Contingent Democrats vs Genuine Democrats preprint

1

Laboratory for Comparative Social Research

Higher School of Economics

Protests under Non-democratic Regimes:

‘Contingent’ Democrats versus ‘Genuine’ Democrats

Exploring the Cross-national Variation at the Individual Level

Progress Report

February, 2014

Margarita Zavadskaya

Associate Researcher, PhD Researcher at the European University Institute (Florence, Italy)

[email protected]

NB! It is a work-in-progress, comments are more than welcome. Please, do not cite without author's permission.

2

Abstract

Much ink has been spilled over elaborating the explanations why and how people mobilize against

the authoritarian rule. The explanation ranges from the class-based theories with the leading role

of either middle or working class to the rational choice theories that emphasize the role of

“information cascades” and “critical mass” of participants in overcoming the problem of collective

action. However, the ideal participant is viewed as “contingent democrat” or the one whose

actions are primarily driven by self-interest. In the first case people would protest only when

current regime does not meet their economic expectations, in the second case they are afraid of

repression and wait until the costs of participation will decrease. I argue that the theory of

emancipative values may be extended in explaining the protest mobilization patterns even under

non-democratic conditions. In the previous research the macro-effect of emancipative values on

anti-incumbent pre-electoral mobilization has been found. In this paper I show that in most of

authoritarian countries those who share to the greater extent emancipative values tend to

increase the overall or chronic level of approval of peaceful forms of protest, however, the size of

this effect varies across countries as well as enabling mechanisms that translate emancipative

values into action. Meanwhile alternative macro-economic explanations do not show such robust

and significant effect.

3

Introduction

Political mobilization in repressive regimes has always been puzzling for scholars. Indeed, the

more repression we observe the less likely people will take into the streets to spell out their

discontent (Koopmans 1997; Davenport 2005; Opp and Roehl 1990). At the same time when

people have nothing to lose “apart from their chains” they probably would also undertake

desperate moves to express their outrage. On the other hand, non-democratic polities differ in

terms of type and scale of repression, as well as redistributive policies: ‘loyalty buying’ and ‘pork

barrel’ prove to be rather efficient “pacifying strategies” for incumbent elites to avoid mass unrest.

Nevertheless citizens under harsh political conditions do still protest and the repertoire is

constantly changing.

The range of theories of mass mobilization varies from the class-based approaches where the

driving forces of democratization include either representatives' of the middle class or workers'

pro-democratic mobilization (Moore 1967; Rueschemeyer, Stephens, and Stephens 1992) to

rational-choice approach where people should overcome the collective action problem in order to

achieve a public good to be provided (Olson 1971). Repression, in turn, decreases the probability

of any collective action, but information cascades with growing number of participating people

changes the perceived costs of collective action and ‘tip the balance’ in favor of protesters

(Lohmann 1994; Kuran 1995; Marwell and Oliver 1993).

When it comes to empirical study many issues occur. Firstly, the operationalization of social

classes still remains a non-trivial task to do for the students of social protests. However we still

can grasp them through income and education that serve as important background variables.

Then, a more tricky part is the notion of rationality or rational choice when a prospective

participant calculates pros and cons of future actions. The rational choice literature on collective

action depicts the Weberian goal-rational considerations, rather than value-rational.

4

Rational-choice perspective offers the image of a typical protester as “contingent democrat” in Eva

Bellin’s terms (Bellin 2000) whose actions are primarily driven by self-interest (and perhaps

fashion?). Resource theories of participation also emphasize money, time and education as

necessary prerequisites for engaging in any collective action (Brady, Verba, and Schlozman 1995),

as well as literature on political regime change in transition societies stress the importance of elite

splits and defections or intra-elite conflicts by completely disregarding the bottom-up protest as

independent political force (O'Donnell 1994; O'Donnell and Schmitter 1986; Robertson 2010). In

other words, people would protest either when the current regime does not meet their economic

expectations or, while being afraid of repression, they wait until the costs of participation

decrease. Finally, citizens are passive until a rebellious counter-elite needs additional support in

the form of mass revolt.

All this literature fails to challenge the basic assumption that the very fact of being pro-

democratic 1 exclusively depends on economic/material self-interest. In other words, if

dictatorship performs well in terms of people’s ability to consume and per capita income, nobody

would protest. However, this is obviously not the case. Various forms of protests take place in

almost all kinds of political regimes with non-democratic rule starting from monarchies to multi

party constellations (see Figure 1). Moreover protests seem to follow cyclical pattern similar to

those in democracies. There are peaks in 1992 and 1995 followed by the slowdown in 1997. The

next big wave of mobilization takes place from 2002 to 2005. The rates for riots are in general

lower that for peaceful demonstrations, though, coincides with them.

In this paper emancipative values are used as a value-rational base for political mobilization and

thereby proxy for new participants that support democratic rule (broadly understood) as a

principle rather than a set of institutional arrangements that bring about individual economic

prosperity. In this paper I’d like to look at the mobilization patterns from the perspective of

1 Here it means “pro-democratic” in broad sense without differentiating between various normative models of democracy.

I’d define democracy as “rule of the people, by the people and for the people” (Abraham Linkoln).

5

emancipative values. Rational individuals should be sensitive to the level of repression as well as

the individual and/or overall economic situation (grievance or relative deprivation theories).

Emancipative values, in turn, are expected to dampen the effect of repression (‘If you have right

on your side, you have no fear’) (Welzel 2013:215). Thus, economic grievances are secondary

motive vis-a-vis emancipative orientations. Violations of rule of law, disrespectful practices in

terms of peoples’ rights are the main reasons for those with prevailing emancipative values that

help to overcome the problem of collective action.

It is illuminating to think about the circumstances under which group interests overcome this motivational block. For this to happen, the interest’s instrumentality must fade. The interest must eventually become an inherent part of people’s social identity, in which case voicing the interest obtains intrinsic value (Welzel 2013:218).

At the same time there are grounds to argue that these two diverging motivations may not

necessarily be mutually exclusive. Both kinds of considerations – economic discontent and

emancipative values as a part of identity - may coincide. Other things being equal, respondents

with more pronounced emancipative values join peaceful protests more willingly (readily,

eagerly) compared to those who share emancipative values to the lesser extent. No doubt, the

very attempt to dissect these motives seems unrealistic; we can’t literally open respondent’s head

and have a look at his or her preferences or “true” motives. Moreover, often respondents are not

able to detect the genuine motives of their decisions and actions themselves and reconstruct then

post factum in socially desired fashion when asked. Maybe protests at some point obtain

fashionable traits of an emerging leisure class that positively characterize an individual (from

stigmatized action to fashion?)?

On the other hand, the translation of individual or group considerations into collective action

differs from the general approval of a specific type or repertoire of protest actions. In some

societies protest per se might be viewed as something intrinsically bad and, the other way round,

in others – protesters are perceived as public heroes. If the chronic level of protest approval

ceteris paribus is relatively high, the enacting of real action tends to be easier. For this reason a

6

distinction between approval and already taken action is required, as various factors may bring

about shifts in the overall approval of protests and outbursts of protests in specific place and time.

The latter reflects quite a different dimension of social protests. In this paper I draw a line

between generally positive attitudes towards peaceful protest and actual experience of taking to

the streets.

Furthermore, the repertoire of protesters varies across time and countries (Tilly 2009). What is

more acceptable in one country and tolerated by one regime can be absolutely unacceptable in

another regime or protesters. Cross-country variation also needs to be taken into account.

Another critical point is that emancipative values enable exclusively peaceful forms of protests,

rather than violent and aggressive riots or armed insurgencies (Welzel and Deutsch 2012), while

more interest-driven protests may take violent forms as well. At least there is no clear theoretical

expectation regarding this nexus of protest and motivations.

The best ground to test these hypotheses is modern relatively prosperous stable autocracies or

the so-called electoral authoritarian regimes (Schedler 2006) that rarely resort to blatant massive

repression, but systematically abuse the existing democratically designed institutions, rules and

norms (Levitsky and Way 2010). One might admit that mass mobilization can still be observed in

repressive regimes, being supportive, though, of current political elite and its policies, rather than

opposing them. In other words in some types of authoritarian regimes mass mobilization is still

going on but under the pro-incumbent loyal banners. I use Arthur Banks Cross-National Time-

Series Dataset (CNTS) in which the information on contentious events is being collected from the

New York Times. Obviously, this might sound ambiguous if to use only one mass media source

since newspapers also pursue their internal policies and have some specific biases. On the other

hand, theses biases are normally in favor of the large-scale events with clear anti-incumbent

slogans. Thus, these estimations even without covering the whole spectrum of protests and paying

attention only to big events, are still demonstrating significant variation.

