Replication code for:

Mobilizing Europe’s citizens to take action on migration and climate change: Behavioral evidence from 27 EU member states (JEPP 2025)

Author

Hieko Giebler, Johannes Giesseke, Macartan Humphreys, Swen Hutter, Heike Klüver

Abstract

This paper investigates how two politicized issues—migration and climate change—mobilize citizens across European countries. Building on the concept of issue-specific mobilization potentials, we examine citizens’ willingness to support petitions related to the two issues using an original behavioral measure embedded in the 2024 European Parliament Election Study. We document variation in political engagement and examine how opposing stances on issues owned by the left or the right mobilize citizens, how citizens’ agreement with issue positions affects support, and whether grievances, participation cultures, politicization levels, and the ideology of the national government can explain national-level variation. Our results indicate substantial variation in petition support across countries and issues, with the right-wing petition on migration attracting the most support. However, our country-level measures do not explain this variation well. Overall, our findings highlight the need for more nuanced, issue-specific approaches to understanding cross-national patterns of political participation.

1 Setup

1.1 Import data

Dataset from EES 2024 merged with country level indicators.

Code

# source("data_prep.R")
data <- read_rds("assets/ees_petition_data.rds")

1.2 Labels

Labels defined for graphs.

Code

# Treatment group labels in data and for figures
tgg <- c("Climate (left)",
         "Climate (right)",
         "Migration (left)",
         "Migration (right)")
names(tgg) <- tgg

tg <- c("CLI left", "CLI right", "MIG left", "MIG right")
treatment_groups <- as.list(tg)
names(treatment_groups) <- tgg


# Core country covariates

vs <- c(
  "grievances",
  "protest_participants",
  "politicization",
  "lfirstp",
  "lsecondp",
  "gov_lr"
)
vs_labels <- c(
  "Grievances",
  "Protest participants\n(per million per year)",
  "Politicization\n(polarization x salience)",
  "Lijphart 1",
  "Lijphart 2",
  "Left / Right \n(government orientation)"
)

vars <- as.list(vs)
names(vars) <- vs

# Additional country covariates
vs_other <-
  c(
    "sal",
    "pol",
    "vulnerability",
    "cri",
    "ims_abs",
    "ims_perc",
    "ref_abs",
    "ref_perc",
    "protest_events",
    "protest_intensity"
  )
vs_other_labs <-
  c(
    "Salience",
    "Polarization",
    "Vulnerability",
    "Climate risk index",
    "Immigrants (absolute)",
    "Immigrants (per capita)",
    "Refugees (absolute)",
    "Refugees (per capita)",
    "Protest events",
    "Protest intensity"
  )

1.3 Prepare country df

Country level data frame.

Code

country_df <-
  data |> 
  group_by(country, left, topic, treatment_grp) |>
  summarize(position = mean(position, na.rm = TRUE), 
            outcome1b = mean(outcome1b, na.rm = TRUE),
            grievances = mean(grievances, na.rm = TRUE),
            politicization = mean(politicization, na.rm = TRUE),
            protest_participants = mean(protest_participants, na.rm = TRUE)) |>
  ungroup() |>
  mutate(groups = haven::as_factor(treatment_grp),
         Left = 1*(left == "Left"),
         Left = Left - mean(Left),
         Climate = 1*(topic == "Climate change"),
         Climate = Climate - mean(Climate)
         )

2 Results

2.1 Figure 1: Distribution of positions in migration and climate, country averages

Code

data |> 
  group_by(Country) |> 
  summarize(
    antimigration = mean(pos_mig, na.rm = TRUE)/10,
    anticlimate = mean(pos_cli, na.rm = TRUE)/10) |> 
  ggplot(aes(antimigration, anticlimate, label = Country)) +
  geom_text()  + 
  xlab("average support: right  position on migration")  + 
  ylab("average support: right  position on climate")

2.2 Figure 2: Share of respondents supporting petitions

Code

data |> 
  group_by(Country, topic, left) |> 
  summarize(petition = mean(outcome1b), .groups = 'drop') |> 
  pivot_wider(names_from = left, values_from = petition) |> 
  ggplot(aes(Right, Left, label = Country)) +  
    geom_point(size = 3) +  # Change the size of the points as needed
    geom_text_repel() +  # Use ggrepel to repel text labels
    facet_grid(~topic) + 
    geom_abline(color = "red") + 
    theme_minimal()  # Optional: change theme for better appearance

Share signing petitions of each type. Each dot represents a country average participation rate for Left and Right sides of a petition.

2.3 Figure 3: Petition support as a function of citizens’ position

Code

data |>  
  ggplot(aes(position, outcome1b)) +  
  geom_smooth()  + 
  facet_grid(topic ~  left) + ylim(0, NA)

Petition signing as a function of position on each issue. All are increasing but are remarkable high on the left of the x axis

2.4 Figure 4: Predicted probabilities

Sample estimation:

Code

# Fit the logistic mixed-effects model
if(run)
  rstanarm::stan_glmer(outcome1b ~ 1 + (1 | Country), data = data, family = binomial(link = "logit")) |>
  write_rds("saved/M_1.1.rds")

posterior_samples <- read_rds("saved/M_1.1.rds") |> as.data.frame()

Code

# Extract the posterior samples of the random effects

# Get the unique country names
Countries <- unique(data$Country)
Countries_list <- as.list(Countries)
names(Countries_list) <- Countries

