Presentation groups

Remember: topic to me at least by Week 4

Opening questions

How is the ‘average’ party structured? What do members, representatives, leaders do?

Percentage of women MPs - hunch

md`## Poll results (Respondents: ${respondentCount})`

import { liveGoogleSheet } from "@jimjamslam/live-google-sheet";
import { aq, op } from "@uwdata/arquero";

// UPDATE THE LINK FOR A NEW POLL
surveyResults = liveGoogleSheet(
  "https://docs.google.com/spreadsheets/d/e/" +
    "2PACX-1vQu1_SNVBYH6TAZ0ZUBFpwmrjzRVsvGgJcL8Y_gN7ec7pIlf70c6SOhWdOz44R7SDzzAdL06cHU4jzt/" +
    "pub?gid=1194358505&single=true&output=csv",
  10000, 1, 7); // adjust the last number to select all relevant columns 

respondentCount = surveyResults.length;

highest_womenCounts = aq.from(surveyResults)
  .select("highest_women")
  .groupby("highest_women")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
highest_women_maxCountRE = Math.max(...highest_womenCounts.objects().map(d => d.count));

plot_highest_women = Plot.plot({
  marks: [
    Plot.barY(highest_womenCounts, {
      x: "highest_women",
      y: "count",
      fill: "highest_women",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(highest_womenCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, highest_women_maxCountRE]
  },
  facet: { data: highest_womenCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

lowest_womenCounts = aq.from(surveyResults)
  .select("lowest_women")
  .groupby("lowest_women")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
lowest_women_maxCountRE = Math.max(...lowest_womenCounts.objects().map(d => d.count));

plot_lowest_women = Plot.plot({
  marks: [
    Plot.barY(lowest_womenCounts, {
      x: "lowest_women",
      y: "count",
      fill: "lowest_women",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(lowest_womenCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, lowest_women_maxCountRE]
  },
  facet: { data: lowest_womenCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

Oldest/youngest MPs - hunch

youngestCounts = aq.from(surveyResults)
  .select("youngest")
  .groupby("youngest")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
youngest_maxCountRE = Math.max(...youngestCounts.objects().map(d => d.count));

plot_youngest = Plot.plot({
  marks: [
    Plot.barY(youngestCounts, {
      x: "youngest",
      y: "count",
      fill: "youngest",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(youngestCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, youngest_maxCountRE]
  },
  facet: { data: youngestCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

oldestCounts = aq.from(surveyResults)
  .select("oldest")
  .groupby("oldest")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
oldest_maxCountRE = Math.max(...oldestCounts.objects().map(d => d.count));

plot_oldest = Plot.plot({
  marks: [
    Plot.barY(oldestCounts, {
      x: "oldest",
      y: "count",
      fill: "oldest",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(oldestCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, oldest_maxCountRE]
  },
  facet: { data: oldestCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

Percentage of MPs with migration background - hunch

highest_migCounts = aq.from(surveyResults)
  .select("highest_mig")
  .groupby("highest_mig")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
highest_mig_maxCountRE = Math.max(...highest_migCounts.objects().map(d => d.count));

plot_highest_mig = Plot.plot({
  marks: [
    Plot.barY(highest_migCounts, {
      x: "highest_mig",
      y: "count",
      fill: "highest_mig",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(highest_migCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, highest_mig_maxCountRE]
  },
  facet: { data: highest_migCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

lowest_migCounts = aq.from(surveyResults)
  .select("lowest_mig")
  .groupby("lowest_mig")
  .count()
  .derive({ measure: d => "" })

// Calculate the maximum count from your dataset
lowest_mig_maxCountRE = Math.max(...lowest_migCounts.objects().map(d => d.count));

plot_lowest_mig = Plot.plot({
  marks: [
    Plot.barY(lowest_migCounts, {
      x: "lowest_mig",
      y: "count",
      fill: "lowest_mig",
      stroke: "black",
      strokeWidth: 1
    }),
    Plot.ruleY([respondentCount], { stroke: "#ffffff99" })
  ],
  color: {
    domain: [
      "CDU.CSU",
      "AfD",
      "SPD",
      "Gruene",
      "Linke"
    ],
    range: [
      "black",
      "#009EE0",
      "#EB001F",
      "#64A12D",
      "#BE3075"
    ]
  },
  marginBottom: 80,
  x: { label: "", tickSize: 2, tickRotate: -25,
    domain: ["CDU.CSU", "AfD", "SPD", "Gruene", "Linke"]
  },  
  y: {
    label: "",
    tickSize: 10,
    tickFormat: d => d,
    tickValues: Array.from(
      new Set(lowest_migCounts.objects().map(d => d.count))
    ).sort((a, b) => a - b),
    domain: [0, lowest_mig_maxCountRE]
  },
  facet: { data: lowest_migCounts, x: "measure", label: "" },
  marginLeft: 140,
  style: {
    width: 1350,
    height: 500,
    fontSize: 30,
  }
});