7

[Figure 1 about here]

In the previous analysis I focused exclusively on the post-electoral protests where voters take into

the streets being assaulted by blatant electoral fraud, manipulations and hostile behavior of

electoral commissions and police officers. It has been shown that emancipative values increase the

probability of post-electoral protests (Zavadskaya and Welzel 2013). Following the line of the

modernization theory, with the increase in personal economic wellbeing and satisfaction of the

survival needs, people’s demands start being reshaped from purely economic and constrained by

private sphere towards more public issues – the quality of government, political emancipation,

political rights and civil liberties (Inglehart and Welzel 2007; Inglehart 2002). Emancipative

versus survival values also tend to increase the frequency and amplify the scale of peaceful forms

of protests (Welzel and Deutsch 2006). The most striking finding is that emancipative values have

some impact even in non-democratic spectrum of political regimes which are in average poorer

performers in economic terms (Przeworski et al. 2000). In other words, values seem to be one of

the important omitted variables affecting the probability of peaceful pro-democracy mobilization.

However, these findings are relevant for modern authoritarian countries only on aggregate level.

As to the individual level, the effect has been discovered only using the overall sample of countries

regardless of political regime features. Firstly, bearing in mind the potential “ecological fallacy”

trap, we cannot automatically extend the inference from country-level to individual level

mechanisms. Secondly, there is no clear prediction regarding the individual values-profile and

protest activities, since the level of economic development is lower and, thus, average values of

emancipative values are lower in autocracies. Thus, the question is what is the key motivation

for protesters in authoritarian regimes? Are they “true believers” in democratic ideals or

“contingent democrats”, primarily preoccupied by their material interests? Finally, how

and when values translate into political protest under non-democratic conditions at the

individual level?

8

In this paper I try to fill the gap by opening up the black box and to explore the individual-level

correlates of three forms of protests – petitions, demonstrations and illegal occupation of

buildings and factories taken from the 5th wave of World Values Survey (WVS) project. The sample

consists of 21 authoritarian countries for which the survey data are available. The time-span is

from 2000 to 2008, thus, I do not account for the impact of the “great recession” and take into

account the prosperous 2000s.

The paper consists of three parts and proceeds as follows: the first psection is devoted to the core

argument, brief literature overview and hypotheses; in the second – I describe the sample,

operationalization details and method, the third section provides the results from more detailed

“nested” analysis of the probabilities of switching between “would never do” vs “would do” and

“would do” vs “have already done”; I conclude by discussion of the results.

Mobilization under non-democratic conditions: Theory and argument

Any authoritarian regime is by definition a repressive polity. Many scholars assume various

relations between protest and repression. The most established in literature hypothesis suggests

the negative relationship: if repression escalates, the probability of collective action decreases,

thus, repression is a cost or a negative incentive (Opp and Roehl 1990; McAdam, Tarrow, and Tilly

2001). Moreover, some interaction effects can occur between the timing of elections and

repression (Gandhi and Bhasin 2012; Norris 2012). First empirical tests have revealed that the

use of repression is more effective before elections, but its ability to sustain autocratic rule

diminishes immediately after electoral results are announced (Hafner-Burton, Hyde, and Jablonski

2012). Davenport shows that during elections the level of repression tends to be lower and

censorship is not that tight (Davenport 1995). Gandhi and Bhasin, in turn, found the cycles of

state-sponsored repression depending on time left before the month of elections and after (2012).

Repression especially increases before elections that aims at the opposition leaders, and has

proven to be more efficient in suppressing potential uprising. However, in the month of election

9

repression drops to its minimum level, because repression of voters is fairly costly and

undermines the legitimacy.

Another important remark touches upon different types of the regime violence. If Kricheli,

Magaloni and Livne (2011) draw no line between electoral repression, various categories of

repression and regime repressive capacity in general, Hafner-Burton, Hyde and Jablonski in their

analysis clearly differentiate between pre-and post-electoral repressive actions as well as overall

level of violence in a polity (2011). Koopmans, in turn, speaks about two types of repression:

institutional and situational that corresponds to the violent or non-violent mode of mass

mobilization (Koopmans 1997). Following this line of argumentation, institutional repression or

restrictive formal rules – in my view – clearly relate to the so-called structural power in

Lindblom’s terms, whereas situational repression is an instance of instrumental power (Lindblom

1988). Institutional repression is viewed as more legitimate and deeply embedded in society’s and

voters’ perceptions, whereas, situational or ad hoc repression might be seen as a signal of

weakening or lack of legitimacy, thus being more dangerous for the incumbent. In this paper I

address solely situational repression measured as Cingranelli and Richards Physical Integrity

Index which varies from 0 to 8 (2010). There is an alternative measure of repression that is more

finely tuned to the electoral timing, but unfortunately it is not yet available for analysis. I do not

use the most conventional indicator – the relative change in Political Rights Index which is a part

of Freedom House democracy measure

(http://www.freedomhouse.org/template.cfm?page=351&ana_page=364&year=2010), because

this index is based mostly on electoral outcomes and ex post evaluations.

The default assumption is that repression tends to discourage the expression of mass

opposition (Opp & Roehl, 1990; McAdam & Snow, 2010). Still, the link between repression and

opposition is not a simple inverse relation such that opposition decreases in a one-to-one

relationship with increasing repression. Instead, some scholars depict the repression-opposition

10

relationship as an inverted-U curve, with opposition rising from low to medium repression and

then falling from medium to high repression (Davenport, 2005). However, the inverted-U curve

does not always hold, for even massive repression sometimes increases rather than decreases

mass opposition (Francisco 1995; Slater 2010).2 One explanation may be found within the

“bounded rationality” framework when the power of demonstration effect sparks the waves of

protests in such circumstances where any protest activity is doomed to failure. This is not

imprudent protesters’ fault but rather the lack of credible information and psychological effects

that produce disproportionate effects on attention and memory. “People overweigh dramatic

appearance relative to actual importance, deviating from logical criteria” (Weyland 2012, 921).

My explanation is that not only economic wellbeing and rationality (goal-rational actions)

account for the probability of protests, but some people inspired by values are able to openly

oppose the repressive regime and even sacrifice their life for the sake of their ideals. Obviously,

“true believers” do not form the majority group of population, and their protest behavior might be

even stigmatized by their fellow citizens. “Expressive utility” of protest action grows in direct

response to the level of repression because acts that express opposition gain in significance under

more severe repression. Thus, if the logic of expressive utility governs, more people engage in

opposing actions precisely when repression is intensifying. With the country’s economic growth

and relative improvement of individual socio-economic situation, the number of altruistically

oriented citizens grows and protest gradually ceases to be the deviation from “normal”

conventional political behavior. Probably, it even may become fashionable. With more people

voicing opposition, the solidarity experience grows, which amplifies the expressive utility of

showing support of opposition. This finally can result in a “surprise” for authoritarian incumbent

(Kuran 1995). I suppose that this dynamic becomes more likely as emancipative values are more

widespread because then people’s joint utility from expressing opposition grows. Peaceful forms

2 See Davenport (2005) and Opp (2009) for comprehensive reviews of the literature.

11

of protests gradually conventionalize or normalize among the citizens and then pull the rest of

population after themselves (Tarrow 1995).

One qualification must be added to this claim. Emancipative values only encourage

nonviolent opposition, not violent opposition. As Deutsch and Welzel (2011) argue, emancipative

values involve an inherently humanitarian orientation that considers human lives sacrosanct and

rejects violence against them. Thus, I do not expect emancipative values to be associated with

violent forms of discontent.

In this paper I seek to contribute by, firstly, testing the importance of emancipative values

on the individual level and respondent’s propensity to engage in any kind of protest event.

Secondly, I wish to test whether emancipative values are of any significance under various non-

democratic settings.

To sum up, my theoretical expectations boil down to the following hypotheses:

H0: emancipative values are significantly associated with participation in lawful/peaceful

demonstration;

H1: higher scores on emancipative values are associated with higher propensity of

participation in lawful/peaceful demonstrations;

H2: emancipative values are not expected to have any link with violent forms of protest.

Alternative hypotheses:

H3: more positive assessments of financial situation are less associated with both higher

propensity and actual participation in lawful/peaceful demonstrations;

H4: the growth of inflation rate is positively associated with protests (violent and non-violent)

Protest as the dependent variable

12

Main dependent variable is respondent’s approval of three forms of protest behavior:

signing petitions, participation in peaceful demonstrations and illegal occupation of buildings or

factories. The main form of protest for us is demonstration; however, in order to account for the

variegated repertoire of available actions I also include petitions as less contentious and costly

form as well as illegal occupation of building and factories as more nonconventional and violent

form of protest. The codebook is available in the Appendix. Unfortunately, I cannot take into

account by using these data whether a declared political action was opposing the government and

regime in general or was it pro-regime mobilization.