# Function to calculate predicted probabilities for each country
get_predicted_probs <- function(country, fit = posterior_samples) {
  random_intercept <-  fit |> 
    select(starts_with(paste0("b[(Intercept) Country:", country))) 
  common_intercept <- fit$`(Intercept)`
  log_odds <- (common_intercept + random_intercept)[[1]]
  # Convert log-odds to probabilities
  data.frame(probs = exp(log_odds) / (1 + exp(log_odds)))
}

# Create a dataframe to store the results
results <- lapply(Countries_list, get_predicted_probs) |>
  bind_rows(.id = "Country")

# Summarize the predicted probabilities
summary_results <- results %>%
  group_by(Country) %>%
  summarize(
    expectation = mean(probs),
    min_prob = min(probs),
    max_prob = max(probs)
  ) |>
  left_join(data |> group_by(Country) |> summarize(raw = mean(outcome1b))) |>
  mutate(Country = fct_reorder(Country, expectation))


# Plot the results
ggplot(summary_results, aes(x = Country, y = expectation)) +
  geom_point() +
  geom_errorbar(aes(ymin = min_prob, ymax = max_prob)) +
  labs(title = "Predicted Probabilities of Petition Signing by Country",
       y = "Predicted Probability",
       x = "Country") +
  theme_minimal() + coord_flip() +
  geom_point(aes(Country, raw), color = "red")

Code

# Prep data so that position is in line with the petition content

decomp_model <- 
  function(g) 
  rstanarm::stan_glmer(
    outcome1b ~ 1 + position + (1 + position | Country),
    family = binomial(link = "logit"),
    data = data |> filter(groups == g) 
    )

if(run)
  treatment_groups |> 
  lapply(decomp_model)|>
  write_rds("saved/M_1.2.rds")

M_1.2 <-  read_rds("saved/M_1.2.rds")

posterior_1.2 <- lapply(M_1.2,  as.data.frame)

2.5 Figure 5: Model predictions on support probabilities

Code

lapply(posterior_1.2, 
       function(df)
         df|> select(b = position, a = `(Intercept)`)) |> 
  bind_rows(.id = "petition") |>
  mutate(y0 = 1/(1+exp(-a)),
         y1 = 1/(1+exp(-a-b)),
         change = y1 - y0) |>
  ggplot() + 
  geom_histogram(aes(x = y0), fill = "red", alpha = .5)  +
  geom_histogram(aes(x = y1), fill = "blue", alpha = .5) + facet_wrap(~petition) + 
  xlab("Signing propensity")

2.6 Figure 6: Decomposition of variation

Decomposition function:

Code

decomposition <- function(g) {
  
 names(posterior_1.2) <- treatment_groups
 posterior <- posterior_1.2[[g]]

 c_df <- country_df |>  filter(groups  == g)

 X_diff <- c_df$position - mean(c_df$position)
  

delta_Y  <- {

  aj <- 
    (posterior |> select(starts_with("b[(I")) |> apply(2, mean)) + 
    (posterior |> select(starts_with("(I")) |> apply(2, mean)) 
  
  bj <- 
    (posterior |> select(starts_with("b[po")) |> apply(2, mean)) + 
    (posterior |> pull(position) |> mean()) 
  
  Y = aj + bj*c_df$position 
  
  Y - mean(Y)
}

delta_X  <- {

  bj <- 
    (posterior |> select(starts_with("b[po")) |> apply(2, mean)) + 
    (posterior |> pull(position) |> mean()) 
  
  bj*X_diff
  }

delta_b <-  {
 a_diff <- posterior |> select(starts_with("b[(I")) |> apply(2, mean)
 b_diff <- posterior |> select(starts_with("b[po")) |> apply(2, mean)
 a_diff + b_diff*mean(c_df$position) 
 }


bind_cols(
  country_df |> filter(groups == g),
  data.frame(
    delta_Y  = delta_Y,
    delta_X  = delta_X,
    delta_b  = delta_b
  )
)  |>
  select(-left, -topic) |>
  mutate(Country = haven::as_factor(country)) 

}

Code

decomps <- lapply(treatment_groups, decomposition)

decomps |> bind_rows(.id = "condition") |> 
  select(condition, Country, starts_with("delta")) |>
  gather(var, val, -condition, - Country,  - delta_Y) |> 
  mutate(var = factor(var, c("delta_b", "delta_X"), c("Difference in average response", 
                                                      "Difference in average position"))) |>
  
  ggplot(aes(delta_Y, val, label = Country, color = var)) + geom_text()+
  xlab("Difference in outcomes") + ylab("contribution") + facet_wrap(~condition) +
  theme(legend.position = "bottom")

2.7 Figure 7: Country level correlates of engagement

Stan model:

