Predictors of participation in protest in the Philippines

ABSTRACT

Social movement theory links protest to psychological (grievance, system justification, uncertainty salience), strategic (costs, trust, norms), resource-based (sex, age, income, group membership, political interest, media consumption), and structural (political opportunity structures) predictors. This paper answers the question, “Why do people protest in the Philippines?” by assessing whether these variables predict protest in the Philippines using data from the 2012 World Values Survey. Specifically, it tests the following hypotheses: (1) more altruistic people are more likely to engage in protest; (2) more trusting respondents are more likely to engage in protest; (3) the relationship between trust and protest differ among active group members, non-active group members, and non-members; and (4) the association between trust and protest and between altruism and protest will differ between low- and high-cost protests. Estimates using binary logistic regression suggest that sex, group membership, political interest, and consumption of certain types of media consistently predict participation in protest. In addition, there is also evidence that factors that predict high-cost protest action do not predict low-cost protest action. Lastly, trust and system justification were also found to predict protest but in directions opposite those established by previous research.

KEYWORDS:

• Social movement theory
• World Values Survey: Philippines
• collective action
• protest

Acknowledgements

The literature review and operational framework of this paper are largely based on a dissertation submitted in partial fulfillment of the MSc Political Science and Political Economy program of the London School of Economics and Political Science in August 2014. The author thanks Prof. Torun Dewan, Prof. Fiona Steele, Dr. Athena Lydia Casambre, Dr. Steve Rood, Dr. Rogelio Alicor Panao, Dr. Jean Encinas-Franco, Dr. Maria Ela L. Atienza, Clarinda L. Berja, Alexander Fouirnaies, Patrick Alcantara, Matthew Ho, Andreas Lind, Jacob Nyrup, Soh Guo Zhong, Esther Calvo, Carlo Borromeo, Gaylyn Brucal, Joshua Aguilar, Marian Valbuena, Samuel Cruz, Christopher Cuevas, Aaron Marquez, Patricia Del Rosario, Floricar Manalo, Shamille De Vera, Paula Morota, Earl Matammu, Maxine Joven, the participants of the panel on Democratic Transition in the 2016 International Conference of the Philippine Political Science Association, and two anonymous reviewers for their inputs. Any errors are the author’s.

Disclosure statement

The author’s maternal uncle is Benigno S. Aquino, III, President of the Philippines at the time that data for the 2012 World Values Survey was gathered.

Notes

1. EDSA refers to Epifanio de los Santos Avenue, the site of the 1986 revolt that removed the dictator Ferdinand Marcos from power (EDSA I) and the 2001 revolt (EDSA II) that removed President Joseph Estrada from power over corruption allegations.

2. This is the proportion of people who have signed a petition, joined a boycott, or attended a peaceful demonstration across the three waves. Joining a strike was not included as the 1996 and 2001 waves asked about joining unofficial strikes while the 2012 wave does not make this qualification.

3. Colloquially, the Philippines is divided into three island groups: Luzon, Visayas, and Mindanao with the National Capital Region (NCR) found in Luzon. Administratively, the Philippines is divided into 17 administrative regions: 8 in Luzon, 3 in Visayas, and 6 in Mindanao. However, the highest local government unit in the Philippines is the province, which is found within a region. Geographically, all cities and municipalities outside NCR are found within provinces, though administratively, some cities such as Davao City are not overseen by a province. The barangay is found within a city or municipality and is the smallest unit of local government in the country.

4. An alternative approach was used by Benson and Rochon (2004) where protest was treated as an ordinal variable with the following categories: (1) signed a petition, (2) signed a petition and joined a boycott, (3) signed a petition, joined a boycott and a peaceful demonstration, and (4) signed a petition, joined a boycott, peaceful demonstration, and unofficial strike, and (5) signed a petition, and joined a boycott, peaceful demonstration, and unofficial strike and occupied a building. This approach, however, is not used here because it does not account for those who, for instance, have joined a boycott but not signed a petition. It also imposes a hierarchy on the types of protest that may not apply in the Philippine case. For instance, while Benson and Rochon (2004) consider a demonstration more costly than a boycott, joining a demonstration is more frequent than joining a boycott in the Philippines. To analyze these types of participation in a single model would require a multinomial logistic regression model whose dependent variable would have 16 categories, and thus, 16 coefficients per independent variable, which would make interpretation unclear. Also, the binary formulation is used as opposed to the aggregated scales used by Dalton, van Sickle, and Weldon (2010) to avoid ambiguous equivalencies (e.g. treating signing a petition and joining a boycott as equivalent to attending a lawful demonstration and joining an official strike).

5. The author thanks an anonymous reviewer for suggesting this model.

6. There may also have been a translation issue with the binary trust question. In the English back translation of the Filipino questionnaire, the response “most people can be trusted” to V24 was back translated as “almost all people can be trusted.” Only 2.8% of Filipino respondents gave this response compared to 25% overall. On the other hand, the word “try” is omitted in the back translations of the responses for V56. Among all respondents in the 2012 World Values Survey, V24 and V56 are significantly correlated, but among Philippine respondents, the correlation between V24 and V56 is weaker.

7. Odds are defined as the ratio of the probability of one outcome to that of another outcome. In this case, the odds being modeled are the ratio of the probability of being protester to that of being a non-protester. Odds ratios describe the multiplicative difference between the odds of different groups. Findings using probit regression are qualitatively similar and are available upon request.

8. Beyond these six, several other models were fit using additional interaction terms, square terms, or variables with alternative specifications. Though discussed when relevant, these additional estimates are not shown here for the sake of brevity but are available upon request. Specifically pertaining to interaction terms, the author only shows the results for interaction terms that are explicitly suggested by theory and that improve model fit. This is because even though adding interaction terms can improve model fit and address problems of multicollinearity, doing so will prevent an overall relationship between the predictor and the outcome from being estimated, making coefficient interpretation unclear (Brambor, Clark, and Golder 2006). Thus, interaction terms for group membership and trust are included (Oliver 1984) while an interaction term between life satisfaction and democracy (Gurr 1968; Eisinger 193) and an interaction term between trust and altruism (Carter and Castillo 2003) were estimated but excluded because they did not significantly improve model fit. Lastly, models with island group interactions were included to account for the over-representation and under-representation of certain island groups in the data. The author also performed a variance inflation factor test to check for multicollinearity and found that no variable was significantly contributing to multicollinearity in the model.

9. Another set of models for protest without petition and with life satisfaction re-specified as a categorical variable with three categories (high: 7–9, medium: 3–6, and low: 0–2) show that the interaction term between democracy rating and the dummy variable for medium life satisfaction (baseline: low) is significant but negative, while both the term for democracy rating and the interaction term between democracy rating and high life satisfaction is insignificant. This suggests that at extreme values of life satisfaction on either end, there is no relationship between protest and democracy rating but at moderate values, a higher democracy rating is associated with lower odds of engaging in protest, which fits Kitschelt’s (1986) findings. However, the author refrains from drawing broader conclusions due to the arbitrary cutoffs for high, low, and medium life satisfaction.

10. For example, on Table 6 the estimated odds of protest in Balance Luzon when all independent variables are set to zero is e^{log(0.03) + log(1.19)} = 0.04, while a one-unit increase in self-rated income is associated with a 5% decrease (e^{log(0.63) + log(1.51)} = 0.95) in the odds of protest, holding all other variables constant.

Additional information

Notes on contributors

Francis Joseph A. Dee

Francis Joseph A. Dee The author is currently Assistant Professor in the Department of Political Science, University of the Philippines, Diliman. He completed his MSc in Political Science and Political Economy in the London School of Economics and Political Science.