Contingent vs Genuine democrats: Emancipative Values and Self-interest as the main

independent variables

A “genuine” claim for democracy and more open politics is approximated through

‘protective-vs.-emancipative values,’ or in short: emancipative values. This set of values is based

on twelve items which are summarized in a two-step procedure. In the first step, groups of three

items are summarized into four sub-indices. In the second step, the four sub-indices are

summarized into the two encompassing values indices. This is done by averaging item scores,

after having recoded all items into the same polarity and having standardized their coding

schemes into the same scale range between minimum 0 and maximum 1.

I adapted the aggregated level indicated to the individual level by using the same items and

weights. As a result each respondent has his/her own score on protective – emancipative values.

Those sharing emancipative values to the greater extent are considered as “genuine democrats” as

opposed to those sharing more protective or survival values – “contingent democrats”.

The counterargument is that protesters can be those who are driven primarily by their

economic self-interest. I measure ‘contingent’ democrats through indicators that reflect their

individual financial wellbeing and ability to consume, as well as the general economic situation. I

13

use the WVS question (c006) that reflects the subjective satisfaction with financial situation of

respondent’s household. The response items vary from 1 to 10 where 1 stands for “Unsatisfied at

all”. At the aggregate level the changes in inflation rates and

The last variable that characterizes protesters is repression: “contingent democrats” are

fearful of repression while emancipative values tend to be less scared by possible reaction from

the regime.

Alternative explanations or different mechanisms that facilitate mobilization?

Different values may or may not translate into protest action. The theory suggests that

values indicate just a chronic level of approval of protest action rather that mechanism that

transform the latter into the reality. Therefore, I include the variables that depict various

mechanisms or paths that facilitate mobilization. Social movement literature is abundant with

various approaches; however, I will use the most important of them: grievance theory (Gurr), SES

models (Verba, Brady and Schlozman 1995), social capital approach (Putman), political

opportunity structure or necessary infrastructure for protest. These variables can act both at the

individual and aggregate levels, as control variables or variables that mediate the effects of values

and form different pathways from the predictor to an outcome.

Probability to take part in political action may depend on many individual features from

age and gender to educational level, occupational status and economic evaluations. Apart from

individual peculiarities one should account for intra-regime variation and country-specific traits

that might alter the probabilities of protest actions. In order to handle the possible omitted

variable biases I include control variables at two levels: individual and aggregate indicators by

country.

Individual features

14

Younger people are more prone to take part in protest actions when women in average are less

exposed to be mobilized. More educated people tend to take part in lawful demonstrations

compared with their less educated counterparts (Brady, Verba, and Schlozman 1995). Finally,

those who read newspapers or watch analytical programs more often are expected to be more

exposed to political involvement. I do not have clear expectation regarding active internet-users,

since some literature suggests the important role of Internet as an alternative source of

information, efficient tool for protest mobilization and maintaining protest infrastructure (Ayres

1999; Earl 2010; Van Laer 2007). On the other hand, Internet might be controlled by state

authorities (e.g. the infamous Chinese firewalls) or even in the absence of such control Internet

does not seem to possess any ‘democratizing’ features per se.

On the other hand, following the line of social capital theory, the membership in civic

organizations such as human rights, ecological societies or other peace movements weakens the

motivational block in order to become a part of collective action. Experience of voluntary activities

builds trust networks that can transcend into political action (Putnam).

Satisfaction with financial situation at the individual level, in turn, captures economically

motivated approval and participation as opposed to the actions inspired by the growth in

emancipative values. Together with reactions to the shifts in inflation rate and adverse economic

situation at the country level reflect the self-interest dimension of a collective action.

Country-level explanations

Macro-economic conditions also affect the protest dynamics. If we assume that citizens

protest only when they suffer economic grievances, so we expect inflation and unemployment

rates to have an impact on the propensity of mass protests, especially given that these issues are

usually at stake during the elections (Przeworski et al. 2000; Boix 2003; Acemoglu and Robinson

15

2005; Pepinsky 2009). Voters tend to punish their governments through elections if the latter

perform poorly in socio-economic policy during their term and vice versa (Alesina, Roubini, and

Cohen 1997). When the rules of electoral game are systematically abused by the incumbent or the

ruling party, the punishment by electoral means becomes an option to voters and if their

dissatisfaction is ignored by institutional channels it is likely to be streamed into protest activities.

Thus, poor economic performance is expected to increase the probability of post-electoral

protests. I use GDP per capita calculated constant US$ prices of 2005 retrieved from the Penn

World Table (Heston, Summers, and Aten 2012) which has the widest country coverage. I also

include the aggregate number of Internet-users per 100 people to account for the overall spread

of Internet-technologies in the country.

Descriptive statistics

In order to test aforementioned hypotheses I choose the countries for which available survey data

exist. The sample includes 21 countries from the 5th wave of the World Values Survey project. All

these countries score poorly on most conventional democracy indices. I set the threshold of 3 and

higher for Freedom House or 5 and lower for Polity IV (Marshall, Jaggers, and Gurr 2004) for the

year before the survey takes place. Thus, the dataset contains different types of authoritarian

countries from poor and neo-patrimonial Zimbabwe and Zambia, countries of Maghreb to highly

modernized regimes in Singapore and Malaysia. Unfortunately, the questions on protest

participation have not been asked in some of the countries covered by WVS – China 2001, Iran

2000, 2007, Iraq 2004 and Saudi Arabia 2003. The data on illegal occupation of buildings and

factories also exists only for nine from twenty-one country surveys. Thus, I will focus mainly on

demonstrations and petitions; however, I provide the analysis for illegal actions as well. As it is

16

shown on the bar charts by country, the patterns of protest participation differ across countries,

but signing petitions and participating in demonstrations remain the dominant forms of protest3.

[Figure 2 about here]

Emancipative values are also distributed unequally across country-years. The highest mean score

is observed in Ethiopia, when the lowest is in Iraq (see Figure 3). On the other hand for the most

of cases emancipative values are not distributed normally, in some countries there many ‘outliers’

or respondents with relatively high scores (e.g. China, Singapore, Malaysia, Iran). In others – the

distribution of values seem polarized, as it the case of Ethiopia, where the outlying dots are

concentrated in the left- and right-hand parts of the box plot. The latter is possibly pulling the

country mean higher. I would even speculate that these dots with significantly higher scores than

the country country’s average are those subgroups of population with more tolerant, pro-

democratic orientations and caring less about consumption. Russia, Rwanda and Morocco, for

instance, look less polarized, and most of the respondents are grouped around the country mean

scores.

[Figure 3 about here]

Finally, a brief look at the joint distributions of aggregate rates of those who are more prone to

take part in lawful demonstrations with countries’ average emancipative values scores and GDP

per capita. As it stems from Inglehart and Welzel’s modernization theory, overall economic

development slowly but surely leads to the formation of more civic-oriented values among

3 As to signing petitions, one might object that it can hardly be viewed as protest action at all, rather a conventional act

which is unlikely to threaten political regime. Probably, it holds for established democracies, however for some democratic countries it could be the only possibility to express their concern with public issues which is not that costly as opposed to demonstrations and more radical actions.

17

citizens. On the scatter plots (figures 4A and 4B) with the subsample of authoritarian countries

the demonstration frequency is positively correlated with aggregate country score on

emancipative values which is predictable. However, the GDP (logged), vice versa, reveals the

negative association with demonstrations. If to put it differently, in wealthier autocracies people

protest less, but it does not imply the same relationship with emancipative values. On the scatter

plot 4C the non-linear relationship between values and GDP per capita is shown. Singapore, Saudi

Arabia and Malaysia seem to be the outliers where favorable economic situation dampens the

probability of civic unrest against arbitrary political regime4. This tells that the economic

development is important variable that may affect the probability of values to translate into

political action.

[Figure 4A,B,C about here]

Variables and operationalization

As it is mentioned above, I use three dependent variables: to what extent do respondents approve

the participation in peaceful/lawful demonstrations, signing petitions or illegally occupying

buildings and factories. These entire variables are coded as three-item Likert-scale where the

responses might be either ‘already done’, ‘Would do’ or ‘Would never do’. I inverted the scale in

such a way that lower values stand for rejection of any possibility of protest action and higher

values mean approval or actual participation in such actions.

Emancipative values are measured at the individual level using the procedure suggested by

Welzel (2013). The details of the index construction are described earlier. I use the standardized

4 This resonates with recent findings from the survey experiments carried out by Klasnja and Tucker (2012) who argue that

in high corruption countries the periods with growing economy make voters insensitive toward the instances of blatant corruption. Thus, prospering economy creates some sort of ‘vicious circle’ where people tend to disregard politicians’ ‘misconduct’ and not to punish them.