Code

library(rstan)

stan_code <- "
data {
    int<lower=1> N;                // Number of observations
    int<lower=1> m;                // Number of countries
    int<lower=1,upper=m> Country[N]; // Country indices for each observation
    vector[N] y;                   // Observed data
    vector[m] X;                   // Country level predictor variable
}

parameters {
    vector[m] alpha_raw;           // Non-centered group-level intercepts
    real beta;                     // Coefficient for X
    real<lower=0> sigma;           // Error term
    real<lower=0> tau;             // Standard deviation of group-level intercepts
    real mu;                       // Mean of group-level intercepts
}

transformed parameters {
    vector[m] alpha;               // Centered group-level intercepts
    alpha = mu + tau * alpha_raw;  // Non-centered parameterization
}

model {
    // Priors
    alpha_raw ~ normal(0, 1);       // Standard normal for non-centered intercepts
    beta ~ normal(0, 2.5);          // Prior for beta
    sigma ~ normal(0, 1);           // Prior for sigma
    mu ~ normal(0, 10);             // Prior for mu 
    tau ~ normal(0, 5);             // Prior for tau
    
    // Likelihood
    for (i in 1:N) {
        y[i] ~ normal(alpha[Country[i]] + beta * X[Country[i]], sigma);
    }
}
"


f <- function(df, var, y = "outcome1b", ...) {
  
    names(df)[names(df) == var] <- "X"
    names(df)[names(df) == y] <- "y"
    
    # Select relevant columns and drop NA
    df <- df |> 
      select(X, y, Country) |> drop_na() |>
      mutate(Country = as.integer(factor(Country))) 

    stan_data <- list(
        N = nrow(df),
        m = length(unique(df$Country)),
        Country = df$Country,
        y = df$y,
        X = df |> group_by(Country) |> summarize(X = mean(X)) |> 
          pull(X) |>  as.vector()
    )

    fit <- stan(model_code = stan_code, data = stan_data, chains = 4, ...)

    return(fit)
}

Code

# Core variables
if(run)
  vars |>
  lapply(function(v) {
    treatment_groups |> lapply(function(g) {
    f(data |> filter(groups==g & position > .5), v, iter = 5000)})}) |>
  write_rds("saved/M_1.3_by_group.rds")

M_1.3_by <-  read_rds("saved/M_1.3_by_group.rds")


sumy <- function(v, g, model_list = M_1.3_by)
  model_list[[v]][[g]] |>
  as.data.frame() |> summarize(mean =  mean(beta), sd =  sd(beta), 
                               lower = quantile(beta, probs = .025),
                               upper = quantile(beta, probs = .975)) |>
  mutate(max_rhat = max(rhat(model_list[[v]][[g]])))


fig.1.3.by <- 
  vars[c(1:3, 6)] |> lapply(function(v)
  tgg |>
    lapply(function(g) sumy(v,g)) |> 
    bind_rows(.id = "group")) |>
    bind_rows(.id = "var") |>
  mutate(var = factor(var, vs, vs_labels)) |>
  ggplot(aes(mean, group)) + geom_point() + facet_wrap(~var, scales = "free_x") + 
  geom_errorbarh(aes(xmin = lower, xmax = upper, height = .2)) +
  geom_vline(xintercept = 0, color = "red")

fig.1.3.by

3 Appendices

3.1 Descriptives

3.1.1 Summary statistics on country level variables

Code

vlong <- c("position", "grievances", "politicization", "gov_lr",
          "protest_participants", "sal", "pol", "vulnerability", "cri",  
          "ims_perc",   "ref_perc", "protest_events",   "protest_intensity",
          "outcome1b")

cdf <-   data |> 
  group_by(country, left, topic, treatment_grp) |>
  summarize(across(all_of(vlong), \(x) mean(x, na.rm = TRUE)), .groups = 'drop')  |>
  ungroup() |>
  rename(support_petition = outcome1b)


cdf |>
st()

Summary Statistics
Variable	N	Mean	Std. Dev.	Min	Pctl. 25	Pctl. 75	Max
left	108
... Right	54	50%
... Left	54	50%
topic	108
... Migration	54	50%
... Climate change	54	50%
position	108	0.5	0.093	0.3	0.44	0.56	0.72
grievances	108	0.35	0.25	0	0.15	0.53	1
politicization	108	0.24	0.2	0	0.1	0.34	1
gov_lr	108	0.21	0.61	-1	-0.26	0.75	1
protest_participants	104	0.24	0.28	0.0027	0.055	0.3	1.1
sal	108	0.51	0.053	0.41	0.48	0.55	0.69
pol	108	3	0.22	2.6	2.9	3.2	3.5
vulnerability	108	0.33	0.033	0.27	0.3	0.35	0.4
cri	108	76	27	18	58	106	106
ims_perc	108	13	9	2.2	6.1	17	48
ref_perc	108	1.8	1	0.46	0.8	2.6	4.2
protest_events	104	639	594	38	270	725	2324
protest_intensity	104	-0.00048	0.87	-0.85	-0.56	0.069	2.4
support_petition	108	0.21	0.061	0.099	0.17	0.25	0.41

3.1.2 Descriptives on petition versions

Code

country_summary

Numbers and shares signing by petition type
left	topic	n	signing
Right	Migration	6456	0.259
Right	Climate change	6569	0.216
Left	Migration	6442	0.184
Left	Climate change	6437	0.184

text 3

Code

data |> group_by(left, topic) |> 
  filter(position > .5) |> 
  summarize(n = n(), 
            signing = mean(outcome1b), 
            .groups = "drop") |> ungroup() |>
  kable(digits = 3, caption = "Numbers and shares signing by petition type among supporters (position >0.5)")

Numbers and shares signing by petition type among supporters (position >0.5)
left	topic	n	signing
Right	Migration	3021	0.312
Right	Climate change	2030	0.253
Left	Migration	2144	0.212
Left	Climate change	2889	0.217