_

Now, let’s see how those hunches measure up to the facts

Percentage of women MPs

Oldest/youngest MPs

Percentage of MPs with migration background

Refresher on party membership concepts

• party activists or members tend to hold more radical views than the average party voter
  • median voter theorem: parties should converge toward the centre to win elections, but…
  • activists/members: typically more committed and engaged; typically have stronger ideological
    • impact through party primaries and internal decision-making

Variation among far-right parties

• Common traits
  • predominantly male activist/member bases
  • typically less (but not ‘no’) ethnic/racial diversity in membership and representation

Factionalism within far-right parties

• frequent divisions between national and regional/local level representatives
• open question of whether local representatives more radical or more pragmatic

Party leaders - functions (Salvini)

Party leaders - support, symbolism, cooperation

What (party) leaders do

• ‘leading tasks’ (cf. Earl 2007 on leadership in movements):
  1. articulating vision and ideology,
  2. engaging the political environment,
  3. framing the party and its issues,
  4. managing relations with other parties and non-party actors,
  5. making strategic decisions,
  6. organizing specific actions/campaigns,
  7. managing the internal life of the party,
  8. innovating and entrepreneurial activity, and
  9. providing social capital (relationships and networks)

When things go wrong…

Party leaders - punitive

Le Pen 2025 embezzlement conviction - 5-year ban on holding office

Party leaders - prohibitive

Golden Dawn leader Nikos Michaloliakos arrested (Greece)

Marine Le Pen (Front National/Rassemblement National)

• took over party leadership from father (J.M. Le Pen) in 2011
  • FN history of under-performing with women voters
    • ‘radical right gender gap’ → decreasing with Marine as leader

Radical right voters (Weeks et al. 2022)

do these patterns hold outside Europe? Outside ‘the West’?

Women far-right representatives (Weeks et al. 2022)

When Do Männerparteien Elect Women?

Women far-right representatives

Weeks et al. (2022)

RQ: Under what conditions do we see an increase in RRP parties’ women MPs? Given that RRP parties have traditionally held very conservative views on the role of women in society, what accounts for this increase in women’s representation?

• Theory: strategic descriptive representation: “attract new voters without necessarily altering core issue positions” (p2)
• Data: European women MPs and voter support: 187 parties, 30 countries, 1985-2018
  • three sets of factors: party ideology, party organizational structures, and women’s activism within the party
    
• Methods: Multilevel random intercept models (inferential stats); two typical, “on-the-line” studies: PVV in 2017 and SVP in 2015

Weeks et al. (2022) - only radical right

Weeks et al. (2022)

Reading a regression table

Remember: regression is a tool for understanding a phenomenon as a linear function (generally) → (y = mx + b)

1. Numbers not in parentheses next to a variable: regression coefficient: expected change in DV for a one-unit increase in IV. NB: ositive or negative relationship?

2. Numbers inside parentheses next to a variable: standard error: estimate of the standard deviation of the coefficient

3. Asterisks/‘stars’: statistical significance: probability of results as extreme as observed result, under the assumption that the null hypothesis is correct. Smaller p-value means such an observation would be less likely under null hypothesis; hence, significance. Statistical significance suggests more precise estimates—NOT necessarily that one IV is more important than another.

Weeks et al. (2022)

Weeks et al. (2022)

Replication and rebuttal (Guinaudeau and Jankowski 2024)

Reply (Weeks et al. 2024)

• Guinaudeau and Jankowski (2024) misunderstand treatment of outliers
• should not ‘arbitrarily delete one influential case but not another’
  • “When all influential cases are removed, and not just the observation of FN in 2012, our key findings do not change.”

Deleting correctly-coded higher influence observations makes the sample an inaccurate representation of the population, and thus can lead to improper inferences about a population, particularly in a small population like the data on radical right populist parties in Europe.