How rebelliousness are Kate Hoey and Frank Field?

rebellion.R

library(reshape2)
library(tidyverse)
library(hrbrthemes)
library(ggridges)

most_common <- function(x) {
    retval <- names(sort(table(x, useNA = "no"),
                         decreasing=TRUE))[1]
    retval <- ifelse(is.null(retval),
                     NA_character_,
                     retval)
}


years <- c("1997", "2001", "2005", "2010",
           "2015", "2017")

## You will need first to have downloaded the contents of 
## https://www.publicwhip.org.uk/data/

holder <- list()
for (y in years) {
    infile <- paste0("votematrix-", y, ".dat")
    tmp <- read.delim(infile, sep = "\t", quote = "")
    tmp$X <- NULL
    tmp <- melt(tmp,
                id.vars = c("rowid", "date", "voteno", "Bill"),
                variable.name = "mpid",
                value.name = "vote") %>% 
        mutate(vote = dplyr::recode(vote,
                                    `1` = "Aye",
                                    `2` = "Aye",
                                    `-9` = NA_character_,
                                    `4` = "Nay",
                                    `5` = "Nay",
                                    `3` = "Abstain",
                                    .default = NA_character_),
               mpid = gsub("mpid", "", mpid))

    ## Get in the IDs
    aux <- read.delim(paste0("votematrix-", y, ".txt"),
                      skip = 19, sep = "\t")
    aux$mpid <- as.character(aux$mpid)
    old <- nrow(tmp)
    tmp <- merge(tmp, aux, by = "mpid", all.x = TRUE, all.y = FALSE)
    new <- nrow(tmp)
    stopifnot(old == new)
    ## Calculate rebellion rates
    tmp <- tmp %>%
        group_by(party, rowid) %>%
        mutate(party_line = most_common(vote)) %>%
        ungroup() %>%
        mutate(rebelled = vote != party_line) %>%
        group_by(mpid, firstname, surname, party) %>%
        summarize(rebelliousness = mean(rebelled, na.rm = TRUE))
    tmp$session <- y
    holder[[y]] <- tmp
}

dat <- bind_rows(holder)

plot.df <- dat %>%
    filter(party == "Lab")

frank.field.ids <- c("197", "883", "1410",
                     "40079", "41187", "41397")
kate.hoey.ids <- c("282", "966", "1897",
                   "40624", "40828", "41992")

highlight.df <- subset(plot.df, mpid %in% c(frank.field.ids, kate.hoey.ids))

p1 <- ggplot(plot.df, aes(x = session, y = rebelliousness,
                    group = interaction(firstname, surname, party))) +
    geom_line(alpha = 0.1) +
    geom_line(data = highlight.df, alpha = 1, colour = 'red') +
    geom_text(data = subset(highlight.df, session == 2017),
              aes(label = surname),
              hjust = 0,
              nudge_x = 0.1) + 
    scale_x_discrete("Parliament") +
    scale_y_sqrt("Rebelliousness (%)\n(square-root scale)") +
    labs(title = "Rates of rebellion amongst Labour MPs, 1997-2018",
         subtitle = "Hoey and Field have been consistently rebellious",
         caption = "Data: www.publicwhip.org.uk") +
    theme_ipsum_rc()

png(file = "rebrates.png", width = 800, height = 640)
p1
dev.off()