Code

av <- 
  data |> 
  group_by(Country) |> 
  summarize(`migration left` = mean(outcome1b[topic =="Migration" & left =="Left"], na.rm = TRUE),
            `migration right` = mean(outcome1b[topic =="Migration" & left !="Left"], na.rm = TRUE),
            `climate left` = mean(outcome1b[topic !="Migration" & left =="Left"], na.rm = TRUE),
            `climate right` = mean(outcome1b[topic !="Migration" & left !="Left"], na.rm = TRUE)) 

cor(av[,-1]) |> kable(caption = "Country level correlations in signing across petitions", digits = 2)

Country level correlations in signing across petitions
	migration left	migration right	climate left	climate right
migration left	1.00	-0.08	0.45	0.18
migration right	-0.08	1.00	0.58	0.72
climate left	0.45	0.58	1.00	0.57
climate right	0.18	0.72	0.57	1.00

3.2 Tables for core analyses

We provide tables of estimates provided in the text.

3.2.1 Figure 4

Code

summary_results |>  
  kable(digits = 2, col.names = c("Country", "Predicted mean", "Lower", "Upper", "Raw mean"),
                caption = "Table of results used for Figure 4")

Table of results used for Figure 4
Country	Predicted mean	Lower	Upper	Raw mean
Austria	0.19	0.15	0.23	0.18
Belgium	0.23	0.19	0.28	0.24
Bulgaria	0.23	0.19	0.28	0.23
Croatia	0.24	0.19	0.28	0.24
Cyprus	0.27	0.21	0.33	0.28
Czech	0.22	0.17	0.26	0.22
Denmark	0.23	0.18	0.28	0.23
Estonia	0.20	0.16	0.25	0.20
Finland	0.13	0.10	0.17	0.12
France	0.18	0.14	0.22	0.17
Germany	0.21	0.16	0.25	0.21
Greece	0.21	0.16	0.25	0.21
Hungary	0.24	0.19	0.29	0.25
Ireland	0.21	0.17	0.25	0.21
Italy	0.17	0.13	0.21	0.17
Latvia	0.20	0.16	0.25	0.20
Lithuania	0.20	0.16	0.24	0.20
Luxembourg	0.18	0.13	0.23	0.17
Malta	0.28	0.22	0.36	0.30
Netherlands	0.24	0.20	0.29	0.24
Poland	0.20	0.16	0.26	0.20
Portugal	0.22	0.18	0.27	0.22
Romania	0.26	0.20	0.30	0.26
Slovakia	0.19	0.16	0.24	0.19
Slovenia	0.20	0.16	0.25	0.20
Spain	0.25	0.20	0.30	0.25
Sweden	0.18	0.14	0.22	0.18

3.2.2 Figure 5

Code

lapply(posterior_1.2, 
       function(df)
         df|> select(b = position, a = `(Intercept)`)) |> 
  bind_rows(.id = "petition") |>
  mutate(y0 = 1/(1+exp(-a)),
         y1 = 1/(1+exp(-a-b)),
         change = y1 - y0) |> 
  group_by(petition) |>
  summarize(y_0 = mean(y0),
         sd_0 = sd(y0),
         y_1 = mean(y1),
         sd_1 = sd(y1),
         difference = mean(change),
         sd_difference = sd(change), .groups = 'drop') |>
  kable(digits = 2, col.names = c("Petition", "Rates | opposed", "sd", "Rates given support", "sd", "Difference", "sd"), align = c("l", "c", "c", "c", "c", "c", "c"),
                caption = "Table of results used for Figure 5")

Table of results used for Figure 5
Petition	Rates \| opposed	sd	Rates given support	sd	Difference	sd
Climate (left)	0.13	0.01	0.24	0.01	0.12	0.02
Climate (right)	0.20	0.01	0.25	0.01	0.05	0.02
Migration (left)	0.16	0.01	0.22	0.02	0.07	0.02
Migration (right)	0.19	0.01	0.33	0.02	0.14	0.02

3.2.3 Figure 6

Code

decomps |> bind_rows(.id = "condition") |> 
  select(Petition = condition, Country, starts_with("delta")) |>
  kable(digits = 2, col.names = c("Petition", "Country", "delta Y", "delta X", "delta b"),
                caption = "Table of results used for Figure 6")

