PreprintPDF Available

Did the Women's March Work? Re-Evaluating the Political Efficacy of Protest

May 2019

DOI:10.13140/RG.2.2.24332.95369

Authors:

University of Texas at Dallas

Preprints and early-stage research may not have been peer reviewed yet.

The Women's March on Washington and its over 600 "sister marches" were likely the single largest day of protest in American history to date, and were followed by a wave of political organizing that re-invigorated progressive politics after the 2016 election of President Donald Trump. Yet to what degree can we attribute the emergence of this Anti-Trump "Resistance" to the Women's Marches themselves? Do public protests such as the Women's March truly change political outcomes or do they simply reflect underlying public opinion? There is a growing literature arguing that protest has important effects independent of its endogenous relationship to public opinion. In this paper, I test this argument on the scale of the Women's March. I instrument Women's March participation using rainfall data from the day of the march and measure the effects of instrumented march size on three dependent variables: the creation of "Indivisible" groups, shifts in voting by congressional representatives, and the shift towards Democratic congressional candidates in the 2018 elections. I find that the instrumented size of Women's March protests significantly increased the Democratic vote share in the 2018 election. These findings provide strong evidence that the Women's Marches were a significant transformative event in American politics, with real political consequences, and speak to the power of peaceful protest as a social movement tactic.

Content uploaded by Jonathan Pinckney

Content may be subject to copyright.

Submission to 2019 Mobilization Conference Page 1 of 22

Did the Women’s March Work?

Re-Evaluating the Political Eﬃcacy of Protest

JONATHAN PINCKNEY

Norwegian U. of Science and Technology (NTNU)

e Women’s March on Washington and its over 600 “sister marches” were likely

the single largest day of protest in American history, and were followed by a wave

of political organizing that re-invigorated progressive politics after the 2016 election

of President Donald Trump. Yet to what degree can we attribute the emergence of this

Anti-Trump “Resistance” to the Women’s Marches themselves? Do public protests such as the

Women’s March truly change political outcomes or do they simply reﬂect underlying public

opinion? There is a growing literature arguing that protest has important eﬀects independent

of its endogenous relationship to public opinion. In this paper, I test this argument on the

scale of the Women’s March. I instrument Women’s March participation using rainfall data

from the day of the march and measure the eﬀects of instrumented march size on three

dependent variables: the creation of “Indivisible” groups, shifts in voting by congressional

representatives, and the shift towards Democratic congressional candidates in the 2018

elections. I ﬁnd that the instrumented size of Women’s March protests signiﬁcantly increased

the Democratic vote share in the 2018 election. These ﬁndings provide strong evidence that

the Women’s Marches were a signiﬁcant transformative event in American politics, with real

political consequences, and speak to the power of peaceful protest as a social movement

tactic.

Word Count: 5202

Jonathan Pinckney is a Post-Doctoral Research Fellow, Department of Sociology and Political Science, NTNU

(jonathan.pinckney@ntnu.no

The author would like to thank the Department of Sociology and Political Science at NTNU for generous

support, and Thea Johansen for excellent research assistance.

This is a very early draft working paper.

Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper Working Paper

Pinckney

INTRODUCTION

e Women’s March of January 21st, 2017 is one of the most prominent examples of public

protest in recent American history, and was the “trigger” for a massive outpouring of social

dissent that became known as “The Resistance.” (Meyer and Tarrow 2018). The size of

the protests was nearly unprecedented, and a signiﬁcant deviation from the norm of protests in the

United States in the recent past. Women’s March events averaged nearly 7,000 participants (Chenoweth

and Pressman 2017). For comparison, a representative sample of protests in the United States in the

recent past found protest events in the United States had a mean participation of 61 participants. The

average turnout was also more than six times higher than the April 15, 2009 Tax Day protests, which

jumpstarted the Tea Party movement (Beyerlein et al. 2018).

In the immediate aftermath of the protests, Democratic activists claimed that the marches indicated

widespread disapproval of President Trump’s policies, and could spark a movements against him.

Columnist Eugene Robinson observed: “The millions who participated nationwide now constitute the

kind of broad-based network that can be harnessed into eﬀective political action” (Robinson 2017).

In the 2018 mid-term elections, the ﬁrst chance for the national electorate to weigh in on the Trump

presidency, the Democratic party experienced signiﬁcant gains - winning the popular vote by more

than seven percent and capturing control of the House of Representatives. Many took the Democratic

shift as indicative in part of the power of the Resistance to achieve political change, and contrasted it

with earlier more diﬀuse movements such as the “Occupy Wall Street” movement, that mobilized large

numbers of people but failed to lead to major political changes.

Yet disentangling the eﬀects of protest is inherently challenging. To what degree did the Women’s

Marches eﬀect political change and to what degree were they simply indicative of a deeper shift in

public opinion in the country (or simply symptomatic of increasing partisan polarization). Does

political protest actually work? And if so, how? The social movements literature gives us conﬂicting

answers on the potential eﬀectiveness of protest, while the nonviolent resistance literature suggests that

peaceful protest can be eﬀective, but typically measures only highly aggregated outcomes and does not

adequately address questions of endogeneity and reverse causality.

In this paper I address this question. I look at the eﬀect of the Women’s Marches on three key areas:

Did the Women’s March Work?

movement building, as measured by the creation and size of “Indivisible” groups; policymaking, as

measured by shifts in legislator DW-NOMINATE scores, and electoral outcomes, as measured by the

shift in Democratic party vote share in 2018. I address the endogeneity of protest size to underlying

political dynamics through an instrumental variable analysis. I instrument the size of participation in

the 2017 Women’s March with a plausibly exogenous instrumental variable: the weather on the date on

the marches. I show ﬁrst that weather signiﬁcantly predicts decreased turnout, and is thus a reliable

instrument for predicting protest size that has no plausible alternative path to aﬀecting my dependent

variables. I show secondly that, when instrumented using good weather, large protests on the day of

the original Women’s March predicts signiﬁcantly increased movement activity, left-ward shifts in

congressional voting scores, and a greater swing to the Democrats in the 2018 midterm elections.

The paper has important implications for our understanding of the power of protest to bring about

political change, and the mechanisms through which that change occurs. I argue, following Madestam

and his co-authors 2013 that the power of protest comes primarily not from its ability to signal

discontent to elites, but rather with the eﬀects of protest on the protesters themselves. Participation in

major protest events such as the 2017 Women’s March socialized protest participants towards greater

left-wing attitudes, helped jump-start local organizing eﬀorts, and socialized the participants to greater

political participation. These combined eﬀects helped cement the growth of the anti-Trump “Resistance”

leading to signiﬁcant gains for liberal politics in the United States.