18

emancipative values scores, i.e. they vary from 0 to 100. The same applies to financial satisfaction

that has been also normalized.

The coding procedure, nature of the variables (continuous, binomial or ordinal) and the sources

can be found in the Appendix.

Estimation procedure

If the dependent variables are ordinal-scale ones, thus, linear models are not applicable.

Therefore, I use logit and ordinal logit models to test the hypotheses where I estimate the

probability of making transition from one response category to another (Scott and Freese 2006).

In order to account for country-specific variables that potentially affect the probability of the

outcomes, I use clustered standard errors as well as country fixed effects. Thus, in this paper I

estimate the pooled ordinal logit models with clustered standard errors and fixed effects and I

also run the same models for each country-year separately.

Analysis

Evidence from the “nested” analysis

It is well known that being supportive of protest actions does not necessarily lead to actual

participation. Emancipative values do not immediately translate into political action; rather the

whole process from citizens’ attitudes to the very decision to engage in protest is mediated by

other variables. In this part of the paper I present the results of the nested analysis of three types

of protest activities: signing petitions, participation in lawful demonstrations and illegal

occupation of buildings and factories. Petitions seem to be less risky and more conventional action

relative to the other forms of protest. Therefore I expect to observe smaller effect of emancipative

values compared with participation in demonstration and no effect on illegal actions.

19

Consequently, the discrepancy between being supportive and real participation should be more

striking as one move from more conventional actions to more violent forms. At the same time, the

effect of emancipative values is expected to be strong in the case of demonstration, while illegal

actions should not be associated with emancipative values.

The preliminary estimates are presented in Tables 1, 2 and 3 of the “nested” analysis of propensity

to protest. Emancipative values are consistently and positively associated with the probability to

approve participation in peaceful demonstrations, one unit change in the emancipative values

profile induces a .01 or .02 increase in the log-odds of switching to the “pro-demonstration” camp.

The probability of making transition from the support to the real action, though, seems lower and

even insignificant. Quite a similar pattern is observed in case of petitions: respondents with higher

scores on emancipative values tend to be more willing to sign petitions. The effect of emancipative

values on the odds of transition from mere approval to signing petitions proved to be larger than

for demonstrations. Probably, under repressive conditions signing petitions are perceived as less

dangerous.

[Tables 1,2,3 about here]

The satisfaction with an individual financial situation shows almost zero effect on the propensity

to be supportive of demonstrations and taking part in demonstration. However, the direction of

relationship is opposite: negative – for making transition from non-approval to approval and

positive – from approval to action. The explanation can be that the less successful are unhappy

with their situation, though, are not ready to undertake any action to voice their grievances. Those

who do demonstrate are subjectively better off in financial terms. The effect of satisfaction with

financial situation is not significant for signing petitions. The same applies to illegal actions.

Another important predictor is state-level repression that should scare ‘contingent democrats’

away and be insignificant for ‘genuine democrats’. For demonstration we observe the striking

20

results: repression does frighten people when they state the attitude towards protest action, while

they have an opposite effect when it comes to real actions. This finding confirms the theoretical

expectation that repression does not only prevent people with higher emancipative values to

demonstrate, but may even trigger protest action. The same link is found for petitions; however,

the b-coefficients are not significant for the models that fit that probability of transition to action.

The effect of repression on illegal action is constantly negative for both sets of models.

In other words, adepts of peaceful forms of protest are not scared away by political repression.

This fact, in its turn, speaks to the initial research whether protesters are “contingent” democrats

or they are, indeed, ready to run the risk of being detained or even beaten or tortured. The results

of the ‘nested analysis’ demonstrate more detailed picture of how repression, values and protests

are interrelated. Thus, the differentiation between the levels of general approval and protest

action in specific time and space is required when one intends to look at the impact of repression

on protest activities.

The last variable that also captures the consuming capacity of citizens is the annual change of

inflation rate. As to demonstrations, the effect is only significant and positive when a propensity

transforms into decision to rise up. This finding confirms the alternative view that citizens react

not to their dismal economic situation, but relative decrease in their consuming capacity. At the

same time it does not tell us that protesters are necessarily deprived of any sort of luxury

consumption. It is likely that protesters may react to the rise in prices on basic food supplies or

growth of the utility bills’ costs. The similar pattern is observed for petitions. The effect on illegal

actions holds only for the approval and fades away for action.

Referring to the control variables, there is evidence that women are constantly less prone to

engage in any kind of protest activities, however, this difference becomes twice as smaller when it

comes to switching from aptitude to take part in demonstration to real participation. Almost the

same applies to petition signing. But the effect of gender withers away in the case of unsanctioned

21

collective actions. Membership in civic organization enables both more a positive attitude towards

demonstrations and decision to take into the streets. The same effect is observe for petitioning

and no effect on illegal actions.

Education is positively associated with the probability of switching from less supportive to more

protest-oriented behavior. Although, among the respondents who reveal their potential consent to

participate only those with higher education are more likely to protest, compared to those with

secondary and primary education. Interesting fact is that the probability making transition to

support of illegal actions is strongly associated with higher education; however, if one takes only

the subsample of potential illegal protesters, more educated fellow citizens are significantly less

likely to engage with it.

Income has stronger effect in the subsample of potential demonstrators, i.e. wealthier people

protest more and are less fearful of potential outcomes. On the other hand, there is no robust

effect of income found for petitioners and nonconventional protesters.

Overall economic development is significant and positive in the case of demonstrations and

petitions, while for illegal actions its effect proved to be not that robust. The effect of the spread of

Internet seems less salient. For the most of the models its effects is inconsistent, however, it

revealed some unexpectedly negative effect in the case of making transition from approval of

demonstrating and petitioning to action.

The pathway from emancipative values to protest can be also facilitated by the access to various

sources of information which triggers the probability to switch from loyalists to more critical

standpoints, from passive and often salient supporters to protesters. Surprisingly, there is no

effect of the Internet access on the probability to switch from non-approval to more supportive

attitudes towards demonstrations. This finding is to be investigated further since the causal

pathway is still unclear and needs more robustness tests.

22

Another important point is that there might be important country-specific patterns and the effect

of emancipative values as well as other controls may differ. In order to be more specific the same

‘nested’ analysis procedure has been applied to each country in the sample. Looking at the

country-level is even more insightful since the various configurations of emancipative and more

‘selfish’ values become visible. Firstly, there are dramatic discrepancies in the effect of

emancipative values and satisfaction with financial situation between the countries. Secondly, the

effects diverge between the first analysis where the probabilities of being supportive of

demonstrations are fitted and the second step where the subsamples of pro-demonstration

respondents are explored.

[Tables 4, 5 about here]

Let us start from the analysis of the full country samples where the outcome variable indicates

whether a respondent would or would never take a part in lawful/peaceful demonstration. The

effect of values is statistically significant only in 50% of all countries where the survey data were

available. The effect, though, is always positive. As opposed to the values, the effect of satisfaction

with financial situation is either insignificant or negative. An interesting fact is that emancipative

values and financial satisfaction act in opposite directions with a couple of exceptions (Morocco

2007). Therefore, it resonates with the hypothesis that these motivations are more substitutive

than complementary or mutually reinforcing. However, this pattern is not that clear-cut as there

are cases where both predictors turn out insignificant (Iraq, Zimbabwe, Egypt 2000, Tanzania,

Zambia).

As to the second step of the ‘nested analysis’ or the analysis of the subsamples where the

probability of real action is being fitted, the significance of the b-coefficients of emancipative

values almost mirror the results of the first step, i.e. most of the countries have switched their

places. The effect of emancipative values becomes significant for Iraq, Nigeria, Rwanda, Tanzania

and Zambia where previously this effect was insignificant. And, vice versa, in Jordan, Kyrgyzstan,

23

Malaysia, Singapore and Egypt the effect fades away. This suggests that the transition from non-

approval to approval and from approval to action is not the same. Then, there two ways how

values affect the probability of peaceful demonstrations: either they help increasing the chronic

level of positive social attitude towards non-violent forms of protest or facilitate the translation of

approval into real demonstration. Only in Algeria, Morocco and Burkina Faso the effect remain

significant and positive for both stages. Meanwhile, Albania, Ethiopia, Russia and Zimbabwe seem

immune to emancipative values.

At the same time, the impact of financial satisfaction proved to be less important and mostly

negative or insignificant. Only in Morocco, Kyrgyzstan and Rwanda this effect is positive and

significant. I other words, economic motivation is less robust and is highly contingent upon other

circumstances, when the effect of values is mostly unconditional and robust. In most of the

countries only ‘genuine democrats’ do protest, while ‘contingent democrats’ show up only in some

cases. Just in couple in a few cases these variables act in the same direction: Morocco and

Rwanda.