Table of results used for Figure 6
Petition	Country	delta Y	delta X	delta b
Climate (left)	Belgium	0.09	-0.02	0.11
Climate (left)	Bulgaria	0.19	0.01	0.18
Climate (left)	Czech Republic	-0.26	-0.04	-0.23
Climate (left)	Denmark	0.21	-0.01	0.21
Climate (left)	Germany	0.00	-0.06	0.06
Climate (left)	Estonia	-0.17	-0.10	-0.07
Climate (left)	Ireland	0.05	-0.04	0.08
Climate (left)	Greece	0.04	0.04	0.00
Climate (left)	Spain	0.25	0.02	0.22
Climate (left)	France	-0.29	0.00	-0.29
Climate (left)	Croatia	-0.02	0.08	-0.10
Climate (left)	Italy	-0.15	0.09	-0.24
Climate (left)	Cyprus	0.45	0.07	0.37
Climate (left)	Latvia	-0.09	-0.08	-0.01
Climate (left)	Lithuania	-0.23	-0.05	-0.18
Climate (left)	Luxembourg	-0.10	0.01	-0.11
Climate (left)	Hungary	0.18	0.06	0.12
Climate (left)	Malta	0.36	0.13	0.22
Climate (left)	Netherlands	0.14	-0.08	0.21
Climate (left)	Austria	-0.16	-0.02	-0.14
Climate (left)	Poland	-0.20	-0.05	-0.15
Climate (left)	Portugal	0.29	0.08	0.21
Climate (left)	Romania	0.27	0.00	0.27
Climate (left)	Slovenia	-0.15	0.00	-0.15
Climate (left)	Slovakia	-0.18	-0.04	-0.14
Climate (left)	Finland	-0.33	0.00	-0.33
Climate (left)	Sweden	-0.16	0.00	-0.16
Climate (right)	Belgium	0.09	0.01	0.08
Climate (right)	Bulgaria	0.12	0.00	0.12
Climate (right)	Czech Republic	-0.02	0.04	-0.05
Climate (right)	Denmark	0.01	0.00	0.00
Climate (right)	Germany	0.02	0.00	0.01
Climate (right)	Estonia	0.04	0.04	0.00
Climate (right)	Ireland	-0.16	0.01	-0.17
Climate (right)	Greece	-0.05	0.00	-0.04
Climate (right)	Spain	0.08	0.00	0.08
Climate (right)	France	-0.02	0.00	-0.01
Climate (right)	Croatia	0.18	-0.02	0.20
Climate (right)	Italy	-0.30	-0.03	-0.26
Climate (right)	Cyprus	0.05	-0.02	0.08
Climate (right)	Latvia	0.01	-0.01	0.02
Climate (right)	Lithuania	-0.03	0.03	-0.06
Climate (right)	Luxembourg	-0.05	-0.02	-0.03
Climate (right)	Hungary	0.16	-0.01	0.17
Climate (right)	Malta	0.21	0.00	0.21
Climate (right)	Netherlands	0.02	0.01	0.01
Climate (right)	Austria	-0.14	0.00	-0.14
Climate (right)	Poland	0.07	0.02	0.05
Climate (right)	Portugal	-0.13	-0.04	-0.08
Climate (right)	Romania	0.18	0.00	0.18
Climate (right)	Slovenia	0.09	-0.01	0.10
Climate (right)	Slovakia	0.01	0.01	0.00
Climate (right)	Finland	-0.30	0.01	-0.31
Climate (right)	Sweden	-0.13	0.00	-0.13
Migration (left)	Belgium	0.04	-0.01	0.05
Migration (left)	Bulgaria	-0.06	0.02	-0.08
Migration (left)	Czech Republic	-0.28	0.03	-0.31
Migration (left)	Denmark	0.14	0.01	0.13
Migration (left)	Germany	0.10	0.06	0.04
Migration (left)	Estonia	-0.19	-0.03	-0.16
Migration (left)	Ireland	0.15	0.00	0.15
Migration (left)	Greece	0.13	-0.03	0.16
Migration (left)	Spain	0.24	0.01	0.23
Migration (left)	France	-0.15	-0.02	-0.13
Migration (left)	Croatia	0.19	0.01	0.17
Migration (left)	Italy	-0.03	0.02	-0.05
Migration (left)	Cyprus	-0.10	-0.05	-0.05
Migration (left)	Latvia	-0.05	0.00	-0.06
Migration (left)	Lithuania	-0.05	-0.03	-0.02
Migration (left)	Luxembourg	-0.29	0.00	-0.29
Migration (left)	Hungary	-0.19	0.00	-0.18
Migration (left)	Malta	0.11	-0.04	0.15
Migration (left)	Netherlands	0.01	0.01	0.00
Migration (left)	Austria	0.19	0.01	0.18
Migration (left)	Poland	0.04	-0.01	0.05
Migration (left)	Portugal	0.16	-0.01	0.17
Migration (left)	Romania	0.34	0.06	0.28
Migration (left)	Slovenia	-0.32	0.00	-0.32
Migration (left)	Slovakia	-0.04	0.00	-0.05
Migration (left)	Finland	-0.30	0.03	-0.33
Migration (left)	Sweden	0.23	-0.03	0.26
Migration (right)	Belgium	0.14	-0.02	0.16
Migration (right)	Bulgaria	0.17	-0.03	0.20
Migration (right)	Czech Republic	0.60	-0.06	0.66
Migration (right)	Denmark	0.08	-0.03	0.11
Migration (right)	Germany	-0.15	-0.05	-0.10
Migration (right)	Estonia	0.08	0.04	0.04
Migration (right)	Ireland	-0.20	0.05	-0.25
Migration (right)	Greece	-0.18	0.04	-0.22
Migration (right)	Spain	0.21	0.03	0.18
Migration (right)	France	-0.24	0.02	-0.26
Migration (right)	Croatia	0.13	-0.02	0.16
Migration (right)	Italy	-0.22	-0.01	-0.21
Migration (right)	Cyprus	0.51	0.13	0.39
Migration (right)	Latvia	-0.04	0.01	-0.04
Migration (right)	Lithuania	0.00	0.07	-0.06
Migration (right)	Luxembourg	-0.03	0.00	-0.02
Migration (right)	Hungary	0.41	-0.01	0.43
Migration (right)	Malta	0.54	0.14	0.41
Migration (right)	Netherlands	0.28	-0.06	0.35
Migration (right)	Austria	-0.44	0.00	-0.44
Migration (right)	Poland	-0.15	0.01	-0.15
Migration (right)	Portugal	-0.12	0.00	-0.12
Migration (right)	Romania	0.05	-0.06	0.11
Migration (right)	Slovenia	0.03	0.00	0.03
Migration (right)	Slovakia	-0.20	-0.07	-0.12
Migration (right)	Finland	-0.71	-0.02	-0.68
Migration (right)	Sweden	-0.55	-0.01	-0.54