The paper proceeds as follows. In section 2 I discuss what we know so far about the political

eﬀectiveness of protest. In section 3 I discuss the origin and dynamics of the 2017 Women’s March. In

section 4 I introduce my research design, including the use of good weather as an instrument for protest

size. In section 5 I present and discuss my ﬁndings. Section 6 concludes.

THE POWER OF PROTEST

Existing Literature

The eﬀectiveness of protest is a core question for the study of extra-institutional politics. Social

movements’ scholars have long wrestled with the ways in which challengers from outside the polity

Pinckney

can challenge those within (Tilly 1978; Tarrow 1998).

Some scholars conclude that social movements

do have a strong impact on political outcomes (Baumgartner and Mahoney 2005; Piven 2006). For

example, Htun and Weldon (2012) ﬁnd that feminist social movements are the key factor driving

advancement in violent against women policies. Yet others ﬁnd little or no inﬂuence (Soule et al.

1999; Giugni 2007). For instance, McAdam and Su (2002) ﬁnd that the US anti-war movement in the

1960s had diﬃculty pressuring elites while also shifting public opinion. Branton et al. (2015) ﬁnd that

exposure to immigration protests shifted public opinion in favor of the protesters, but that this eﬀect

was limited to immigrants.

A more singularly optimistic tone comes from the literature on nonviolent resistance. Early

work from Sharp (1973) and Ackerman and Kruegler (1994) suggested that nonviolent resistance

movements, when skillfully employing tools of strategy and nonviolent discipline could bring about

major political transformations, even in the most unfavorable circumstances. Schock (2005) pointed

to a similar dynamic, arguing that so-called “unarmed insurrections” could oust dictatorships if they

could successfully foster points of leverage over those opposing regimes and deploy a dynamic micture

of tactics of concentration and dispersion. Building on this work, Chenoweth and Stephan (2011) show

that nonviolent resistance campaigns seeking “maximalist” goals of regime change, secession, or an

end to a military occupation succeeded in achieving their goals in roughly 50% of cases from 1900 to

2006.

Scholars are divided on the mechanisms whereby protest might aﬀect political outcomes. McAdam

and Su (2002) identify three mechanisms: disruptive protest, signaling and public opinion shift.

Chenoweth and Stephan (2011) emphasize the importance of participation, and how large, diverse

participation in protest campaigns can lead to both greater tactical innovation and more points of

connection with those in power, leading to greater opportunities to undermine “pillars of support”

(Helvey 2004). Similarly, Uba (2005) ﬁnds that the size and degree of disruption of anti-privatization

protests in India signiﬁcantly slowed the pace of privatization.

In a classic treatment of the subject, Lohmann (1993) argues that protest acts as a signal to political

For an excellent summary of major research eﬀorts on the eﬀectiveness and long-term outcomes of social

movements, see Amenta et al. (2010)

Did the Women’s March Work?

elites of underlying shifts in public opinion. Size is a clear indicator of the strength of this signal, with

larger protests acting as stronger signals that the underlying demands of the protest are widely shared.

Protest may also lead to political change through the eﬀects that it has on its participants.

Participation in protest tends to lead to more political participation in the future, as well as socialization

into the attitudes of the protest group (Opp and Kittel 2010). Protest participants have a powerful

experience and potentially a new social network on which to draw while they engage in future political

action. This level of organization and socialization can then lead to greater pressure on elected oﬃcials.

Finally, protest can aﬀect political outcomes by imposing direct economic costs on its opponents.

Labor strikes and other methods of noncooperation are the most prominent form of action that rely on

this mechanism. I expect that public demonstrations such as the Women’s March would be unlikely to

operate through this mechanism. Peaceful public demonstrations are often intentionally designed to

not disrupt economic activity. The Women’s Marches were no exception, with close cooperation with

police and attempts to minimize the economic disruption of the event. Events with high turnout may

even have boosted economic activity in their surrounding environment.

Challenges to Inference

Both the social movements literature and the nonviolent resistance literature have tended to focus on

larger movements or campaigns, and given less attention to the disaggregated impact of particular

protest events. Even most analyses based on protest events data tend to aggregate events to protest

counts temporally or geographically. For example, McAdam and Su (2002) look at the count of various

types of protest events aggregated to the national level. Walgrave and Vliegenthart (2012) aggregate all

demonstrations on a particular issue area in Belgium when looking at agenda-setting eﬀects of protest.

Gillion (2012) disaggregates protests in the United States by congressional district, but aggregates

them temporally to an annual level, and Banaszak and Ondercin (2016) aggregate protests on women’s

movement issues across the United States by quarter.

This tendency towards aggregation makes it diﬃcult to tease out the causal relationships between

In an experimental study of the eﬀects of protest on policymakers, Wouters and Walgrave (2017) ﬁnd that

numbers and agreement within the protest group are the most persuasive features of a protest for policymakers

Pinckney

tactics and political outcomes. We have strong evidence that nonviolent resistance can work, and that

social movements may, dependent on the political opportunity structure, and availability of resonant

frames, lead to signiﬁcant political change. Yet it is diﬃcult to say exactly why, since our independent

variables are measured at such a high level of aggregation. We know a lot about movements, but less

about speciﬁc protests in particular locations.

In addition to the aggregation problem, most studies of protest face a fundamental endogeneity

problem that is often ignored or only partially addressed. The occurrence and intensity of protest are

unlikely to indicate a fully exogenous shock to the system but rather part of an ongoing process of

political contention. Many of the observable correlates of protest and political eﬀectiveness overlap.

Insofar as these correlates can be measured, this issue can be remedied through the inclusion of

appropriate control variables. However, many of the factors that plausibly inﬂuence both protest and

political outcomes are not so easily observed. Thus, correlations between protest and political shifts in

the direction of the protesters are empirically suspect.

Recent work has begun to address these challenges, but leaves many questions unanswered. Biggs

and Andrews (2015) ﬁnd that lunch counter sit-ins made desegregation signiﬁcantly more likely in

cities across the American South in 1960, controlling for several determinants of protest. Yet their

measure of protest is a simple binary variable that does not allow for disaggregation in terms of size or

intensity of protest.

