft und Polititk.
———. 1990. Cross-border mass communication from West to East Germany. European Journal of Communi-
cation 5:355–71.
Holland, Paul W. 1986. Statistics and causal inference. Journal of the American Statistical Association
81:945–60.
Holzweissig, Gunter. 2002. Die schärfste Waffe der Partei: eine Mediengeschichte der DDR. Köln, Germany:
Böhlau.
Huntington, Samuel P. 1991. The third wave: Democratization in the late twentieth century. Norman, OK:
University of Oklahoma Press.
Imbens, Guido W., and Joshua D. Angrist. 1994. Identification and estimation of local average treatment effects.
Econometrica 62:467–76.
Imbens, Guido W., and Jeffrey M. Wooldridge. 2008. Recent developments in the econometrics of program
evaluation. NBER Working Paper 14251.
Kalathil, Shanti, and Taylor C. Boas. 2003. Open networks, closed regimes: The impact of the internet on

authoritarian rule. Washington, DC: Carnegie Endowment for International Peace.
Katz, Elihu, and David Foulkes. 1962. On the use of the mass media as ‘escape’: Clarification of a concept. Public
Opinion Quarterly 26:377–88.
Kopstein, Jeffrey S., and David A. Reilly. 2000. Geographic diffusion and the transformation of the postcom-
munist world. World Politics 53:1–37.
Kuran, Timur. 1991. Now out of never: The element of surprise in the East European revolution of 1989. World
Politics 44:7–48.
Lazarsfeld, Paul F., and Robert K. Merton. 1948. Mass communication, popular taste and organized social action.
In Communication of ideas, ed. Lyman Bryson, 95–118. New York: Harper & Brothers.
Lisann, Maury. 1975. Broadcasting to the Soviet Union: International politics and radio. New York: Praeger
Publishers.
Linz, Juan J., and Alfred Stepan. 1996. Problems of democratic transition and consolidation: southern Europe,
South America, and post-communist Europe. Baltimore, MD: Johns Hopkins University Press.
Lohmann, Susanne. 1994. The dynamics of informational cascades: The Monday demonstrations in Leipzig, East
Germany, 1989–1991. World Politics 47:42–101.
O’Loughlin, John, Michael D. Ward, Corey L. Lofdahl, Jordin S. Cohen, David S. Brown, David Reilly,
Kristian S. Gleditsch, and Michael Shin. 1998. The diffusion of democracy, 1946–1994. Annals of the Asso-
ciation of American Geographers 88:545–74.
Mayer, Wolfgang. 2002. Flucht und Ausreise: Botschaftsbesetzungen als Form des Widerstands gegen die
Politische Verfolgung in der DDR. Berlin: Anita Tykve Verlag.
Meyen, Michael. 2001. Haben die Westmedien die DDR stabilisiert? Siegener Periodicum zur internationalen
empirischen Literaturwissenschaft 20:117–33.
———. 2003a. Einschalten, Umschalten, Auschalten? Das Fernsehen im DDR-Alltag. Leipzig: Leipziger
Universitätsverlag.
———. 2003b. Denver Clan und Neues Deutschland: Mediennutzung in der DDR. Berlin: Ch. LinksVerlag.
Morgan, Stephen L., and Christopher Winship. 2007. Counterfactuals and causal inference: Methods and
principles for social research. New York: Cambridge University Press.
Nelson, Michael. 1997. War of the black heavens: The battles of Western broadcasting in the Cold War. Syracuse,
NY: Syracuse University Press.
Nye, Joseph S., Jr. 2004. Soft power: The means to success in world politics. New York: Public Affairs.
———. 2008. Public diplomacy and soft power. Annals of the American Academy of Political and Social Science
616:94–109.
O’Donnell, Guillermo, Philippe C. Schmitter, and Laurence Whitehead, eds. 1986. Transitions from authoritarian
rule: prospects for democracy. Baltimore, MD: Johns Hopkins University Press.
Parta, R. Eugene. 2007. Discovering the hidden listener: An assessment of Radio Liberty and Western Broad-
casting to the USSR during the Cold War. Stanford, CA: Hoover Institution Press.
Pfaff, Steven. 2006. Exit-voice dynamics and the collapse of East Germany. Durham and London: Duke University
Press.
Pridham, Geoffrey, ed. 1991. Encouraging democracy: The international context of regime transition in Southern
Europe. Leicester: Leicester University Press.
———. 1997. The international dimension of democratization: theory, practice, and inter-regional comparisons.
In Building democracy? The international dimension of democratization in Eastern Europe, eds.
Geoffrey Pridham, Eric Herring, and George Sanford, 1–22. Leicester: Leicester University Press.
Puddington, Arch. 2000. Broadcasting freedom: The Cold War triumph of Radio Free Europe and Radio Liberty.
Lexington: University Press of Kentucky.
Quester, George H. 1990. The international politics of television. Lexington, MA: Lexington Books.
Roberts, Adam. 1999. Civil resistance in the East European and Soviet revolutions. Albert Einstein Institution
Monograph Series no. 4.
Rosenbaum, Paul R. 1984. The consequences of adjustment for a concomitant variable that has been affected by
the treatment. Journal of the Royal Statistical Society Series A 147:656–666.
———. 2002. Observational studies. 2nd edn. New York: Springer.
Rubin, Donald B. 1974. Estimating causal effects of treatments in randomized and nonrandomized studies.
Journal of Educational Psychology 66:688–701.
———. 1978. Bayesian inference for causal effects: The role of randomization. Annals of Statistics 6:1–26.
Rohrschneider, Robert. 1994. Report from the laboratory: The influence of institutions on political elites’
democratic values in Germany. American Political Science Review 88:927–41.
———. 1996. Institutional learning versus value diffusion: The evolution of democratic values among parlia-
mentarians in Eastern and Western Germany. Journal of Politics 58:422–46.