3.2.4 Figure 7

Code

  vars[c(1:3, 6)] |> lapply(function(v)
  tgg |>
    lapply(function(g) sumy(v,g)) |> 
    bind_rows(.id = "group")) |>
    bind_rows(.id = "var") |>
  kable(digits = 2, 
        col.names=c("Covariate", "Petition", "Posterior mean", "Posterior sd", "Cred lower" , "Cred upper", "Rhat"),
        caption = "Table of results used for Figure 7. Rhat is a `stan`  convergence diagnostic.")

Table of results used for Figure 7. Rhat is a `stan` convergence diagnostic.
Covariate	Petition	Posterior mean	Posterior sd	Cred lower	Cred upper	Rhat
grievances	Climate (left)	0.09	0.04	0.02	0.17	1
grievances	Climate (right)	0.06	0.04	-0.03	0.14	1
grievances	Migration (left)	-0.02	0.08	-0.18	0.13	1
grievances	Migration (right)	0.02	0.10	-0.19	0.22	1
protest_participants	Climate (left)	-0.01	0.04	-0.09	0.06	1
protest_participants	Climate (right)	-0.03	0.04	-0.11	0.04	1
protest_participants	Migration (left)	0.06	0.05	-0.05	0.16	1
protest_participants	Migration (right)	-0.04	0.07	-0.19	0.10	1
politicization	Climate (left)	0.06	0.05	-0.05	0.16	1
politicization	Climate (right)	0.07	0.05	-0.04	0.17	1
politicization	Migration (left)	-0.05	0.08	-0.20	0.11	1
politicization	Migration (right)	0.10	0.10	-0.11	0.30	1
gov_lr	Climate (left)	-0.02	0.02	-0.05	0.02	1
gov_lr	Climate (right)	-0.01	0.02	-0.05	0.02	1
gov_lr	Migration (left)	-0.01	0.02	-0.06	0.04	1
gov_lr	Migration (right)	-0.05	0.03	-0.11	0.01	1

3.3 Frequentist results

3.3.1 Country level models, across all petition types

Code

lm_models <-
  list(
    lm_robust(outcome1b ~ position, data = country_df),
    lm_robust(outcome1b ~ Left*Climate, data = country_df),
    lm_robust(outcome1b ~ Left*Climate + position, data = country_df),
    lm_robust(outcome1b ~ Left*Climate*grievances + position,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*politicization + position,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*protest_participants + position, data = country_df),
    lm_robust(outcome1b ~ Left*Climate*grievances + Left*Climate*politicization + Left*Climate*protest_participants + position, data = country_df)
)

lm_models |> texreg::htmlreg(include.ci = FALSE)

Statistical models
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
(Intercept)	0.13^***	0.21^***	0.16^***	0.14^**	0.15^**	0.15^**	0.13^**
	(0.04)	(0.01)	(0.04)	(0.04)	(0.05)	(0.05)	(0.05)
position	0.17^*		0.11	0.11	0.09	0.13	0.11
	(0.07)		(0.09)	(0.09)	(0.09)	(0.09)	(0.09)
Left		-0.06^***	-0.06^***	-0.06^**	-0.04^*	-0.06^***	-0.05^*
		(0.01)	(0.01)	(0.02)	(0.02)	(0.02)	(0.02)
Climate		-0.02^*	-0.02^*	-0.06^***	-0.03	-0.02	-0.05^*
		(0.01)	(0.01)	(0.02)	(0.02)	(0.02)	(0.02)
Left:Climate		0.05^*	0.02	0.03	-0.01	0.03	-0.01
		(0.02)	(0.03)	(0.04)	(0.03)	(0.03)	(0.05)
grievances				0.04			0.03
				(0.03)			(0.03)
Left:grievances				0.00			0.01
				(0.06)			(0.06)
Climate:grievances				0.10			0.08
				(0.06)			(0.06)
Left:Climate:grievances				-0.02			0.01
				(0.12)			(0.12)
politicization					0.05		0.04
					(0.03)		(0.03)
Left:politicization					-0.05		-0.06
					(0.05)		(0.06)
Climate:politicization					0.05		0.01
					(0.05)		(0.06)
Left:Climate:politicization					0.13		0.14
					(0.11)		(0.12)
protest_participants						-0.01	-0.01
						(0.02)	(0.02)
Left:protest_participants						0.03	0.04
						(0.04)	(0.04)
Climate:protest_participants						-0.01	-0.00
						(0.04)	(0.04)
Left:Climate:protest_participants						-0.06	-0.07
						(0.08)	(0.08)
R²	0.06	0.29	0.30	0.36	0.35	0.31	0.41
Adj. R²	0.06	0.27	0.27	0.31	0.30	0.25	0.30
Num. obs.	108	108	108	108	108	104	104
RMSE	0.06	0.05	0.05	0.05	0.05	0.05	0.05
^*p < 0.001; ^p < 0.01; ^*p < 0.05