Madestam et al. (2013) show that protest size on Tax Day, 2009, which they instrument using

rainfall, signiﬁcantly predicts several indicators of movement success for the Tea Party. Counties with

large Tea Party protests saw large follow-up groups, more donations to major right-wing PACs, and a

swing toward the Republicans in the 2010 mid-term elections. Their strategy addresses many of the

methodological challenges of determining the impact of individual protest events, but the limited scale

of the Tea Party tells us little about protest more broadly.

Omar Wasow (2017) examines the attributes of eﬀective protest in more detail with an examination

of the civil rights movement in the 1960s. Wasow shows that peaceful protests signiﬁcantly increased

public attention to the issue of civil rights, and that counties within 100 miles of a peaceful protest

had major shifts towards the Democratic party in the 1968 presidential election, while counties within

Did the Women’s March Work?

100 miles of violent riots in the aftermath of the assassination of Dr. Martin Luther King, Jr. had

signiﬁcant shifts towards the Republicans. Simulating the 1968 election with and without peaceful and

violent protests, Wasow argues there is good reason to believe that the post-assassination riots played a

crucial role in giving an election victory to Richard Nixon. Yet Wasow does not include any measures

of protest intensity to distinguish between the eﬀects of small protests and large protests.

In this paper, I build on the insights from these early papers looking at the impact of speciﬁc protest

events through examining one of the most recent major protest events: the 2017 Women’s March on

Washington and its associated “sister protests.”

The Women’s March provides an ideal environment in which to evaluate the power of protest. First,

the march was an extreme outlier in American protest politics, both in size and dispersion. Thus, it

provides a strong “best-case scenario” test of the power of protest. If protest can be eﬀective, the

Women’s March protests should be eﬀective. If the Women’s March failed to have a signiﬁcant political

impact, it provides strong evidence for the skeptics of the power of protest. Second, the large number

of events taking place on the same day provides an ideal environment for a natural experiment. Some

localities experienced the “treatment” of a Women’s March on January 21, 2017, and others did not.

Once I have addressed the endogenous aspects of generating the protest itself (which I discuss below in

section 4), this simultaneous shock provides strong grounds for causal inference.

What eﬀects should we expect based on the character of the Women’s March? I answer this by

examining the precipitating causes of the march and its immediate aftermath.

THE 2017 WOMEN’S MARCH

Planning for a “Million Woman March” on Washington to protect women’s rights began immediately

following the 2016 election of US President Donald Trump. Teresa Shook, a retired lawyer from

Hawaii, created a Facebook group planning an event after discussions on the pro-Hillary Clinton

Facebook Group “Pantsuit Nation.” The event, publicized in the midst of the beginning of a wave of

protest against Trump’s election, grew rapidly, with over 10,000 people saying they would participate

in the ﬁrst 24 hours and over 100,000 soon afterwards (Stein 2017).

While the march began with little organizational backing, it grew quickly thanks to assistance from

Pinckney

several diﬀerent “organizational tributaries” (Berry and Chenoweth 2018). The eventual “Women’s

March on Washington” had an organization behind it, and over 400 co-sponsors from several diﬀerent

strains of the American left.

In addition to the primary event in Washington, DC, however, over 600 “sister marches” were

organized across the United States and indeed around the world. A comprehensive tally of events

from the day of the Women’s March includes everything from the estimated 1,600,000 who attended

the three largest marches in DC, New York, and Los Angeles, to ten brave souls in the tiny town of

Adak, Alaska, a village on a remote Aleutian island that has the distinction of being the Westernmost

municipality in the United States.

The march was not explicitly partisan, and the organizers went to some lengths to ensure that their

organization was not seen as simply an arm of the Democratic party. However, the causes around which

march participants organized were overwhelmingly on the left of the American political spectrum

(Fisher et al. 2017).

The marches were also highly geographically dispersed. While the largest events took place in

major left-leaning urban centers, marches took place across the country, in small relatively rural

areas and in purple and red states as well as blue states. In a study of a representative sample of

Women’s Marches, McKane and McCammon (2018) found that marches were most likely to occur

in Democratic cities in predominately Republican states, and in places with a large existing social

movement organization infrastructure, however, the eﬀects of these variables are fairly small, indicating

just how widespread the marches were. The marches also varied signiﬁcantly in terms of their levels

of attendance. Beyerlein and his co-authors 2018 estimate that while the largest protest had roughly

750,000 to a million attendees, the average was roughly 7,000, and around a quarter of events had fewer

than 100 participants.

Anecdotally, the Women’s Marches played a signiﬁcant role in jump-starting the anti-Trump

opposition. They were the ﬁrst in a series of semi-regular major protest marches around various

themes that took place during the ﬁrst two years of the Trump presidency. Major protests took place

at American airports in support of Muslim refugees after the issuing of the ﬁrst “Muslim ban,” and

included the “March for Science,” the “People’s Climate March” and widespread marches focused on

Did the Women’s March Work?

healthcare during the summer of 2017 when the Republican-led congress was attempting to repeal the

Aﬀordable Care Act. The Women’s March itself has become an annual tradition, with widespread

demonstrations on the anniversary in 2018 and 2019. While the numbers in these follow-up marches

have not matched the overwhelming numbers in the initial 2017 protest, they have remained some

of the largest days of protest in American history. For instance, the 2019 Women’s March, while

largely described in the media as a failure due to divisions in the Women’s March organization and

accusations of anti-Semitism among the leadership of the Women’s March organization, actually had

more participants than the 2009 “Tax Day” protests that initiated the Tea Party movement (Chenoweth

and Pressman 2019).

The impact of the Women’s March on the subsequent anti-Trump “Resistance” can be seen clearly

in the demographic characteristics of the resistance. As Putnam and Skocpol (2018) identify, the

resistance is a movement that tends to be dominated not by the traditional “activist class” of the

young, highly-educated, and urban. Instead, the movement has been spearheaded by local groups of

predominately middle-aged educated white women in the suburbs, similar to the dominant demographics

of the Women’s March identiﬁed by Fisher et al. (2017).

Unlike prior progressive movements such as “Occupy Wall Street,” the anti-Trump resistance has

been focused on electoral change from the outset. As Putnam and Skocpol (2018) report: “Many of

the local groups whose emergence was linked to Indivisible and the March a year ago are already ten

months into an electoral‘turn.’ They have one election cycle under their belt and concrete targets in

their sights for 2018, 2019, and 2020.”

Hypotheses on Women’s March Eﬀects

One key beneﬁt of large protests is as mobilizing moments for building larger movement infrastructures.

