of dissidents appears to be on the rise ( 47 ). For
example, as the Arab Spring protests began
to diffuse from Tunisia, dictators—fearing
that similar protests would take hold in their
own countries—were able to adapt strategies
to more effectively suppress prodemocracy
protests ( 34 ). Although the protests in Tunisia
and Egypt succeeded in overthrowing dicta-
torial regimes, no other Arab Spring protest
movements that followed were able to over-throw their governments ( 34 ). Recent research
has documented Chinese and Russian strate-
gic censorship and internet filtering by key-
words that is aimed at minimizing the spread
of information about protests and inoculating
their governments from the diffusion of pro-
tests to their own citizens ( 36 ). Another recent
study has similarly documented regional diffu-
sion of illiberal norms that restrict civil society,
including cases in which restrictive new regu-
latory language was copied verbatim from
neighboring countries ( 46 ). Just as mass protests
have proven regionally and globally conta-
gious, inspiring citizens in other countries to
take to the streets, some authoritarian elites
have learned how to neutralize challenges to
their rule from the experiences of their peers
( 34 , 46 , 47 ). Similarly, extremist political parties
with authoritarian tendencies may emulate
rhetoric and tactics aimed at hollowing out
democracy in their countries, which represents
a globalized threat to democracy from below.
A third threat to democracy, led by Russia
and China, aims to undermine the interna-
tional norm of democracy and the legitimacy
of democracy promotion efforts in part by
privileging state security over individual rights
( 18 ). As the international norm of democracy
erodes and associated international benefits
decline, related prescriptive international
norms—like the expectation that leaders will
invite international election observers to scruti-
nize their elections ( 3 , 4 )—may also weaken. This
will result in fewer opportunities for international
actors to effectively penalize stolen elections ( 3 )
and the loss of a periodic opportunity for inter-
national support of democratic elections. Inter-
national election observation has been linked to
less election fraud ( 3 ) and more durable peace
agreements after civil war ( 49 ). These emergent
threats to democracy combine to create poten-
tially divergent consequences for heads of state,
other political elites, and citizens, as would-be
autocrats seek to manage potential elite and
mass challenges to their continued hold on
power ( 50 , 51 ) under shifting global constraints.Citizens and democratic backsliding in the
international environment
The recent, massive protests against democratic
erosion in Hong Kong, initiated in 2019, stand
out as a movement that would have been met
with substantial international support just
3 years earlier. Such support would likely have
included diplomatic pressure on leaders in
Hong Kong and China to moderate their stance
and rhetorical and material support for protes-
tors from democracies around the world. In-
stead, external support for the prodemocracy
movement was underwhelming. Even so,
hundreds of thousands of citizens were willing
to put their lives at risk to demand democracy
and protest curtailments of individual free-
doms. This stands as an example of how citizenHyde,Science 369 , 1192–1196 (2020) 4 September 2020 3of5
1976198919992019Electoral
democracy level>0.25-0.50-0.25>0.5-0.75
>0.75-1.0Fig. 2. Regional and temporal clusters in democracy and backsliding around the world.Cartograms
illustrate the regional and temporal diffusion of electoral democracy around the world in four snapshots of
independent states: 1976 (the beginning of the most recent wave of democratization, initiated in Portugal), 1989
(showing the regional clustering of transitions to democracy in Latin America), 1999 (showing the clusters of
transitions in Eastern Europe and sub-Saharan Africa), and 2019 (in which democracy scores have decreased
in many regions). The measure of democracy is the Electoral Democracy Index v10 from the Varieties of
Democracy project, which measures a minimalist definition of democracy and is available at http://www.v-dem.net/en/
data/data-version-10/. Replication data and code are available from Harvard Dataverse ( 62 ). Microstates and
nonindependent territories are not shown, as defined by Weidemanet al.( 37 ) in the CShapes dataset.
DEMOCRACY IN THE BALANCE