preferences for democracy can develop under
limited democratization and how citizen-led
protest movements can be powerful even
without international support. However, the
strong Chinese reaction, U.S. and British capi-
tulation, and Hong Kong’suncertainfuture
raise questions about whether citizens alone
can successfully defend democracy in the cur-
rent international environment ( 16 ). Recently,
the 2020 Chinese security law quashed the
1Julyprodemocracyprotests,whichhavemarked
the anniversary of the handover of political con-
trol of Hong Kong to China since 1997 ( 52 ).
Although the Hong Kong case is unparalleled
in many ways, it illustrates how elite-driven
democratic backsliding was met with surpris-
ingly robust citizen-led resistance. Yet even
with massive public resistance to backsliding,
and likely because of declining global support
for democracy, political elites ordered a bla-
tantly and unapologetically brutal crackdown
against protestors. What would have happened
if international actors had quickly and force-
fully supported the protestors’cause?
Democracy’s new global headwinds will
likely have heterogeneous effects across types
of democracy, including opportunities to learn
about the consequences of pseudodemocracy
for citizens. Some consequences of backsliding
will depend on whether members of power-
sharing coalitions also gain from autocratiza-
tion ( 50 ) and whether business, military, and
religious elites are content with changes to their
statuses under democratic backsliding ( 12 ). For
citizens, the global turn away from democracy
could result in several divergent potential im-
plications that future research should consider.
One area for further study is how the col-
lective action potential of citizen movements
to defend democracy is influenced by varying
forms of autocratization. Thus far, documented
autocratization has been gradual and subtle
rather than sudden and blatant, which may
undermine citizens seeking to defend democ-
racy against backsliding ( 14 ):“[E]lections are
being hollowed out as autocracies find ways to
control their results while sustaining a veneer
of competitive balloting. Polls in which the
outcome is shaped by coercion, fraud, gerry-
mandering, or other manipulation are increas-
ingly common....indicators for elections have
declined at twice the rate of overall score totals...”
( 53 ). Similarly, Lührmann and Lindberg, dis-
cussing the challenge in measuring autocrat-
ization, have suggested that“[r]uling elites
shy away from sudden, drastic moves to auto-
cracy and instead mimic democratic institu-
tions while gradually eroding their functions”
( 14 ). Abrupt transitions to dictatorship, like
military coups, remain rare ( 14 ).
Thus, a critical juncture for the potential
power of citizen movements will be whether
the competitive veneer of democracy is sus-
tained or leaders drop the act entirely, includ-
ing the elimination of multiparty elections.
Once introduced, regular elections provide an
established focal point for collective action
to enforce democracy when it is threatened
( 54 , 55 ). Since the mid-1970s, some leaders
whosought to manage the introduction of
multiparty elections without losing their hold
onpowermiscalculatedinamannerthat
ultimately led to their downfall. Such downfalls
have included either losing outright, in what
Huntington has called“stunning”elections ( 20 ),
or cheating to win, getting caught, and facing
postelection protests and electoral revolutions,
often combined with international pressure to
step down or hold new elections ( 28 , 56 ).
Figure 3 shows the total number of elections
held annually over time, which fluctuates
naturally from year to year. But after steady
increases throughout the post–World War II
period, the number of countries holding elec-
tions hit a plateau and may be in decline.
Although 2016 matched the highest number
of elections ever held, in 2017 the number of
elections globally dropped to a number not
seen since the late 1980s. This could be a
result of idiosyncratic fluctuations that result
from differences in the length of electoral cy-
cles and parliamentary systems with variable
election timing, but it could also be related
to backsliding trends. As partial evidence of
backsliding, the lower line in Fig. 3 also shows
how many elections each year are character-
ized by government harassment of opposition
parties and/or media bias in favor of the in-
cumbent, both of which have increased since
the mid-2000s.
In 2020, the global pandemic gave many
governments a legitimate public health reason
to postpone or suspend elections, and these
suspensions may serve as effective cover for
some leaders to more permanently eliminate
elections, which have represented a power-ful risk to leaders’hold on power. Although
citizens could use the cancellation of elections
as a onetime focal pointto coordinate protest,
theperiodicnatureofelections—even in electoral
authoritarian regimes—facilitates a regular and
collectively anticipated opportunity for citizen
coordination that would be hard to replace if
elections were eliminated.
Thesecondareainwhichchangesinthe
international environment provide new re-
search opportunities relates to where citizen
support for democracy will prove more dura-
ble in the face of backsliding. Is exposure to
even superficial democratic institutions suffi-
cient to generate either elite-led backlash and/
or mass protests against democratic erosion?
Or are such reactions much more likely in
countries with long-term experience with democ-
racy? Recent research has suggested that threats
to democracy can, on average, cause increased
public support for democracy ( 57 ),butitisnot
yet clear whether all forms of backsliding are
similarly likely to trigger countermobilization.
Democracy protests and citizen-led mass mobili-
zation remain powerful forces ( 30 , 58 , 59 ). As
backsliding spread and international support
for democracy declined, citizens in some coun-
tries took to the streets to demand political
freedom, including the massive protests in
Algeria, Bolivia, and Sudan ( 11 ). Sudan’srecent
popular uprising unseated a dictator who had
withstood decades of international pressure,
including an International Criminal Court
indictment. Indeed, some of the notable cases
of democratic progress amidst the more gen-
eral trend of backsliding in 2019 were linked to
mass protests. It may not be easy for elites to
put the genie back in the bottle. But whether
citizen movements alone, absent external sup-
port, will be sufficient to check newly embold-
ened autocrats will not be apparent for years
to come.Hyde,Science 369 , 1192–1196 (2020) 4 September 2020 4of5
010203040506070Total number per year1945 1950 1955 19601965 1970 1975 19801985 19901995 2000200520102015Total elections per yearTotal elections with government harassment of opposition and/or media bias in favor of the incumbent
Smoothed trend line (lowess) Smoothed trend line (lowess)Fig. 3. Total national elections per year with increasing trends in opposition harassment and/or media bias,
1945 to 2018.There has been a recent plateau or ceiling in the total number of elections held each year from 1945
to 2018 (top lines). At the same time, there has also beena recent increase in elections characterized by
government harassment of their political opposition and/or media bias in favor of the incumbent candidate or party,
with a notable uptick since 2006. Smoothed trendlines overlay the annual counts. Data are from the NELDA project
( 9 ). Government harassment of opposition is measured as NELDA15, and media bias in favor of the incumbent
is measured as NELDA16. The full data and codebook are available at https://nelda.co/, and code and data used to
produce the figure are available from Harvard Dataverse ( 62 ). lowess, locally weighted scatterplot smoothing.