Eﬀective protests should not simply remain on the streets but instead turn into long-term organizing

for future events. Thus, if the Women’s March was eﬀective, we should see it translating into more

grassroots organization in the aftermath of the event. Since the Women’s March was closely associated

with the anti-Trump “Resistance” movement, if the march was eﬀective in movement-building we

should expect to see more movement-related activities in locations where large Women’s Marches

Pinckney

occurred.

: Counties with large 2017 “Women’s Marches” will have more “Resistance” activities than

those without such marches.

While most Resistance groups do not explicitly identify themselves as members of the Democratic

party, and encourage membership from Independents and Republicans, the substance of their agenda

tends to be quite closely associated with a progressive agenda. Resistance groups have also been directly

involved in recruiting Democratic candidates to run for oﬃce, interfacing with local Democratic party

infrastructures. Thus, the ﬁrst outcome that I measure is whether the Women’s March impacted the

Democratic Party’s vote share in the 2018 election.

: Counties with large 2017 “Women’s Marches” will have larger 2018 Democratic vote shares,

ceteris paribus.

One key metric for the eﬀectiveness of protest is whether policy changes are enacted in line with

the preferences of the protesters. Participants in the Women’s March did not speak with a single voice,

or focus on a single issue. However, the issues motivating the majority of the participants to attend tend

to be associated with more left-wing politics in the United States. In a random sample of participants in

the Women’s March Fisher et al. (2017, 2) found that the top ﬁve reasons participants gave for attending

the Women’s March were Women’s Rights, Equality, Reproductive Rights, the Environment, and

Social Welfare. If the Women’s March was eﬀective in motivating changes in policymaker behavior,

we should expect to see elected oﬃcials voting in a stronger left-leaning direction.

: Representatives from districts with large 2017 “Women’s Marches” will vote more left-wing

than those without such marches, ceteris paribus.

Did the Women’s March Work?

RESEARCH DESIGN

Independent Variables

My primary independent variables are the occurrence and size of a women’s march on January 21,

2017. My data source for the occurrence of a march and the number of marchers comes from the

Erica Chenoweth and Jeremy Pressman’s Crowd Counting Consortium data, which records 656 distinct

marches with between one and 725,000 total participants, including eight marches with over 100,000

participants. The Crowd Counting Consortium data is based on aggregating multiple sources, including

media reports (which in turn primarily rely on police or government estimates of protest size), social

media posts, and activist self-reporting.

In my primary tests I look at the number of marchers per capita in a county or congressional district.

My population estimates (and thus my calculation of the per capita number of marchers) comes from

the US census.

Dependent Variables

I look at two speciﬁc avenues through which we can measure protest eﬀectiveness: movement-building

and political change. Within political change I look at both shifts in policymaking and in future

electoral outcomes.

For movement-building, I look at the creation, size, and activity of groups associated with the

“Indivisible” movement (Brooker 2018). Indivisible was started in December 2016 by a group of

congressional staﬀers interested in spreading eﬀective strategies of political engagement for people

opposed to the “Trump agenda.” Their “Indivisible: A Practical Guide for Resisting the Trump Agenda”

encouraged concerned citizens to create local groups that would pressure elected oﬃcials to resist the

Trump agenda (Bethea 2016). Indivisible groups often became the most prominent part of the activist

space directly devoted to anti-Trump resistance.

I selected Indivisible groups as my measure of Resistance activity for several reasons. First, I was

interested in the origins of new social movement activity, rather than mobilization through existing

social movement organizations. Second, Indivisible is one of the few national networks of grassroots

Pinckney

organizations focused on the “Resistance agenda.” Third, summary data on the geographic dispersion

and size of groups was easily available for analysis.

My data on Indivisible groups comes from the listing of Indivisible groups on the main Indivisible

website.

For rough estimates of the size of groups I investigated the public social media accounts

associated with each group (typically a Facebook page) and recorded the number of members, as well

as the levels of activity (frequency of posts and date of the most recent post). While the number of

members in a Facebook group is certainly not a direct reﬂection of the absolute number of members

active in a group, it is a reasonable rough proxy for the number of people in a local area who have

expressed some interest in the Resistance and a ceiling on the number of likely active participants in

Resistance activism in a local area.

For groups where no public social media data was available, a research assistant contacted the

group at the email address publicly listed on the Indivisible website and requested information on the

group’s founding date, number of members, and the date of their most recent meeting. If Hypothesis 1

is correct, I expect that counties with Women’s Marches should be much more likely to have Indivisible

groups, and that these groups will in turn have larger numbers of members reported on social media

and be more active.

For electoral outcomes, I look at the county or congressional-district vote share for the Democratic

party in the 2018 elections to the House of Representatives. My data come from Dave Leip’s election

atlas of the United States. If Hypothesis 2 is correct, I expect that the number of Democratic votes in

the 2018 election should be signiﬁcantly higher in counties and congressional districts that experienced

larger Women’s Marches.

For policy shifts, I look at the DW-NOMINATE scores of congressional representatives in the

115th Congress (From January 2017 - January 2019), controlling for their score in the 114th congresss.

DW-NOMINATE assigns scores along two ideological dimensions for each legislator based on roll-call

voting. I select their scores along DW-NOMINATE’s ﬁrst dimension, which is roughly equivalent to a

“left-right” or “liberal-conservative” spectrum (Poole 2005). The DW-NOMINATE scores come from

the Voteview project at the University of California: Los Angeles (Lewis et al. 2019). If Hypothesis

3www.indivisible.org

Did the Women’s March Work?

3 is correct, I expect that house members in districts with large Women’s Marches should shift their

DW-NOMINATE scores leftwards in response to the pressure

Control Variables

I control for several plausible alternative explanations for both leftward political shifts in policymaking

and election results and Resistance movement-building. First, I control for the democratic vote share in

elections to the House of Representatives in 2014. I control for three demographic characteristics: the

percentage of the population that identiﬁes as white, black, and hispanic, based on census data. I also

control for several economic indicators: median income, the unemployment rate, and the poverty rate.

Instrumental Variable Analysis

As described above, one of the key drawbacks in much of the existing work on the eﬀectiveness of

protest is that, insofar as studies examining the impact of discrete protest events have been done, they

are not able to fully account for the fact that the occurrence and size of protest is endogenous to

existing political conditions. Thus any ﬁndings on the eﬀectiveness of protest are subject to potential

spuriousness. Studies have typically attempted to address this through the use of control variables to

close oﬀ other observable explanations. However, this does not address potential unobservable factors

inﬂuencing both protest size and political outcomes.