24

Discussion

Protests under repressive regimes do significantly differ from those under more liberal

conditions. The classic collective action predicts that no one would protest because of high costs of

participation. Nevertheless, we regularly observe mass outbursts of people’s discontent all over

the world. The most recent waves – the Color revolutions and the Arab spring – bring the evidence

that the repression-protest nexus is more complicated that it looks at the first glance. Another

puzzle is what motivates people to take into the streets: extreme economic grievances and relative

deprivation or sincere claims for genuine democratic change driven by the growth in the so-called

emancipative values? Who are those who push forward pro-democratic slogans: “contingent

democrats” in Eva Bellin’s terms or “true believers” (“message-seekers”)?

Cross-sectional comparison with country-level unit of analysis often suppresses significant intra-

case variation. Theoretically authoritarian countries are in average poorer, so that they often

cannot satisfy economic demands of population. However, some regimes have managed to

maintain relatively high per capita income and even improve their citizens’ well-being. As it is the

case of not only Singapore, but also Malaysia and Russia. At hard times probably both -

“consumerists” and “true democrats” - may protest, however, in time of economic stability and

even growth, only those with predominantly emancipative values are prone take into the streets.

I have taken the period from 2000 to 2008, the time of relative economic stability and for some

countries – the time of unprecedented economic growth. The evidence from individual-level

analysis suggests five important findings:

1) Emancipative values have revealed in average positive effect on the probability of peaceful

forms of political protest even in authoritarian countries. At the same time emancipative

values have nothing to do with intentionally violent and illegal protest actions; the average

effect seems robust and unconditional, though, not big in magnitude;

25

2) Individual satisfaction with financial situation coupled with inflation rates proved to be

weaker predictors of peaceful protests. In other words, the effect of financial satisfaction

seems contingent upon many other country specific predictors;

3) The evidence from the country-level analysis confirms the hypothesis that emancipative

values and subjective satisfaction with economic conditions create rather controversial

stimuli for the respondents, i.e. the configuration of both is mostly substitutive rather than

complimentary or mutually reinforcing (with a few exceptions);

4) The evidence from the “nested” analysis reveals that there are differences between various

degrees of approval of protest action by respondents. The transition from non-approval to

approval of protest action does not reflect exactly the same phenomenon as transition from

passive support to real action.

5) There is significant intra-country variation in ways how values may translate into political

action. There are different modes of how emancipative values affect the protest action: in

some cases values are crucial at the stage of being more positive towards the very fact of

lawful demonstrations, in others – at the stage of taking a decision to demonstrate. In the

first case they reflect the general degree of societal attitudes towards peaceful protest, in

the other – the mechanism of mobilization.

Relatively low coefficients of determination indicates that the puzzle of mass mobilization

under non-democratic conditions still remained to significant extent unexplained not only by

emancipative values, but by conventional theories as well. Repression, inflation rates and the

spread of the Internet do not shed much light on the individual level patterns of mobilization

as well. At the same time, Internet communication is not a ‘democratizing’ or ‘mobilizing’ tool

per se, rather a channel facilitating mechanism. However, the pace of Internet spread seems to

matter: at the earlier stages of proliferation Internet may be a powerful instrument of

mobilization, however, at later stages the effect seems even negative (see in Appendix).

26

Authoritarian countries differ to a much greater extent than democracies. Citizens living under

non-democratic rule vary according their not only their socio-economic positions, but their

personal value-profiles as well. Some societies are more homogenous, others, vice versa,

extremely polarized and divided. In some countries protesting is viewed as heroic deed, while

in others any action is severely stigmatized and even anti-socially punished by their fellow

citizens. As Dmitriev and Treisman describe the situation of Russia:

‘the urban activists who have appeared on the front pages of newspapers around the world constitute at most a tiny fraction of the Russian population -- a few hundred thousand people in a country of 143 million. The big question that will determine Russia's political future is how much support this politicized vanguard can hope for from the quiet majority that lives outside Moscow and St. Petersburg. (…) The stereotype of the provincial Russian is of a politically apathetic conformist who is resentful of pampered Muscovites, socially conservative, generally pro-Putin, suspicious of the West, and nostalgic for Soviet order’(Dmitriev and Treisman 2012).

Perhaps, other instances of an ambiguous rule share similar patterns: citizen vanguard with

predominant emancipative values bears the costs of pro-democratic pioneers, when the rest join

at the later stages when the costs are not so high. However, another story also may be true:

bottom-up citizens inspired by the events in other countries apply the same frame to their own

situation. The lack of sufficient expertise makes them less fearful of potential repression from the

incumbent’s side. Anyway, these are the questions for future research.

As to the project, the next steps will be the introduction of additional country-level predictors:

type of institutional design, the longevity of authoritarian rule. At the individual level I’m planning

to test the impact of mass media more systematically, by using the question on how often and

from where people get the information. Finally, in terms of method, there is a need for multilevel

models of protest actions and more scrutiny on the interaction between the individual-level

emancipative values and country-specific features.

27

References

Acemoglu, D., and J.A. Robinson. 2005. Economic origins of dictatorship and democracy: Cambridge

University Press.

Alesina, A., N. Roubini, and G.D. Cohen. 1997. Political cycles and the macroeconomy: MIT press.

Ayres, Jeffrey M. 1999. "From the streets to the Internet: The cyber-diffusion of contention." The

ANNALS of the American Academy of Political and Social Science no. 566 (1):132-143.

Bellin, Eva. 2000. "Contingent democrats: Industrialists, labor, and democratization in late-

developing countries." World Politics no. 52 (02):175-205.

Boix, C. 2003. Democracy and redistribution: Cambridge University Press.

Brady, Henry E, Sidney Verba, and Kay Lehman Schlozman. 1995. "Beyond SES: A resource model

of political participation." American Political Science Review:271-294.

Cingranelli, D.L., and D.L. Richards. 2010. "The Cingranelli and Richards (CIRI) Human Rights Data

Project." Human Rights Quarterly no. 32 (2):401-424.

Davenport, C. 1995. "Multi-dimensional threat perception and state repression: An inquiry into

why states apply negative sanctions." American Journal of Political Science:683-713.

———. 2005. Repression and mobilization. Vol. 21: University of Minnesota Press.

Dmitriev, M., and D. Treisman. 2012. "The Other Russia. Discontent Grows in the Hinterlands."

Foreign Affairs (September/October).

Earl, Jennifer. 2010. "The dynamics of protest-related diffusion on the web." Information,

Communication & Society no. 13 (2):209-225.

Francisco, R.A. 1995. "The Relationship between Coercion and Protest An Empirical Evaluation in

Three Coercive States." Journal of Conflict Resolution no. 39 (2):263-282.

Gandhi, J., and Bhasin. 2012. Government Accommodation and Repression in Non-Democratic

Elections. In Pre-IPSA Workshop, Challenges of Electoral Integrity. Madrid.

28

Hafner-Burton, E., S. Hyde, and R. Jablonski. 2012. "When Governments use Election Violence to

Stay in Power." British Journal of Political Science, Forthcoming.

Heston, Alan , Robert Summers, and Bettina Aten. 2012. Penn World Table Version Center for

International Comparisons of Production, Income and Prices at the University of

Pennsylvania.

Inglehart, R. 2002. "Trends in Political Action: The Developmental Trend and the Post-Honeymoon

Decline." International Journal of Comparative Sociology no. 43 (3-5):300-316. doi:

10.1177/002071520204300305.

Inglehart, R., and C. Welzel. 2007. "Mass Beliefs and Democratic Institutions. En." The Oxford

Handbook of Comparative Politics.

Koopmans, R. 1997. "Dynamics of repression and mobilization: The German extreme right in the

1990s." Mobilization: An International Quarterly no. 2 (2):149-164.

Kricheli, R., Y. Livne, and B. Magaloni. 2011. Taking to the Streets: Theory and Evidence on

Protests under Authoritarianism. Paper read at APSA 2010 Annual Meeting Paper.

Kuran, Timur. 1995. Private truths, public lies : the social consequences of preference falsification.

Cambridge, Mass.: Harvard University Press.

Lindblom, C.E. 1988. Democracy and market system: Norwegian University Press Oslo.

Lohmann, Susanne. 1994. "The dynamics of informational cascades." World Politics no. 47 (1):42-

101.

Marshall, M.G., K. Jaggers, and T.R. Gurr. 2004. Polity IV project: Center for International

Development and Conflict Management at the University of Maryland College Park.

Marwell, Gerald, and Pamela Oliver. 1993. The critical mass in collective action: Cambridge

University Press.

McAdam, Doug, Sidney Tarrow, and Charles Tilly. 2001. Dynamics of Contention. Cambridge:

Cambridge University Press.

29

Moore, Barrington. 1967. Social origins of dictatorship and democracy: lord and peasant in the

making of the modern world. London,: Penguin P.