3.3.2 Country level models, across all petition types, including country fixed effects

Code

fe_models <-
  list(
    lm_robust(outcome1b ~ position, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate + position, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*grievances + position, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*politicization + position, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*protest_participants + position, fixed_effects = ~country,  data = country_df),
    lm_robust(outcome1b ~ Left*Climate*grievances + Left*Climate*politicization + Left*Climate*protest_participants + position, fixed_effects = ~country,  data = country_df)

)
fe_models |> texreg::htmlreg(include.ci = FALSE)

Statistical models
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
position	0.17^**		0.10	0.11	0.09	0.13^*	0.12
	(0.06)		(0.06)	(0.06)	(0.06)	(0.06)	(0.06)
Left		-0.06^***	-0.06^***	-0.06^***	-0.04^**	-0.06^***	-0.05^**
		(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)
Climate		-0.02^**	-0.02^**	-0.03	-0.02	-0.02	-0.02
		(0.01)	(0.01)	(0.03)	(0.01)	(0.01)	(0.04)
Left:Climate		0.05^**	0.02	0.03	-0.01	0.04	-0.01
		(0.02)	(0.02)	(0.03)	(0.02)	(0.03)	(0.03)
grievances				0.00			-0.01
				(0.02)			(0.03)
Left:grievances				0.00			0.01
				(0.04)			(0.05)
Climate:grievances				0.03			0.02
				(0.08)			(0.08)
Left:Climate:grievances				-0.02			0.01
				(0.08)			(0.09)
politicization					0.03		0.04
					(0.04)		(0.04)
Left:politicization					-0.05		-0.06
					(0.05)		(0.05)
Climate:politicization					0.02		0.03
					(0.05)		(0.05)
Left:Climate:politicization					0.13		0.14
					(0.10)		(0.11)
protest_participants

Left:protest_participants						0.03	0.04
						(0.02)	(0.02)
Climate:protest_participants						-0.01	-0.02
						(0.02)	(0.03)
Left:Climate:protest_participants						-0.06	-0.07
						(0.04)	(0.05)
R²	0.45	0.67	0.68	0.68	0.70	0.70	0.72
Adj. R²	0.26	0.55	0.56	0.54	0.56	0.56	0.55
Num. obs.	108	108	108	108	108	104	104
RMSE	0.05	0.04	0.04	0.04	0.04	0.04	0.04
^*p < 0.001; ^p < 0.01; ^*p < 0.05

In these models, Left and Migration are centered on zero and so the “main” terms report estimated average effects.

3.3.3 Individual level models, across all petition types

Code

data <- mutate(data, Left = 1*(left == "Left"), Climate = 1*(topic == "Climate change"))

lm_models_i <-
  list(
    lm_robust(outcome1b ~ position, data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate, data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate + position, data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate*grievances + position,  data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate*politicization + position,  data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate*protest_participants + position, data = data, cluster = Country, se_type = 'stata'),
    lm_robust(outcome1b ~ Left*Climate*grievances +  Left*Climate*politicization + 
                Left*Climate*protest_participants + 
                position, data = data, cluster = Country, se_type = 'stata')
  )
    
lm_models_i |> texreg::htmlreg(include.ci = FALSE)

Statistical models
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
(Intercept)	0.17^***	0.26^***	0.21^***	0.22^***	0.19^***	0.22^***	0.20^***
	(0.01)	(0.01)	(0.01)	(0.02)	(0.03)	(0.02)	(0.04)
position	0.10^***		0.09^***	0.09^***	0.09^***	0.09^***	0.09^***
	(0.01)		(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Left		-0.08^***	-0.06^***	-0.08^*	-0.03	-0.08^**	-0.06
		(0.02)	(0.02)	(0.03)	(0.03)	(0.02)	(0.03)
Climate		-0.04^***	-0.03^**	-0.08^***	-0.03	-0.04^*	-0.06
		(0.01)	(0.01)	(0.02)	(0.02)	(0.01)	(0.03)
Left:Climate		0.04^*	0.02	0.04	-0.01	0.04	0.01
		(0.02)	(0.02)	(0.03)	(0.03)	(0.02)	(0.04)
grievances				-0.04			-0.04
				(0.07)			(0.07)
Left:grievances				0.06			0.05
				(0.10)			(0.10)
Climate:grievances				0.13			0.09
				(0.07)			(0.07)
Left:Climate:grievances				-0.06			-0.03
				(0.11)			(0.11)
politicization					0.06		0.08
					(0.09)		(0.09)
Left:politicization					-0.11		-0.12
					(0.11)		(0.11)
Climate:politicization					0.01		-0.03
					(0.09)		(0.09)
Left:Climate:politicization					0.11		0.11
					(0.12)		(0.12)
protest_participants						-0.03	-0.03
						(0.04)	(0.05)
Left:protest_participants						0.06	0.07
						(0.04)	(0.04)
Climate:protest_participants						0.02	0.03
						(0.03)	(0.04)
Left:Climate:protest_participants						-0.06	-0.06
						(0.05)	(0.05)
R²	0.01	0.01	0.01	0.01	0.01	0.01	0.01
Adj. R²	0.01	0.01	0.01	0.01	0.01	0.01	0.01
Num. obs.	24943	25904	24943	24943	24943	23978	23978
RMSE	0.41	0.41	0.41	0.41	0.41	0.41	0.41
N Clusters	27	27	27	27	27	26	26
^*p < 0.001; ^p < 0.01; ^*p < 0.05