To address this objection, in my analysis of the Women’s March I use a two-stage least-squares

model with instrumental variables. Instrumental variables are the typical econometric technique for

addressing the problem of endogenous independent variables. An instrumental variable aﬀects an

endogenous independent variable, but is itself exogenously assigned and only aﬀects the dependent

variable through its eﬀect on the independent variable. While instrumental variables are common in

economics, they are less frequently employed in social movement studies.4

I use average rainfall on January 21, 2017 as my instrument for the size of Women’s March protests.

A search on Google Scholar of the terms "protest" and "instrumental variable" returned only seven articles from

the American Sociological Review, ﬁve articles from Social Forces, and a single article from Mobilization.

Pinckney

The use of rainfall as an exogenous predictor of public dissent, from protests and riots to organized

violence, is well-established (Madestam et al. 2013; Ritter and Conrad 2016; Wasow 2017). Heavy rain

and other forms of bad weather increase the personal discomfort and cost of engaging in protest. On

rainy days fewer people are likely to turn out for a protest. For example, Hong Kong’s pro-democracy

“Umbrella Revolution” sit-in was signiﬁcantly demobilized when its titular accessories failed to protect

activists from sustained torrential downpours (Wan 2014). More recently, turnout for the 2019 third

annual Women’s March was signiﬁcantly depressed by wintry weather associated with Winter Storm

Harper (Ortiz 2019).

Rainfall on the day of the original Women’s March is an ideal instrument for capturing the exogenous

impact of protest size as well because it intuitively satisﬁes the “exclusion restriction,” that is to say it

has no plausible alternative connection to the longer-term outcomes I am measuring. While rainfall on

election day may signiﬁcantly eﬀect voter turnout and swing elections (Gomez et al. 2007; Shachar

and Nalebuﬀ 1999),rainfall nearly two years before an election does not. Nor does a single day of rain

plausibly aﬀect the formation of activist groups and congressional voting patterns, except insofar as it

aﬀects critical events on that day: the Women’s March itself.

My rainfall data comes from the National Oceanic and Atmospheric Administration (NOAA). NOAA

collects data on several weather-related indicators from its more than 12,000 weather stations across

the United States. I use NOAA’s data on weather stations’ location to create a grid of weather-station

area polygons based on Voronoi tessellation (Voronoi 1908). The Voronoi tessellation algorithm draws

the polygons around each weather station such that no point within a weather station’s polygon is closer

to any other weather station. To generate the weighted average rainfall across a county or congressional

district I sum up the rainfall reported from each weather station whose Voronoi polygon intersects the

county or congressional district in question, and then average them, weighting the average based on the

percentage of the county or congressional district’s area accounted for by each Voronoi polygon.5

This provides a more accurate weather estimate than simply giving equal weight to the average

rainfall estimates from all weather stations inside a county or congressional district, as Madestam et al.

Thanks to Martin Smidt for suggesting the use of the Voronoi tessellation algorithm to calculate the geographic

coverage of weather stations.

Did the Women’s March Work?

(2013) do, and addresses the problem that some small urban counties often do not have any weather

stations inside their boundaries.

Table 1 shows summary statistics for the main variables used in the county-level analysis.

TABLE 1. Summary Statistics

Statistic N Mean St. Dev. Min Max

Marchers 3,141 1,089.306 13,940.330 0 451,223

Marchers (per capita) 3,141 0.002 0.014 0 0

Rainfall (in) 3,134 0.148 0.339 0.000 3.425

Dem. Vote Share 2018 3,110 0.361 0.178 0.000 1.000

Dem. Vote Share 2014 3,105 0.330 0.187 0.000 1.000

Percent White 3,141 0.847 0.164 0.039 0.993

Percent Black 3,141 0.101 0.146 0.001 0.864

Percent Hispanic 3,141 0.094 0.137 0.005 0.963

Total Population 3,141 103,111.600 331,986.400 88 10,157,032

Unemployment Rate 3,140 4.619 1.676 1.600 20.100

FINDINGS

At this stage, data collection is only fully complete for the election models. Thus, I am only able to

present preliminary results on the eﬀects of the Women’s March on the 2018 Democratic vote share.

Table 2 contains four models presenting these early results. Model 1 is a “naive” model showing

the direct eﬀect of participants in the 2017 Women’s March per capita on the 2018 Democratic vote

share, with no attend to take into account the endogenous elements of Women’s March size. There is a

positive and signiﬁcant impact, however, this result is questionable because of the likely endogeneity

problem.

Model 2 shows the ﬁrst stage model in my instrumental variable analysis, testing whether rainy

Pinckney

TABLE 2. Democratic Vote Share Models

Dependent variable:

Dem Share 2018 Marchers (pc) Dem Share 2018

OLS OLS instrumental

variable

(1) (2) (3) (4)

Marchers (pc) 0.972∗∗∗ 12.281∗18.925∗∗∗

(0.121) (4.866) (5.117)

Rainfall 0.002∗∗

(0.001)

Rural/Urban 0.055∗∗∗ 0.001∗0.042∗∗∗ 0.039∗∗∗

(0.003) (0.001) (0.009) (0.006)

Dem. Vote Share 2014 0.703∗∗∗ 0.017∗∗∗ 0.518∗∗∗ 0.610∗∗∗

(0.010) (0.001) (0.082) (0.029)

Percent White −0.229∗∗∗ −0.008∗−0.137∗−0.159∗∗∗

(0.022) (0.003) (0.059) (0.035)

Percent Hispanic 0.077∗∗∗ 0.002 0.056∗0.062∗∗∗

(0.012) (0.002) (0.025) (0.016)

Percent Black −0.097∗∗∗ −0.014∗∗∗ 0.051 0.015

(0.023) (0.003) (0.078) (0.045)

Unemployment Rate −0.00002 −0.001∗∗∗ 0.009∗0.004∗

(0.001) (0.0002) (0.005) (0.002)

Constant 0.304∗∗∗ 0.008∗0.210∗∗∗ 0.231∗∗∗

(0.023) (0.003) (0.060) (0.037)

Observations 3,105 3,103 3,103 3,071

R20.738 0.061 0.005 0.555

Adjusted R20.737 0.059 0.002 0.554

Note: ∗p<0.05; ∗∗p<0.01; ∗∗∗p<0.001

Did the Women’s March Work?

weather is a signiﬁcant predictor of marchers per capita. Rainfall is indeed a signiﬁcant predictor,

though the sign on the coeﬃcient is not in the expected direction.

