23/replication_package/Code/replicate_fig_3.R

##--##--##--##
## Figure 3
##--##--##--##

us_survey_wave1$identity_protesters_fac = as.factor(us_survey_wave1$identity_protesters)
levels(us_survey_wave1$identity_protesters_fac) = c("White", "Black")
us_survey_wave1$tactic_fac = as.factor(us_survey_wave1$tactic)
levels(us_survey_wave1$tactic_fac) = c("March in streets", "Shut down traffic", "Destroy police cars")

isr_survey_wave1$identity_protesters_fac = as.factor(isr_survey_wave1$identity_protesters)
levels(isr_survey_wave1$identity_protesters_fac) = c("White", "Ethiopian", "Arab")
isr_survey_wave1$tactic_fac = as.factor(isr_survey_wave1$tactic)
levels(isr_survey_wave1$tactic_fac) = c("March in streets", "Shut down traffic", "Destroy garbage cans")


## US sample
us_survey_wave1$degree_violence_std = scale(us_survey_wave1$degree_violence)
us_survey_wave1$police_action_required_std = scale(us_survey_wave1$police_action_required)
us_survey_wave1$recall_violence2_std = scale(us_survey_wave1$recall_violence2)

march1 = lm_robust(degree_violence_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==0,], weights=weight)
shut1 = lm_robust(degree_violence_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==1,], weights=weight)
destroy1 = lm_robust(degree_violence_std ~ identity_protesters_fac , data=us_survey_wave1[us_survey_wave1$tactic==2,], weights=weight)

march2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==0,], weights=weight)
shut2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==1,], weights=weight)
destroy2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==2,], weights=weight)

march3 = lm_robust(recall_violence2_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==0,], weights=weight)
shut3 = lm_robust(recall_violence2_std ~ identity_protesters_fac, data=us_survey_wave1[us_survey_wave1$tactic==1,], weights=weight)
destroy3 = lm_robust(recall_violence2_std ~ identity_protesters_fac , data=us_survey_wave1[us_survey_wave1$tactic==2,], weights=weight)

# Plot differences

term_degree_violence = c(march1$coefficients[2], shut1$coefficients[2], destroy1$coefficients[2])
se_degree_violence = c(march1$std.error[2], shut1$std.error[2], destroy1$std.error[2])
statistic_degree_violence = c(march1$statistic[2], shut1$statistic[2], destroy1$statistic[2])
pval_degree_violence = c(march1$p.value[2], shut1$p.value[2], destroy1$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
degree_violence = as.data.frame(cbind(term_degree_violence, se_degree_violence, statistic_degree_violence, pval_degree_violence))
degree_violence = cbind(degree_violence, term)
colnames(degree_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(degree_violence) = term
degree_violence$model = "Perceived degree of violence"

term_police_action_required = c(march2$coefficients[2], shut2$coefficients[2], destroy2$coefficients[2])
se_police_action_required = c(march2$std.error[2], shut2$std.error[2], destroy2$std.error[2])
statistic_police_action_required = c(march2$statistic[2], shut2$statistic[2], destroy2$statistic[2])
pval_police_action_required = c(march2$p.value[2], shut2$p.value[2], destroy2$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
police_action_required = as.data.frame(cbind(term_police_action_required, se_police_action_required, statistic_police_action_required, pval_police_action_required))
police_action_required = cbind(police_action_required, term)
colnames(police_action_required) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(police_action_required) = term
police_action_required$model = "Police action required"

term_recall_violence = c(march3$coefficients[2], shut3$coefficients[2], destroy3$coefficients[2])
se_recall_violence = c(march3$std.error[2], shut3$std.error[2], destroy3$std.error[2])
statistic_recall_violence = c(march3$statistic[2], shut3$statistic[2], destroy3$statistic[2])
pval_recall_violence = c(march3$p.value[2], shut3$p.value[2], destroy3$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
recall_violence = as.data.frame(cbind(term_recall_violence, se_recall_violence, statistic_recall_violence, pval_recall_violence))
recall_violence = cbind(recall_violence, term)
colnames(recall_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(recall_violence) = term
recall_violence$model = "Recall violence"

differences_us = rbind(degree_violence, police_action_required, recall_violence)