3.3.4 Individual level models, across all petition types, including country fixed effects

Code

fe_models_i <-
  list(
    lm_robust(outcome1b ~ position, data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate, data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate + position, data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate*grievances + position,  data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate*politicization + position,  data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate*protest_participants + position, data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country),
    lm_robust(outcome1b ~ Left*Climate*grievances +  Left*Climate*politicization + 
                Left*Climate*protest_participants + 
                position, data = data, cluster = Country, se_type = 'stata', fixed_effects = ~country)
 )



fe_models_i |> texreg::htmlreg(include.ci = FALSE)

Statistical models
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
position	0.10^***		0.09^***	0.09^***	0.09^***	0.09^***	0.09^***
	(0.01)		(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Left		-0.08^***	-0.06^***	-0.08^*	-0.03	-0.08^**	-0.06
		(0.02)	(0.02)	(0.03)	(0.03)	(0.02)	(0.03)
Climate		-0.04^***	-0.03^**	-0.05	-0.02	-0.04^**	-0.02
		(0.01)	(0.01)	(0.03)	(0.02)	(0.01)	(0.03)
Left:Climate		0.04^*	0.02	0.04	-0.01	0.04	0.01
		(0.02)	(0.02)	(0.03)	(0.03)	(0.02)	(0.04)
grievances				-0.04			-0.04
				(0.06)			(0.06)
Left:grievances				0.06			0.05
				(0.10)			(0.10)
Climate:grievances				0.06			0.03
				(0.08)			(0.08)
Left:Climate:grievances				-0.07			-0.03
				(0.11)			(0.11)
politicization					0.07		0.09
					(0.08)		(0.07)
Left:politicization					-0.11		-0.11
					(0.11)		(0.11)
Climate:politicization					-0.04		-0.04
					(0.07)		(0.07)
Left:Climate:politicization					0.11		0.11
					(0.12)		(0.12)
protest_participants

Left:protest_participants						0.06	0.07
						(0.04)	(0.04)
Climate:protest_participants						0.02	0.01
						(0.03)	(0.04)
Left:Climate:protest_participants						-0.06	-0.06
						(0.05)	(0.05)
R²	0.01	0.01	0.02	0.02	0.02	0.02	0.02
Adj. R²	0.01	0.01	0.02	0.02	0.02	0.02	0.02
Num. obs.	24943	25904	24943	24943	24943	23978	23978
RMSE	0.41	0.41	0.41	0.41	0.41	0.41	0.41
N Clusters	27	27	27	27	27	26	26
^*p < 0.001; ^p < 0.01; ^*p < 0.05

3.4 Duration and signing

Evidence on whether “speeders” are more or less likely to sign.

Code

data |> mutate(signing = factor(outcome1b)) |> 
  ggplot(aes(duration, linetype = signing)) + 
  geom_density() +  xlim(0, 2500)

3.5 Additional Bayesian analyses

The “grievance” measure in the text is generated using a migration measure for migration petitions and climate measure for climate petitions. The “politicization” measure is generated by combining salience and polarization measures.

The below analyses implement Analysis 3 using these subcomponents and close substitute measures.

Code

if(run)
  vs_other |>
  lapply(function(v) {
    treatment_groups |> lapply(function(g) {
    f(data |> filter(groups==g & position > .5), v, iter = 5000)})}) |>
  write_rds("M_1.3_by_group_other.rds")

M_1.3_by_group_other <-  read_rds("saved/M_1.3_by_group_other.rds")
names(M_1.3_by_group_other) <- vs_other

names(vs_other) <- vs_other

fig.1.3.by_appendix_df <- 
  vs_other |> lapply(function(v)
  tgg |>
    lapply(function(g) sumy(v,g, model_list = M_1.3_by_group_other)) |> 
    bind_rows(.id = "group")) |>
    bind_rows(.id = "var") |>
  mutate(var = factor(var, vs_other, vs_other_labs))

fig.1.3.by_appendix_df |> filter(max_rhat < 1.01) |>
  ggplot(aes(mean, group)) + geom_point() + facet_wrap(~var, scales = "free_x") + 
  geom_errorbarh(aes(xmin = lower, xmax = upper, height = .2)) +
  geom_vline(xintercept = 0, color = "red")

In response to comments by reviewers, the below analyses implement Analysis 3 using additional measures capturing Lijphart’s distinction between majoritarian and consensus democracy to investigate the effect of political opportunity structures.

Code

fig.1.3.by.extra <- 
  vars[c(4,5)] |> lapply(function(v)
  tgg |>
    lapply(function(g) sumy(v,g)) |> 
    bind_rows(.id = "group")) |>
    bind_rows(.id = "var") |>
  mutate(var = factor(var, vs, vs_labels)) |>
  ggplot(aes(mean, group)) + geom_point() + facet_wrap(~var, scales = "free_x") + 
  geom_errorbarh(aes(xmin = lower, xmax = upper, height = .2)) +
  geom_vline(xintercept = 0, color = "red")

fig.1.3.by.extra