Models 3 and 4 present initial results of the instrumental variable analysis. Model 3 replicates

the naive model reported in Model 1, simply replacing Marchers per capita with marchers per capita

instrumented by rainfall. The coeﬃcient is positive and signiﬁcant at the p < 0.05 level.

Model 4 reﬁnes Model 3 by removing outliers from the sample. Ninety-nine percent of counties

have marchers per capita of less than ﬁve percent. However, a small number of counties have numbers

much higher than this. The county with the highest number of marchers per capita is Washington

County, Vermont, where the more than 17,000 reported participants in the Montpelier Women’s March

accounted for nearly 30% of the population. Since these outliers are so extreme relative to the total

population, it is plausible that they might be driving any results. Thus in Model 4 I remove the 16

observations more than six standard deviations above the mean marchers per capita, which is roughly

equivalent to any observation with a number of marchers per capita above 0.09.

The result increases both the size and level of signiﬁcance for the instrumented measure of marchers

per capita, suggesting that the extreme outliers were obscuring the eﬀectiveness of the Women’s

Marches rather than driving it.

CONCLUSION

This paper has used the 2017 Women’s March as an example of a single “shock” event that allows for

signiﬁcant geographic disaggregation. In addition, it addresses the endogeneity of protest to current

political conditions through an instrumental variable analysis using rainfall as an exogenous predictor

of protest participation.

The early analysis provides some support for the argument that political protest can be broadly

eﬀective in leading to political change, and that a key factor in that eﬀectiveness is protest size. The

number of marchers per capita, instrumented by rainy weather, has a strong positive and signiﬁcant

relationship with the Democratic vote share in 2018, controlling for relevant political and demographic

predictors (most importantly the previous mid-term election results).

Only the ﬁrst stage of analysis for this project is currently completed. The next steps are to measure

Pinckney

the policy impact of the Women’s Marches by looking at their eﬀects on policymakers voting records.

I am currently planning to incorporate the liberal-conservative dimension of the DW-NOMINATE

scores for this purpose, but am also looking into voting scorecards from Planned Parenthood and other

Women’s Rights and reproductive rights organizations.

In addition, at this stage I am missing the crucial intervening aspect of movement-building through

which I have argued the Women’s Marches were likely to have their eﬀect. Data collection on the

Indivisible movement is ongoing, allowing tests of whether the Women’s March signiﬁcantly increased

“Resistance” mobilization in the intervening period between the 2017 marches and the 2018 election.

The rainfall instrument has several positive qualities, but also some drawbacks. In particular, the

date in question had remarkably good weather across much of the country. Thus the number of “rainy”

counties is somewhat small. Future analysis could incorporate other measures of weather likely to

inﬂuence protest turnout, for example deviations from typical temperature.

While the analysis is still only in its very early stages, it provides suggestive evidence for the power

of political protest to aﬀect change. Political protest, in particular large protests of the type seen in the

2017 Women’s Marches, can have signiﬁcant political impacts.

REFERENCES

Ackerman, Peter and Christopher Kruegler (1994). Strategic Nonviolent Conﬂict, The Dynamics of People Power

in the Twentieth Century. Boston, MA: Praeger.

Amenta, Edwin , Neal Caren, Elizabeth Chiarello, and Yang Su (2010). The Political Consequences of Social

Movements. Annual Review of Sociology 36(1), 287–307.

Banaszak, Lee Ann and Heather L. Ondercin (2016, September). Public Opinion as a Movement Outcome: The

Case of the U.S. Women’s Movement. Mobilization: An International Quarterly 21(3), 361–378.

Baumgartner, Frank R. and Christine Mahoney (2005). Social movements, the rise of new issues, and the public

agenda. In D. S. Meyer, V. Jenness, and H. M. Ingram (Eds.), Routing the Opposition: Social Movements,

Public Policy, and Democracy, pp. 65–86. Minneapolis, MN: University of Minnesota Press.

Berry, Marie and Erica Chenoweth (2018). Who Made the Women’s March? In D. S. Meyer and S. Tarrow

Did the Women’s March Work?

(Eds.), The Resistance: The Dawn of the Anti-Trump Opposition Movement, pp. 75–89. New York, NY: Oxford

University Press.

Bethea, Charles (2016, December). The Crowdsourced Guide to Fighting Trump’s Agenda.

Beyerlein, Kraig , Peter Ryan, Aliyah Abu-Hazeem, and Amity Pauley (2018). The 2017 Women’s March: A

National Study of Solidarity Events. Mobilization: An International Quarterly 23(4), 425–449.

Biggs, Michael and Kenneth T. Andrews (2015, April). Protest Campaigns and Movement Success: Desegregating

the U.S. South in the Early 1960s. American Sociological Review 80(2), 416–443.

Branton, Regina , Valerie Martinez-Ebers, Tony E. Carey, and Tetsuya Matsubayashi (2015, February). Social

Protest and Policy Attitudes: The Case of the 2006 Immigrant Rallies. American Journal of Political

Science 59(2), 390–402.

Brooker, Megan E. (2018). Indivisible: Invigorating and Redirecting the Grassroots. In D. S. Meyer and

S. Tarrow (Eds.), The Resistance: The Dawn of the Anti-Trump Opposition Movement, pp. 162–184. New

York, NY: Oxford University Press.

Chenoweth, Erica and Jeremy Pressman (2017, February). This is What We Learned by Counting the Women’s

Marches. The Washington Post.

Chenoweth, Erica and Jeremy Pressman (2019, February). Analysis | The 2019 Women’s March was bigger than

you think. https://www.washingtonpost.com/news/monkey-cage/wp/2019/02/01/the-2019-womens-march-

was-bigger-than-you-think/.

Chenoweth, Erica and Maria J. Stephan (2011). Why Civil Resistance Works: The Strategic Logic of Nonviolent

Conﬂict. New York, NY: Columbia University Press.

Fisher, Dana R. , Dawn M. Dow, and Rashawn Ray (2017, September). Intersectionality takes it to the streets:

Mobilizing across diverse interests for the Women’s March. Science Advances 3(9), eaao1390.

Gillion, Daniel Q. (2012, October). Protest and Congressional Behavior: Assessing Racial and Ethnic Minority

Protests in the District. The Journal of Politics 74(4), 950–962.