Norris, P. 2012. Why Electoral Malpractices Heighten Risks of Electoral Violence. Paper read at

APSA 2012 Annual Meeting Paper.

O'Donell, G.A. 1994. "Delegative democracy." Journal of Democracy no. 5 (1):55-69.

O'donnell, Guillermo A, and Philippe C Schmitter. 1986. "Transitions from authoritarian rule:

Tentative conclusions about uncertain democracies."

Olson, Mancur. 1971. The logic of collective action; public goods and the theory of groups, Harvard

economic studies,. Cambridge, Mass.,: Harvard University Press.

Opp, K.D., and W. Roehl. 1990. "Repression, Micromobilization, and Political Protest." Social Forces

no. 69 (2):521-547.

Pepinsky, T.B. 2009. Economic crises and the breakdown of authoritarian regimes: Indonesia and

Malaysia in comparative perspective: Cambridge University Press.

Przeworski, A., M.E. Alvarez, J.A. Cheibub, and F. Limongi. 2000. Democracy and development:

political institutions and well-being in the world, 1950-1990. Vol. 3: Cambridge University

Press.

Robertson, Graeme B. 2010. The politics of protest in hybrid regimes: managing dissent in post-

communist Russia: Cambridge University Press.

Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John D. Stephens. 1992. Capitalist

Development and Democracy: Polity Press.

Scott, Long J, and Jeremy Freese. 2006. "Regression models for categorical dependent variables

using Stata." College Station, Tex.: StataCorp LP.

Slater, Dan. 2010. Ordering Power Contentious Politics and Authoritarian Leviathans in Southeast

Asia. Cambridge: Cambridge University Press.

Tarrow, Sidney. 1995. "The Europeanisation of conflict: Reflections from a social movement

perspective." West European Politics no. 18 (2):223-251.

30

Van Laer, Jeroen. 2007. Internet Use and Protest Participation: How do ICTs affect mobilization?:

Universiteit Antwerpen, Faculteit Politieke en Sociale Wetenschappen.

Welzel, C., and F. Deutsch. 2012. "Emancipative Values and Non-Violent Protest: The Importance

of ‘Ecological’Effects." British Journal of Political Science. Vol.42(2):465-479.

Welzel, C. 2013. Freedom Rising. Cambridge: Cambridge Unviersity Press

Weyland, Kurt. 2012. "The Arab Spring: Why the Surprising Similarities with the Revolutionary

Wave of 1848?" Perspectives on Politics no. 10 (04):917-934. doi:

doi:10.1017/S1537592712002873.

Zavadskaya, M. and C. Welzel (2013). “Values,RepressionandSubversion.” World Values Research

6: 46–75.

31

Figure1. Frequency of protests in authoritarian regimes by year

Source: CNTS

NB! Only authoritarian countries included, i.e. Freedom House (Political rights) t-1 >=3 OR Polity IV < 6.

0

2

4

6

8

10

12

number of demonstrations

number of riots

32

Figure 2. Protest events by country

Source: WVS

0.5

11

.52

0.5

11

.52

0.5

11

.52

0.5

11

.52

0.5

11

.52

albania (2002) algeria (2002) china (2001) china (2007) ethiopia (2007) iran (2000)

iran (2007) iraq (2004) iraq (2006) jordan (2001) jordan (2007) kyrgyzstan (2003)

malaysia (2006) morocco (2001) morocco (2007) nigeria (2000) russian federation (2006) rwanda (2007)

saudi arabia (2003) singapore (2002) zimbabwe (2001) egypt (2000) egypt (2008) tanzania (2001)

burkina faso (2007) zambia (2007)

mean of demonstration mean of petitiondone

mean of occupy

Graphs by country - year

33

Figure 3. Distribution of emancipative values by countries

Source: WVS

0 20 40 60 80 100EVstan

zambia (2007)burkina faso (2007)

tanzania (2001)egypt (2008)egypt (2000)

zimbabwe (2001)singapore (2002)

saudi arabia (2003)rwanda (2007)

russian federation (2006)nigeria (2000)

morocco (2007)morocco (2001)malaysia (2006)

kyrgyzstan (2003)jordan (2007)jordan (2001)

iraq (2006)iraq (2004)iran (2007)iran (2000)

ethiopia (2007)china (2007)china (2001)

algeria (2002)albania (2002)

34

Figure 4A. Average number of potential protesters by emancipative values

Figure 4B. Average number of potential protesters by GDP ppp pc

albania (200

algeria (200

ethiopia (20

iraq (2006)

jordan (2001jordan (2007

kyrgyzstan (

malaysia (20

morocco (200

morocco (200nigeria (200

russian fede

rwanda (2007

singapore (2

zimbabwe (20

egypt (2008)

tanzania (20

burkina faso

zambia (2007

11.

52

20 25 30 35 40 45evstan

demonstration Fitted values

albania (200

algeria (200

ethiopia (20

iraq (2006)

jordan (2001jordan (2007

kyrgyzstan (

malaysia (20

morocco (200

morocco (200nigeria (200

russian fede

rwanda (2007

singapore (2

zimbabwe (20

egypt (2008)

tanzania (20

burkina faso

zambia (2007

11

.52

6 7 8 9 10 11ln(gdpt1-76.35459)

demonstration Fitted values

35

Figure 4C. Logged GDP ppp pc by emancipative values

albania (200

algeria (200

china (2001)

china (2007)

ethiopia (20

iran (2000)

iran (2007)

iraq (2004)iraq (2006)

jordan (2001

jordan (2007

kyrgyzstan (

malaysia (20

morocco (200

morocco (200

nigeria (200

russian fede

rwanda (2007

saudi arabia

singapore (2

zimbabwe (20

egypt (2000)

egypt (2008)

tanzania (20

burkina faso

zambia (2007

20

25

30

35

40

45

6 7 8 9 10 11ln(gdpt1-76.35459)

evstan Fitted values

36

Table 1. Nested models . Demonstrations”5

'Would never do' vs 'Would do' 'Would do' vs 'Have already done'

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

EVstan 0.01***

0.01***

0.01*** 0.01***

0.00

0.00

(0.00)

(0.00)

(0.00) (0.00)

(0.00)

(0.00)

Finsat

-0.00

-0.00** -0.00**

0.00***

0.00*** 0.00***

(0.00)

(0.00) (0.00)

(0.00)

(0.00) (0.00)

Represinverted

0.03* 0.05** 0.06**

-0.06** -0.08** -0.11***

(0.02) (0.02) (0.03)

(0.03) (0.03) (0.04)

2.x001sex

-0.42*** -0.37*** -0.41***

-0.22*** -0.24*** -0.24***

(0.04) (0.04) (0.04)

(0.07) (0.07) (0.07)

x003age

-0.01*** -0.01*** -0.01***

0.02*** 0.02*** 0.02***

(0.00) (0.00) (0.00)

(0.00) (0.00) (0.00)

2.x025redulevel

0.28*** 0.30*** 0.28***

0.06 0.05 0.05

(0.05) (0.05) (0.05)

(0.09) (0.09) (0.09)

3.x025redulevel

0.72*** 0.77*** 0.73***

0.24** 0.25** 0.25**

(0.07) (0.07) (0.07)

(0.10) (0.10) (0.10)

2.x047r_incomelevel

0.12** 0.15*** 0.14***

0.26*** 0.24*** 0.24***

(0.05) (0.05) (0.05)

(0.08) (0.09) (0.09)

3.x047r_incomelevel

0.09 0.15** 0.12**

0.36*** 0.29*** 0.29***

(0.06) (0.06) (0.06)

(0.09) (0.10) (0.10)

1.membership

0.53*** 0.55*** 0.53***

0.36*** 0.35*** 0.35***

(0.07) (0.06) (0.07)

(0.11) (0.11) (0.11)

Loginfl

-0.02 -0.02

0.33*** 0.29**

(0.09) (0.09)

(0.11) (0.12)

loggdpt1

0.06 0.06

0.28* 0.31**

(0.12) (0.12)

(0.15) (0.15)

Logint

0.01

-0.11*

(0.04)

(0.07)

N 30750 30541 11254 11218 11218 12217 12116 4605 4578 4578

5 Note that not all specifications are presented in this paper, but they are available upon request. Constant terms and country dummies are suppressed. Models with country fixed

effects and clustered standard errors by country produce almost identical estimates for all variables.

37

PsR2

Table entries are log-odds, i.e. the natural logarithm of odds, i.e. Log-odds = ln(p/(p-1)). Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

38

Table 2. Nested models . Petitions.

