appendix
Quantitative Data and the Statistical Analysis
of County-Level Exit and Voice Relationships
Protest Event Data
For the analysis of voice, detailed data—especially on protest and state repression—are
di≈cult to come by for socialist states. Many existing studies rely on secondary source
materials such as Western newspapers and previously published accounts or on retro-
spective, self-reported survey data. However, state-controlled media are notoriously
unreliable, and Western newspaper stories are limited by the fact that reporters cannot
venture wherever they wish to (in the GDR, they were generally only present in larger
cities such as East Berlin, Leipzig, or Dresden). Self-reported data on individual par-
ticipation often su√er from social-desirability and retrospective biases. Because of these
considerations, in the quantitative study in this book, secondary source materials were
not employed to code the dependent variables Protest Event Rate (per) and Average
Participation Rate (apr). Instead, a powerful, if not perfect, county-level dataset was
assembled from summary police reports in the archives of the GDR Interior Ministry.
These records cover the whole of the country during the period from September 1989 to
March 1990; thus the events statistically analyzed in this book range from the first
popular demonstrations against the regime to the free elections on March 18, 1990.
The documents coded were composed of crisis reports made by GDR People’s Police
(Volkspolizei) at the county/municipal level (Kreis or Stadtkreis) to the Interior Ministry
in East Berlin. The reports list politically motivated illegal assemblies, demonstrations,
and violations of public order (Störungen der ö√entlichen Ordnung und Sicherheit) oc-
curring within a county. These crisis reports are complete for the period between Sep-
tember 1989 and April 1990 (see Bundesarchiv der Bundesrepublik Deutschland [Berlin-
Lichterfelde], files barch do 1 2.3/053614, do 1 2.3/052445, and do 1 2.3/052449). Only
those events that involved public manifestations of voice—such as demonstrations, as-
semblies, marches, the seizure of public buildings, strikes, rallies, riots, and picketing—
were coded in calculating the dependent variables. This means that, to prevent inflation,
those activities deemed suspicious by the police but did not result in public events—such
as church services, meetings of dissidents, or regime-sponsored propaganda events—are
not coded as protest events. In order to prevent inflation in the apr, when local police
agencies o√ered an estimate of protest participation in a range (e.g., 1,000–2,000 per-
sons), the lowest figure in that range was consistently chosen.
Comparing the resulting list of protest events and participation estimates with reli-
Previous Page Next Page