Pinckney

Giugni, Marco (2007). Useless protest? A time-series analysis of the policy outcomes of ecology, antinuclear,

and peace movements in the United States, 1977-1995. Mobilization: An International Quarterly 12(1),

53–77.

Gomez, Brad T. , Thomas G. Hansford, and George A. Krause (2007). The Republicans should pray for rain:

Weather, turnout, and voting in US presidential elections. The Journal of Politics 69(3), 649–663.

Helvey, Robert L. (2004). On Strategic Nonviolent Conﬂict: Thinking about the Fundamentals. Albert Einstein

Institute.

Htun, Mala and S. Laurel Weldon (2012, August). The Civic Origins of Progressive Policy Change: Combating

Violence against Women in Global Perspective, 1975–2005. American Political Science Review 106(3),

548–569.

Lewis, Jeﬀrey B. , Keith Poole, Howard Rosenthal, Adam Boche, Aaron Rudkin, and Luke Sonnet (2019).

Voteview: Congressional Roll-Call Votes Database. https://voteview.com/.

Lohmann, Susanne (1993). A Signaling Model of Informative and Manipulative Political Action. The American

Political Science Review 87(2), 319–333.

Madestam, Andreas , Daniel Shoag, Stan Veuger, and David Yanagizawa-Drott (2013). Do Political Protests

Matter? Evidence from the Tea Party Movement. Quarterly Journal of Economics 128(4), 1633–1685.

McAdam, Doug and Yang Su (2002). The war at home: Antiwar protests and congressional voting, 1965 to

1973. American sociological review 67(5), 696–721.

McKane, Rachel G. and Holly J. McCammon (2018). Why we March: The Role of Grievances, Threats,

and Movement Organizational Resources in the 2017 Women’s Marches. Mobilization: An International

Quarterly 23(4), 401–424.

Meyer, David S. and Sidney Tarrow (2018). Introduction. In D. S. Meyer and S. Tarrow (Eds.), The Resistance:

The Dawn of the Anti-Trump Opposition Movement, pp. 1–23. New York, NY: Oxford University Press.

Opp, Karl-Dieter and Bernhard Kittel (2010, February). The Dynamics of Political Protest: Feedback Eﬀects and

Interdependence in the Explanation of Protest Participation. European Sociological Review 26(1), 97–109.

Did the Women’s March Work?

Ortiz, Jorge L. (2019, January). Women’s March 2019: Activists press on for Saturday protest despite stiﬀ

headwinds. USA Today Online.

Piven, Frances Fox (2006). Challenging Authority: How Ordinary People Change America. Rowman &

Littleﬁeld Publishers.

Poole, Keith T. (2005). Spatial Models of Parliamentary Voting. New York, NY: Cambridge University Press.

Putnam, Lara and Theda Skocpol (2018, February). Middle America Reboots Democracy.

https://democracyjournal.org/arguments/middle-america-reboots-democracy/.

Ritter, Emily Hencken and Courtenay R. Conrad (2016, February). Preventing and Responding to Dissent: The

Observational Challenges of Explaining Strategic Repression. American Political Science Review 110(1),

85–99.

Robinson, Eugene (2017, January). Make the march a movement. The Washington Post, A15.

Schock, Kurt (2005). Unarmed Insurrections: People Power Movements in Nondemocracies, Volume 22.

Minneapolis, MN: U of Minnesota Press.

Shachar, Ron and Barry Nalebuﬀ (1999). Follow the leader: Theory and evidence on political participation.

American Economic Review 89(3), 525–547.

Sharp, Gene (1973). The Politics of Nonviolent Action. Boston, MA: Porter Sargent.

Soule, Sarah , Doug McAdam, John McCarthy, and Yang Su (1999). Protest events: Cause or consequence of

state action? The US women’s movement and federal Congressional activities, 1956-1979. Mobilization: An

International Quarterly 4(2), 239–256.

Stein, Perry (2017, January). The woman who started the Women’s March with a Facebook post reﬂects: ‘It was

mind-boggling’. https://www.washingtonpost.com/news/local/wp/2017/01/31/the-woman-who-started-the-

womens-march-with-a-facebook-post-reﬂects-it-was-mind-boggling/.

Tarrow, Sidney (1998). Power in Movement: Social Movements and Contentious Politics. New York, NY:

Cambridge University Press.

Tilly, Charles (1978). From Mobilization to Revolution. Boston, MA: Routledge.

Pinckney

Uba, Katrin (2005). Political protest and policy change: The direct impacts of Indian anti-privatization

mobilizations, 1990-2003. Mobilization: An International Quarterly 10(3), 383–396.

Voronoi, Georges (1908). Nouvelles applications des paramètres continus à la théorie des formes quadratiques.

Deuxième mémoire. Recherches sur les parallélloèdres primitifs. Journal für die reine und angewandte

Mathematik 134, 198–287.

Walgrave, Stefaan and Rens Vliegenthart (2012). The Complex Agenda-Setting Power of Protest: Demonstrations,

Media, Parliament, Government, and Legislation in Belgium, 1993-2000. Mobilization: An International

Quarterly 17(2), 129–156.

Wan, William (2014, October). Exhaustion saps fervor of Hong Kong protests. The Washington Post, A13.

Wasow, Omar (2017). Do Protests Matter? Evidence from the 1960s Black Insurgency. Working Paper, 63.

Wouters, Ruud and Stefaan Walgrave (2017). Demonstrating Power: How Protest Persuades Political Represen-

tatives. American Sociological Review 82(2), 361–383.

ResearchGate has not been able to resolve any citations for this publication.

Intersectionality takes it to the streets: Mobilizing across diverse interests for the Women’s March

Article

Full-text available

Sep 2017

Can a diverse crowd of individuals whose interests focus on distinct issues related to racial identity, class, gender, and sexuality mobilize around a shared issue? If so, how does this process work in practice? To date, limited research has explored intersectionality as a mobilization tool for social movements. This paper unpacks how intersectionality influences the constituencies represented in one of the largest protests ever observed in the United States: the Women’s March on Washington in January 2017. Analyzing a data set collected from a random sample of participants, we explore how social identities influenced participation in the Women’s March. Our analysis demonstrates how individuals’ motivations to participate represented an intersectional set of issues and how coalitions of issues emerge. We conclude by discussing how these coalitions enable us to understand and predict the future of the anti-Trump resistance.