'Would never do' vs 'Would do' 'Would do' vs 'Have already done'

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

EVstan 0.01***

0.02***

0.02*** 0.01***

0.01***

0.01***

(0.00)

(0.00)

(0.00) (0.00)

(0.00)

(0.00)

finsat -0.00

-0.00* -0.00 0.00

0.00 0.00

(0.00)

(0.00) (0.00) (0.00)

(0.00) (0.00)

represinverted

0.05** 0.07*** 0.09***

-0.00 0.04 0.01

(0.02) (0.02) (0.03)

(0.03) (0.04) (0.04)

2.x001sex

-0.34*** -0.27*** -0.33***

-0.11 -0.09 -0.11

(0.04) (0.04) (0.04)

(0.07) (0.07) (0.07)

x003age

-0.00*** -0.01*** -0.00***

0.01** 0.00* 0.01**

(0.00) (0.00) (0.00)

(0.00) (0.00) (0.00)

2.x025redulevel

0.20*** 0.23*** 0.20***

-0.03 -0.02 -0.03

(0.05) (0.05) (0.05)

(0.08) (0.08) (0.08)

3.x025redulevel

0.56*** 0.62*** 0.56***

0.15 0.18* 0.16*

(0.06) (0.06) (0.06)

(0.10) (0.10) (0.10)

2.x047r_incomelevel

0.03 0.06 0.04

-0.09 -0.08 -0.09

(0.05) (0.05) (0.05)

(0.08) (0.08) (0.08)

3.x047r_incomelevel

0.10* 0.17*** 0.13**

0.08 0.08 0.06

(0.05) (0.05) (0.05)

(0.08) (0.08) (0.08)

1.membership

0.41*** 0.45*** 0.42***

0.49*** 0.49*** 0.48***

(0.06) (0.06) (0.06)

(0.11) (0.11) (0.11)

loginfl

-0.10 -0.11

0.38** 0.38**

(0.08) (0.08)

(0.16) (0.15)

loggdpt1

-0.03 -0.04

0.56** 0.55***

(0.11) (0.11)

(0.22) (0.21)

logint

0.05

-0.10*

(0.04)

(0.06)

N 32107 31857 11156 11121 11121 13467 13362 5193 5179 5179

R2 . . . . . . . . . . Table entries are log-odds, i.e. the natural logarithm of odds, i.e. Log-odds = ln(p/(p-1)). Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

39

Table 3. Nested models. Unsanctioned occupation of buildings, factories .

'Would never do' vs 'Would do' 'Would do' vs 'Have already done'

(1) (2) (4) (5) (6) (7) (8) (9) (10) (11)

EVstan 0.00

0.00

0.00 -0.00

-0.00

-0.00

(0.00)

(0.00)

(0.00) (0.01)

(0.01)

(0.01)

represinverted

0.08*** 0.06* 0.07 0.00 0.00

(0.02) (0.03) (0.06) (0.00) (0.00)

2.x001sex

-0.29*** -0.28*** -0.29*** 0.33 0.33

(0.07) (0.07) (0.07) (0.64) (0.63)

x003age

-0.01*** -0.01*** -0.01*** 0.00 0.00 0.00 0.00 0.00

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

2.x025redulevel

0.15* 0.16* 0.15* -0.74 -0.75 -0.85* -0.02 -0.06

(0.09) (0.09) (0.09) (0.69) (0.68) (0.51) (0.77) (0.73)

3.x025redulevel

0.30*** 0.32*** 0.31*** -1.23* -1.21* -1.61*** -0.05 -0.36

(0.10) (0.10) (0.10) (0.67) (0.66) (0.55) (1.16) (0.95)

2.x047r_incomelevel

0.10 0.10 0.10 -0.05 -0.05 -0.72* 0.86 0.47

(0.09) (0.09) (0.09) (0.58) (0.58) (0.44) (1.19) (1.02)

3.x047r_incomelevel

0.23** 0.25*** 0.24*** -1.80* -1.76* -1.78** -2.74** -2.19**

(0.09) (0.09) (0.09) (0.92) (0.92) (0.84) (1.09) (1.10)

1.membership

0.04 0.03 0.02 1.24** 1.25** 0.42 1.09* 0.98*

(0.15) (0.15) (0.15) (0.59) (0.59) (0.45) (0.64) (0.59)

finsat -0.00

-0.00 -0.00

0.00 0.00 0.00

(0.00)

(0.00) (0.00)

(0.00) (0.00) (0.00)

loginfl

0.73*** 0.75***

0.87 0.64

(0.21) (0.23)

(0.60) (0.52)

loggdpt1

1.09*** 1.12***

1.22 0.89

(0.30) (0.33)

(0.84) (0.70)

logint

0.02

-0.09

(0.09)

(0.19)

N 13201 13145 9665 9630 9630 1293 1292 1185 1184 1184

R2 . . . . . . . . . . Table entries are log-odds, i.e. the natural logarithm of odds, i.e. Log-odds = ln(p/(p-1)). Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

40

Table 4. Demonstrations: 'Would never do' vs 'Would do' by country

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

Alb

ania

20

02

Alg

eria

20

02

Eth

iopia

20

07

Iraq

2006

Jord

an 2

00

1

Jord

an 2

00

7

Ky

rgy

zsta

n

2003

Mal

aysi

a

2006

Mo

rocc

o

2001

Mo

rocc

o

2007

Nig

eria

20

00

Ru

ssia

200

6

Rw

anda

2007

Sin

gap

ore

2002

Zim

bab

we

2001

Egy

pt

2000

Egy

pt

2008

Tan

zan

ia

2001

Bu

rkin

a F

aso

2007

Zam

bia

200

7

EVstan 0.01 0.01** -0.01 0.00 0.03**

* 0.02** 0.01* 0.01* 0.03**

* 0.03**

* 0.00 0.00 0.00 0.02**

* 0.00 -0.00 0.02**

* -0.01 0.01**

* -0.01

(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.00) (0.01) (0.00) (0.00) (0.00) (0.00) (0.01) (0.00) (0.01) (0.01) (0.01) (0.00)

finsat

-

0.01** -0.00

-

0.02**

* -0.00 -0.01 -0.00

-

0.01**

*

-

0.01**

* -0.00

0.02**

* 0.00**

-

0.00**

* 0.00*

-

0.01** -0.00 -0.00 0.00 -0.00 -0.00 0.00

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

2.x001se

x

-0.96**

*

-0.73**

* -0.23

-1.68**

*

-0.85**

*

-1.46**

*

-0.39**

* -0.22*

-0.74**

*

-0.66**

*

-0.80**

* -0.05 -0.21* -0.09

-0.55**

* 0.09

-0.73**

* -0.25

-0.56**

*

-0.46**

*

(0.18) (0.13) (0.14) (0.09) (0.22) (0.22) (0.14) (0.13) (0.10) (0.14) (0.10) (0.10) (0.12) (0.13) (0.14) (0.10) (0.14) (0.18) (0.14) (0.12)

x003age -0.01 0.00 0.01

-

0.01***

-0.02** -0.01 0.01 -0.01

-

0.02*** -0.01 -0.01

0.01*** -0.01

-

0.03***

-0.01** 0.00

-

0.02*** -0.00 0.00 -0.01*

(0.01) (0.01) (0.01) (0.00) (0.01) (0.01) (0.00) (0.01) (0.00) (0.01) (0.00) (0.00) (0.00) (0.01) (0.01) (0.00) (0.01) (0.01) (0.01) (0.00)

2.x025re

dulevel 0.25 0.36* -0.03 0.09 0.47 0.20

0.91**

* 0.11

0.79**

*

0.51**

*

0.39**

* 0.24

0.75**

*

0.39**

* 0.05 -0.07

0.69**

*

0.53**

*

1.14**

*

0.33**

*

(0.20) (0.19) (0.15) (0.11) (0.29) (0.30) (0.22) (0.18) (0.13) (0.18) (0.11) (0.17) (0.15) (0.14) (0.16) (0.12) (0.18) (0.20) (0.24) (0.12)

3.x025redulevel 0.31

0.83*** 0.20 0.12

1.06***

1.19***

1.79*** 0.56**

1.16***

1.60***

1.02***

0.59*** 0.85**

0.95***

-0.13 1.37**

* 0.80**

* 1.84**

* 1.08**

*

(0.27) (0.21) (0.23) (0.13) (0.27) (0.27) (0.24) (0.23) (0.20) (0.42) (0.13) (0.18) (0.34) (0.21)

(0.14) (0.20) (0.30) (0.52) (0.20)

1.membe

rship 0.25

0.50**

*

0.57**

*

0.33**

0.52**

* 0.12

1.03**

*

(0.19) (0.13)

(0.14)

(0.16)

(0.13) (0.20)

(0.20)

0o.membership

0.00 0.00 0.00 0.00

0.00

0.00 0.00 0.00 0.00

0.00 0.00

0.00 0.00

(0.00) (0.00) (0.00) (0.00)

(0.00)

(0.00) (0.00) (0.00) (0.00)

(0.00) (0.00)

(0.00) (0.00)

3o.x025r

edulevel

0.00

(0.00)

N 784 1175 1082 2302 1195 1175 1024 1198 2042 1062 1936 1876 1405 1490 947 2994 3048 1085 1137 1380

R2 . . . . . . . . . . . . . . . . . . . .