Rustow, Dankwart A. 1990. Democracy: A global revolution? Foreign Affairs 69(4):75–91.
Shirk, Susan L. 2007. China: fragile superpower. Oxford and New York: Oxford University Press.
Schmitt, Karl. 1993. Politische Landschaften im Umbruch: Das Gebiet der ehemaligen DDR 1928–1990. In
Wahlen in Zeiten des Umbruchs, eds. W. Oscar Gabriel and Klaus G. Troitzsch. Frankfurt am Main: Peter Lang.
Staatliche Zentralverwaltung für Statistik. 1989. Statistisches Jahrbuch 1989. Berlin: Staatsverlag der Deutschen
Demokratischen Republik.
Starr, Harvey. 1991. Democratic dominoes: Diffusion approaches to the spread of democracy in the international
system. Journal of Conflict Resolution 35:356–81.
Starr, Harvey, and Christina Lindborg. 2003. Democratic dominoes revisited: The hazards of governmental
transitions, 1974–1996. Journal of Conflict Resolution 47:490–519.
Stiehler, Hans-Jörg. 2001. Leben ohne Westfernsehen. Leipzig: Leipziger Universitätsverlag.
Stroud, Natalie Jomini 2008. Media use and political predispositions: Revisiting the concept of selective exposure.
Political Behavior 30:341–66.
Sükösd, Miklos. 2000. Democratic transformation and the mass media in Hungary: From stalinism to democratic
consolidation. In Democracy and the media, eds. Richard Gunther and Anthony Mughan. Cambridge and New
York: Cambridge University Press.
Welzel, Christian. 2006. Democratization as an emancipative process: The neglected role of mass motivations.
European Journal of Political Research 45:871–96.
Welzel, Christian, and Ronald Inglehart. 2008. The role of ordinary people in democratization. Journal of
Democracy 19:126–40.
Whitehead, Laurence. 1996. Three international dimensions of democratization. In The international dimensions
of democratization, ed. Laurence Whitehead. Oxford and New York: Oxford University Press.
Wolle, Stefan. 1998. Die heile Welt der Diktatur. Alltag und Herrschaft in der DDR 1971–1989. Berlin: Ch. Links
Verlag.
Uunk, Wilfred, Bogdan W. Mach, and Karl Ulrich Mayer. 2005. Job mobility in the former East and West
Germany: The effects of state-socialism and labor market composition. European Sociological Review
21:393–408.
Zelikow, Philip, and Condoleezza Rice. 1995. Germany unified and Europe transformed: a study in statecraft.
Cambridge, MA: Harvard University Press.

Holger Lutz Kern & Jens Hainmueller - Opium For The Masses: How Foreign Media Can Stabilize Authoritarian Regimes

Uploaded by

Copyright:

Available Formats

Holger Lutz Kern & Jens Hainmueller - Opium For The Masses: How Foreign Media Can Stabilize Authoritarian Regimes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Holger Lutz Kern & Jens Hainmueller - Opium For The Masses: How Foreign Media Can Stabilize Authoritarian Regimes

Uploaded by

Copyright:

Available Formats

Advance Access publication July 30, 2009 Political Analysis (2009) 17:377–399

Opium for the Masses: How Foreign Media Can

Holger Lutz Kern

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

2 The Effects of Foreign Media Exposure

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

3.1 Survey Data

3.2 Causal Inference Using Instrumental Variables

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

3.2.1 Identification assumptions

Table 1 West German television exposure and place of residence

How often do you watch Dresden district Other districts

Daily 42 (5) 1820 (65)

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

3.2.2 Threats to the exclusion restriction

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

−3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3

3.2.3 Threats to the ignorability assumption

instrument. Residential mobility however was exceptionally low in East Germany. On

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

covðY; ZÞ E½YjZ 5 12E½YjZ 5 0

Table 2 Outcome variables

Fully Largely Largely Fully

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

3.5 Outcome Variables

Table 3 Effect of West German television exposure on regime support

Estimator Diff. LATE 2SLS LARF 2SLS LARF

Convinced of Leninist/Marxist worldview

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Column 2 displays unconditional LATE estimates. According to these specifications,

Table 4 Deterrence and escapism

Covariate set Limited Extensive

Do you feel attached to West Germany? (mean 5 2.28; SD 5 1.08)

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

5 Exit Visa Applications

Table 5 Exit visa application rates in 1988

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Access to West German television

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

Downloaded from http://pan.oxfordjournals.org/ at Northeastern University Libraries on April 1, 2014

You might also like