Demonstrating Power: How Protest Persuades Political Representatives

Article

Full-text available

Mar 2017

How do public opinion signals affect political representatives’ opinion formation? To date, we have only limited knowledge about this essential representative process. In this article, we theorize and examine the signaling strength of one type of societal signal: protest. We do so by means of an innovative experiment conducted among Belgian national and regional politicians. Elected officials were exposed to manipulated television news items covering a protest demonstration. Following Tilly’s previously untested WUNC claim, four features of the event were manipulated: the demonstrators’ worthiness, unity, numerical strength, and commitment. We argue that these protest features present elected officials with useful cues about what (a segment of) the public wants. We find that these cues affect elected officials’ beliefs. The salience they attach to the protest issue, the position they take, and their intended actions all change as a consequence of exposure. The size of a protest event (numbers) and whether the protesters agree among themselves (unity) are the most persuasive protest factors. The effects of the protest signals come on top of strong receiver effects. We find no evidence that elected officials’ predispositions moderate the effects of the protest features.

Public Opinion as a Movement Outcome: The Case of the U.S. Women's Movement

Article

Full-text available

Sep 2016

We demonstrate that an important outcome of social movements is public opinion change, particularly in the case of the U.S. women's movement. We argue that contentious events associated with the women's movement provide informational cues that prime the public. This process then leads to changes in attitudes regarding gender. We use quarterly time series data on contentious events of the U.S. women's movement ranging from 1960 to 1992 and public opinion about gender attitudes in the United States to examine whether public opinion moves in response to social movement events. Using an error correction model, we demonstrate that social movement events have a significant effect on gender attitudes. Citizens adopt more liberal gender attitudes as the U.S. women's movement increases its activity. These results suggest that social movement scholars should be paying more attention to public opinion when assessing the outcomes of social movements.

THE 2017 WOMEN'S MARCH: A NATIONAL STUDY OF SOLIDARITY EVENTS

Article

Dec 2018

On January 21, 2017, over four hundred cities across the United States organized sister marches in solidarity with the Women's March on Washington. In this paper, we first compare the size of these marches to that of several significant protest-event sources to show how extraordinary turnout was that day. Then, analyzing a nationally representative sample of sister marches, we present univariate statistics for both event-level characteristics (such as demographics of participants or types of speakers) and mobilization processes (such as composition of organizing teams or recruitment efforts). We situate the descriptive findings in the broader literature on protest events and the women's movement to identify how they converge or diverge from established patterns. In addition, our study shows that many event-level characteristics of the sister marches were distinct relative to a recent national study of protests. Also discussed are the ways in which our results contribute to understanding the sister marches' success in mobilizing millions of people to take to the streets.

Why we March: The role of grievances, threats, and movement organizational resources in the 2017 Women's Marches

Article

Dec 2018

We use Metropolitan Statistical Area-level data to investigate the emergence and size of the 2017 Women's Marches. Our findings indicate that the protests can be understood through both grievances and threats felt in the aftermath of Trump's election and movement organizational resources. While the impact of movement resources is as expected (more protest in MSAs with greater resources), the effect of grievances and threats is complex. Cautions concerning the marches in both African American and Hispanic communities result in negative influences on protest in MSAs with larger black and Latino populations. Also, heightened grievances/threats generally do not increase the occurrence of the Women's Marches, but some grievances/threats, specifically among feminists and those voting for Clinton, increase the size of protests. We also do not find significant interactions between grievances and movement resources. Our findings suggest that because post-election grievances/threats were strongly felt, protesters did not need movement organizational leaders to help them define their grievance. They simply needed movement groups to provide venues for protest, that is, a coordinated set of sister marches. We conclude that researchers should consider both the type of grievances and threats and how grievances/threats and organizational resources work alongside one another to promote protest.

From Mobilization to Revolution.

Article

Jan 1980

Preventing and Responding to Dissent: The Observational Challenges of Explaining Strategic Repression

Article

Mar 2016

Although scholarly consensus suggests that dissent causes repression, the behaviors are endogenous: governments and dissidents act in expectation of each other’s behavior. Empirical studies have not accounted well for this endogeneity. We argue that preventive aspects of repression meaningfully affect the relationship between observed dissent and repression. When governments use preventive repression, the best response to dissent that does occur is unclear; observed dissent does not meaningfully predict responsive repression. By contrast, governments that do not engage in ex ante repression will be more likely to do it ex post . We follow U.S. voting scholarship and propose a new instrument to model the endogeneity: rainfall. We couple rainfall data in African provinces and U.S. states with data on dissent and repression and find that dissent fails to have a significant effect on responsive repression in states that engage in preventive repression.

Protest Events: Cause or Consequence of State Action? The U.S. Women's Movement and Federal Congressional Activities, 1956–1979

Article

Sep 1999

Using data compiled from a variety of different sources, we seek to answer questions about the emergence and outcomes of women's collective action in the United States between 1956 and 1979. In particular, we examine hypotheses derived from political opportunity and resource mobilization theories about the emergence of women's protest. We also examine the consequence of women's collective action on congressional hearings and House and Senate roll call votes on women's issues. We find support for arguments about the effects of resources on the emergence of protest. We also find mixed support for arguments about the effects of political opportunity on the emergence of protest. Finally, we find little support for arguments about the effects of women's collective action on congressional hearings and House and Senate roll call votes on women's issues.

The Politics of Nonviolent Action

Article

Mar 1975

Protest Campaigns and Movement Success: Desegregating the U.S. South in the Early 1960s

Article

Mar 2015

Can protest bring about social change? Although scholarship on the consequences of social movements has grown dramatically, our understanding of protest influence is limited; several recent studies have failed to detect any positive effect. We investigate sit-in protest by black college students in the U.S. South in 1960, which targeted segregated lunch counters. An original dataset of 334 cities enables us to assess the effect of protest while considering the factors that generate protest itself—including local movement infrastructure, supportive political environments, and favorable economic conditions. We find that sit-in protest greatly increased the probability of desegregation, as did protest in nearby cities. Over time, desegregation in one city raised the probability of desegregation nearby. In addition, desegregation tended to occur where opposition was weak, political conditions were favorable, and the movement’s constituency had economic leverage.

Did the Women's March Work? Re-Evaluating the Political Efficacy of Protest

Abstract

Recommended publications

The Political Institutions, Processes, and Outcomes Movements Seek to Influence

Polls and Elections Resistance Is Not Futile: Anti‐Trump Protest and Senators’ Opposition to Preside...

News and Geolocated Social Media Accurately Measure Protest Size Variation

Close but Not Too Close: Opposition Network Strategy and Democratization in Zambia