## Israel respondents
isr_survey_wave1$degree_violence_std = scale(isr_survey_wave1$degree_violence)
isr_survey_wave1$police_action_required_std = scale(isr_survey_wave1$police_action_required)
isr_survey_wave1$recall_violence2_std = scale(isr_survey_wave1$recall_violence2)

march1 = lm_robust(degree_violence_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==0,], weights=weight)
shut1 = lm_robust(degree_violence_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==1,], weights=weight)
destroy1 = lm_robust(degree_violence_std ~ identity_protesters_fac , data=isr_survey_wave1[isr_survey_wave1$tactic==2,], weights=weight)

march2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==0,], weights=weight)
shut2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==1,], weights=weight)
destroy2 = lm_robust(police_action_required_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==2,], weights=weight)

march3 = lm_robust(recall_violence2_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==0,], weights=weight)
shut3 = lm_robust(recall_violence2_std ~ identity_protesters_fac, data=isr_survey_wave1[isr_survey_wave1$tactic==1,], weights=weight)
destroy3 = lm_robust(recall_violence2_std ~ identity_protesters_fac , data=isr_survey_wave1[isr_survey_wave1$tactic==2,], weights=weight)


# Ethiopian protesters
term_degree_violence = c(march1$coefficients[2], shut1$coefficients[2], destroy1$coefficients[2])
se_degree_violence = c(march1$std.error[2], shut1$std.error[2], destroy1$std.error[2])
statistic_degree_violence = c(march1$statistic[2], shut1$statistic[2], destroy1$statistic[2])
pval_degree_violence = c(march1$p.value[2], shut1$p.value[2], destroy1$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
degree_violence = as.data.frame(cbind(term_degree_violence, se_degree_violence, statistic_degree_violence, pval_degree_violence))
degree_violence = cbind(degree_violence, term)
colnames(degree_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(degree_violence) = term
degree_violence$model = "Perceived degree of violence"

term_police_action_required = c(march2$coefficients[2], shut2$coefficients[2], destroy2$coefficients[2])
se_police_action_required = c(march2$std.error[2], shut2$std.error[2], destroy2$std.error[2])
statistic_police_action_required = c(march2$statistic[2], shut2$statistic[2], destroy2$statistic[2])
pval_police_action_required = c(march2$p.value[2], shut2$p.value[2], destroy2$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
police_action_required = as.data.frame(cbind(term_police_action_required, se_police_action_required, statistic_police_action_required, pval_police_action_required))
police_action_required = cbind(police_action_required, term)
colnames(police_action_required) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(police_action_required) = term
police_action_required$model = "Police action required"

term_recall_violence = c(march3$coefficients[2], shut3$coefficients[2], destroy3$coefficients[2])
se_recall_violence = c(march3$std.error[2], shut3$std.error[2], destroy3$std.error[2])
statistic_recall_violence = c(march3$statistic[2], shut3$statistic[2], destroy3$statistic[2])
pval_recall_violence = c(march3$p.value[2], shut3$p.value[2], destroy3$p.value[2])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
recall_violence = as.data.frame(cbind(term_recall_violence, se_recall_violence, statistic_recall_violence, pval_recall_violence))
recall_violence = cbind(recall_violence, term)
colnames(recall_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(recall_violence) = term
recall_violence$model = "Recall violence"

differences_isr_black = rbind(degree_violence, police_action_required, recall_violence)

term_degree_violence = c(march1$coefficients[3], shut1$coefficients[3], destroy1$coefficients[3])
se_degree_violence = c(march1$std.error[3], shut1$std.error[3], destroy1$std.error[3])
statistic_degree_violence = c(march1$statistic[3], shut1$statistic[3], destroy1$statistic[3])
pval_degree_violence = c(march1$p.value[3], shut1$p.value[3], destroy1$p.value[3])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
degree_violence = as.data.frame(cbind(term_degree_violence, se_degree_violence, statistic_degree_violence, pval_degree_violence))
degree_violence = cbind(degree_violence, term)
colnames(degree_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(degree_violence) = term
degree_violence$model = "Perceived degree of violence"