Table entries are log-odds, i.e. the natural logarithm of odds, i.e. Log-odds = ln(p/(p-1)). Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

41

Table 5. Demonstrations: 'Would do' vs 'Have already done' by country

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

A

lban

ia 2

002

Alg

eria

20

02

Eth

iopia

20

07

Iraq

2006

Jord

an 2

00

1

Jord

an 2

00

7

Ky

rgy

zsta

n

2003

Mal

aysi

a

2006

Mo

rocc

o

2001

Mo

rocc

o

2007

Nig

eria

20

00

Ru

ssia

200

6

Rw

anda

2007

Sin

gap

ore

2002

Zim

bab

we

2001

Egy

pt

2000

Egy

pt

2008

Tan

zan

ia

2001

Bu

rkin

a F

aso

2007

Zam

bia

200

7

EVstan 0.00 0.01** -0.01 -0.01* 0.03 -0.00 -0.00 0.02 0.02**

* 0.04**

* 0.01* -0.00 0.02**

* -0.00 -0.01 -0.00 0.01 -

0.01** 0.02**

* 0.01*

(0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01)

finsat 0.00 0.00

-

0.01**

*

0.01**

* 0.01 -0.00 0.01* 0.00

0.01**

*

0.02**

* 0.00 0.00

0.02**

* 0.01 -0.00 0.00 0.01 0.00 0.00 0.00

(0.00) (0.00) (0.00) (0.00) (0.01) (0.01) (0.00) (0.01) (0.00) (0.01) (0.00) (0.00) (0.00) (0.01) (0.01) (0.00) (0.01) (0.00) (0.00) (0.00)

2.x001sex

-0.91**

* -0.24 -0.18

-1.09**

*

-1.34**

* 0.02 0.23 -0.31

-0.60**

*

-0.73**

*

-0.70**

* 0.01 -0.17 -0.13

-1.14**

* 0.05 0.13 0.11

-0.60**

*

-

0.31**

(0.20) (0.19) (0.16) (0.16) (0.52) (0.45) (0.23) (0.41) (0.18) (0.22) (0.15) (0.16) (0.20) (0.39) (0.44) (0.26) (0.38) (0.15) (0.17) (0.15)

x003age -0.00

0.05**

* -0.01 -0.00 0.00 0.02

0.04**

* 0.00

0.03**

* 0.01 0.02**

0.02**

* 0.01

0.04**

*

0.05**

* -0.01 0.00 0.00 0.00

0.03**

*

(0.01) (0.01) (0.01) (0.00) (0.02) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01) (0.00) (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01)

2.x025redulev

el 0.37* 0.45 -0.17 0.14 -0.25 -0.24 0.32 0.06 0.44**

1.08**

* 0.46** 0.14 -0.14 0.15

1.54**

* 0.08 0.41

-

0.54**

*

0.60**

*

0.62**

*

(0.21) (0.29) (0.17) (0.15) (0.52) (0.54) (0.47) (0.66) (0.20) (0.25) (0.18) (0.25) (0.25) (0.44) (0.45) (0.32) (0.47) (0.17) (0.19) (0.17)

3.x025redulevel

0.69*** 0.66** 0.25 -0.16 -0.26 0.35 0.39 0.15

0.78***

1.03***

0.84*** 0.10 -0.09 -0.75 1.11 0.13 0.91* 0.07

1.87***

0.89***

(0.26) (0.30) (0.26) (0.17) (0.52) (0.48) (0.48) (0.76) (0.24) (0.37) (0.18) (0.27) (0.48) (0.71) (1.40) (0.37) (0.50) (0.20) (0.30) (0.21)

1.membership 0.09 0.40**

0.47**

0.29

0.38 2.20**

0.02

(0.21) (0.18)

(0.23)

(0.23)

(0.42) (1.05)

(0.21)

0o.membership

0.00 0.00 0.00 0.00

0.00

0.00 0.00 0.00 0.00

0.00 0.00

0.00 0.00

(0.00) (0.00) (0.00) (0.00)

(0.00)

(0.00) (0.00) (0.00) (0.00)

(0.00) (0.00)

(0.00) (0.00)

N 600 592 761 1123 115 133 491 337 838 495 1018 836 445 394 405 525 251 914 837 892

R2 . . . . . . . . . . . . . . . . . . . .

42

Appendix

Figure A. Potential demonstrators by the number of Internet users per 100 people

Source: WVS, World Bank

albania (200

algeria (200

ethiopia (20

iraq (2006)

jordan (2001jordan (2007

kyrgyzstan (

malaysia (20

morocco (200

morocco (200nigeria (200

russian fede

rwanda (2007

singapore (2

zimbabwe (20

egypt (2008)

tanzania (20

burkina faso

zambia (2007

11

.52

-2 0 2 4ln(internetusersper100people+.0484565)

demonstration Fitted values

43

Figure B. Petitioners by the number of Internet users per 100 people -

-

Source: WVS, World Bank

albania (200

algeria (200china (2007)

ethiopia (20

iraq (2006)

jordan (2001

jordan (2007

kyrgyzstan (malaysia (20

morocco (200

morocco (200

nigeria (200 russian federwanda (2007

singapore (2

zimbabwe (20

egypt (2000)

egypt (2008)

tanzania (20

burkina faso

zambia (2007

1.2

1.4

1.6

1.8

2

-2 0 2 4ln(internetusersper100people+.0484565)

petitiondone Fitted values

44

Figure C. Predicted number of demonstrators by emancipative values for each country

-

11

.52

2.5

0 20 40 60 80 100EmanValues

45

Table A. Descriptive statistics

Variable Obs Mean Std. Dev. Min Max Type of variable Transformation in the analysis Source

Individual Level Variables

sex 44926 1.510417 .499897 1 2 binomial WVS, x001

age 44891 36.03615 14.1875 15 97 continuous WVS, x003

education 44686 1.704136 .7366943 1 3 ordinal WVS, x025

income 20333 1.896179 .7883993 1 3 ordinal WVS, x047

EVstan 44987 29.1411 13.63909 0 100 continuous WVS

petitiondone 32107 1.52705 .6815428 1 3 ordinal Inverted WVS

demonstration 30750 1.520033 .7036179 1 3 ordinal Inverted WVS

occupy 13201 1.121279 .3903022 1 3 ordinal Inverted WVS

Aggregate Level Variables

repression inverted 42662 4.904154 1.878104 1 7 continuous inverted, squared CIRI

inflation, annual change% 12.55507 16.16304 -.3916769 76.70727 continuous Logged World Bank

unemployment, annual change% 11.03733 6.804557 3,3 26,8 continuous Logged World Bank

internet users per 100 8.492771 12.23103 .0640808 51.63799 continuous Logged World Bank

GINI 37.9372 6.667577 28.15 54.63 continuous Logged World Bank

FDI / foreign direct investment 2.886624 3.250644 -.2733369 15.32467 continuous World Bank

GDP ppp pc in constant prices 5879.616 6932.719 374.04 36340.95 continuous Logged Penn World Tables

46

GDP t-1 ppp pc in constant prices 5525.264 6712.162 392.92 35519.49 continuous Logged Penn World Tables

EmanValues 29.1411 5.381313 21.59546 43.82969 continuous WVS

PetitionsAGGR 1.537714 .1999096 1.15313 1.920986 continuous WVS

DenAGGR 1.569931 .2948002 1.096425 2.128472 continuous WVS

OccupyAGGR 1.123175 .1598456 1.004996 1.508963 continuous WVS

-

Table B. Dependent variables: coding procedure (output from STATA)

47

-

.

32762 .

11902 3 would never do

997 2 might do

302 1 have done

tabulation: Freq. Numeric Label

unique values: 3 missing .: 32762/45963

range: [1,3] units: 1

label: e029

type: numeric (byte)

e029 political action: occupying buildings or factories

15213 .

18533 3 would never do

8443 2 might do

3774 1 have done

tabulation: Freq. Numeric Label

unique values: 3 missing .: 15213/45963

range: [1,3] units: 1

label: e027

type: numeric (byte)

e027 political action: attending lawful/peaceful demonstrations

13856 .

18640 3 would never do

10012 2 might do

3455 1 have done

tabulation: Freq. Numeric Label

unique values: 3 missing .: 13856/45963

range: [1,3] units: 1

label: e025

type: numeric (byte)

e025 political action: signing a petition

48