term_police_action_required = c(march2$coefficients[3], shut2$coefficients[3], destroy2$coefficients[3])
se_police_action_required = c(march2$std.error[3], shut2$std.error[3], destroy2$std.error[3])
statistic_police_action_required = c(march2$statistic[3], shut2$statistic[3], destroy2$statistic[3])
pval_police_action_required = c(march2$p.value[3], shut2$p.value[3], destroy2$p.value[3])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
police_action_required = as.data.frame(cbind(term_police_action_required, se_police_action_required, statistic_police_action_required, pval_police_action_required))
police_action_required = cbind(police_action_required, term)
colnames(police_action_required) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(police_action_required) = term
police_action_required$model = "Police action required"

term_recall_violence = c(march3$coefficients[3], shut3$coefficients[3], destroy3$coefficients[3])
se_recall_violence = c(march3$std.error[3], shut3$std.error[3], destroy3$std.error[3])
statistic_recall_violence = c(march3$statistic[3], shut3$statistic[3], destroy3$statistic[3])
pval_recall_violence = c(march3$p.value[3], shut3$p.value[3], destroy3$p.value[3])
term = c("1) Minority: March in streets", "2) Minority: Shut down traffic", "3) Minority: Destroy police cars / garbage cans")
recall_violence = as.data.frame(cbind(term_recall_violence, se_recall_violence, statistic_recall_violence, pval_recall_violence))
recall_violence = cbind(recall_violence, term)
colnames(recall_violence) = c("estimate", "std.error", "statistic", "p.value", "term")
rownames(recall_violence) = term
recall_violence$model = "Recall violence"

differences_isr_arab = rbind(degree_violence, police_action_required, recall_violence)

## Result tables for plot
differences_us2 = differences_us
differences_us2$submodel = differences_us2$term
differences_us2$term = "Perception of Blacks \n(United States)"

differences_isr_black2 = differences_isr_black
differences_isr_black2$submodel = differences_isr_black2$term
differences_isr_black2$term = "Perception of Ethiopians \n(Israel)"

differences_isr_arab2 = differences_isr_arab
differences_isr_arab2$submodel = differences_isr_arab2$term
differences_isr_arab2$term = "Perception of Arabs \n(Israel)"

diffs_sm = rbind(differences_us2, differences_isr_arab2, differences_isr_black2)
rownames(diffs_sm) = 1:nrow(diffs_sm)

results_df <- data.frame(term = diffs_sm$term,
                         estimate = diffs_sm$estimate,
                         std.error = diffs_sm$std.error,
                         model = diffs_sm$model,
                         submodel = as.character(diffs_sm$submodel),
                         stringsAsFactors = FALSE)

results_df$submodel[results_df$submodel=="1) Minority: March in streets"] = "March in streets"
results_df$submodel[results_df$submodel=="2) Minority: Shut down traffic"] = "Shut down traffic"
results_df$submodel[results_df$submodel=="3) Minority: Destroy police cars / garbage cans"] = "Destroy property"

results_df$model[results_df$model=="Perceived degree of violence"] = "1. Perceived degree \nof violence"
results_df$model[results_df$model=="Police action required"] = "3. Police action \nrequired"
results_df$model[results_df$model=="Recall violence"] = "2. Recall \nviolence"

small_multiple(results_df, dot_args = list(aes(shape = submodel))) + 
  ylab("Coefficient Estimate (Std. Dev. Units)") + theme_bw()+
  geom_hline(yintercept = 0, colour = "grey60", linetype = 2) +
  #theme(plot.margin=unit(c(0.4,0.1,2.6,1),"cm"))+
  theme(legend.position = "bottom", 
        legend.justification=c(0, 0),
        # legend.background = element_rect(color="white"),
        legend.spacing = unit(-4, "pt"),
        legend.key.size = unit(10, "pt"))+
  scale_shape_discrete(name = "Protester Tactic") + 
  scale_color_manual(values=c("#999999", "#E69F00", "#56B4E9"), name = "Protester Tactic")
ggsave(filename="Figures/fig_3.pdf", width=6, height=6)

table_a14 = results_df