Parliamentary Studies: Speeches and Votes

Dr Eitan Tzelgov

So...

The Art of Political Manipulation (William Riker, 1986)

“They win because they have set up the situation in such a way that other people will want to join them – even without any persuasion at all. And this is what heresthetic is all about: structuring the world so you can win” (Riker, preface)

alt text

Variables

  • As you know, can describe variables in may ways

  • One common way are measures of central tendency such as the mean and the median

  • Another common way are measures of spread such as standard deviation and variance

  • We can combine the two to make inferences about the world

plot of chunk normals_plot

The scientific method

  • We rely on theories (A theory is a set of interrelated concepts, definitions, and propositions that explains or predicts events or situations by specifying relations among variables.)

  • And derive hypotheses – more \( X \) –> more \( Y \)

  • We call X and Y variables

  • Variables can behave in various different ways and come in a variety of shapes and sizes

The Sound of Rebellion (Slapin & Kirkland, 2020)

alt text

Motivation and Questions

  • Rebellion is a costly act (rebels have lower prospects of promotion)

  • Government rebels are generally purists thinking the government should produce policies popular among party base

  • They grandstand

  • They use speeches to communicate their position to constituencies

  • Therefore: rebel speeches should be different

Different in what way? Rebel rhetoric

  • Communicate–>Less complex?

  • Costly act–>Different emotions?

  • “Owning it”–> Different use of pronouns?

bogey

Findings and importance

  • Rebels use less complex speeches and more first-person pronouns, make longer speeches

  • No clear pattern with regards to emotions

  • Speech patterns are predictive of rebellion

  • Tells us something about the psychology of rebellion

  • Opens a door for further research on the topic –> connecting votes and speeches

bogey

Class exercise: Our research Paper

  • Divide into two groups

  • Open your dataset

  • Write down two hypotheses

    • Relationship between Loyalists/Rebel and the length of speeches
    • Relationship between Loyalists/Rebel and tone (negative/neutral/positive)
    • Maybe.. think about the topics emphasized(equality, economy, taxes, poor people, society, discipline etc. )

Test your hypotheses

  • Length:

    • Using excel, calculate the mean length of Yes/No speeches
    • Does this finding support your hypothesis?
    • Produce a plot (barchart)?
    • Email everything to me

  • Sentiment:

    • In your group, code each speech negative/neutral/positive
    • Aggregate the two by vote
    • Does this finding support your hypothesis?
    • Produce a plot (barchart)?
    • Email everything to me

Your speeches analyzed:

Length

bogey

t.test(speeches$length~ speeches$vote)

Polarity

bogey

Specific emotions

bogey

Pronouns

bogey

bogey

Readability: specific speeches

bogey

Readability: overall

bogey

The Public Health Rebellion

bogey

What it was all about

  • Maggie Throup, the Under-Secretary of State for Vaccines and Public Health moved a public health motion in the Commons.

  • It's basically about mandating mask-wearing in certain public space

  • Throup said: “The Scientific Advisory Group for Emergencies (Sage) has found that face coverings are likely to reduce transmission through all routes by partially reducing emissions of and or exposure to the full range of aerosols and droplets that carry the virus.”

bogey

“Today's debate … is about how we react and the kind of nation and civilisation that we are creating in the context of this new disease. What is the relationship between the state and the individual? Are we to be empty vessels or mere automata-things to be managed, as if a problem? Or are we free spirits with, for want of a better term, a soul? This is a fundamental choice between heading towards heaven and heading towards hell…I am afraid that the Government are choosing that downward path towards, frankly, hell… I, for one, intend to chart a course towards heaven, and I hope that hon. Members will come with me.” (Steven Baker MP)

bogey

The vote

bogey

The Rebels

                    mps      parties
1          Adam Afriyie Conservative
2          Steven Baker Conservative
3          Graham Brady Conservative
4        Andrew Bridgen Conservative
5           Steve Brine Conservative
6     Christopher Chope Conservative
7            Greg Clark Conservative
8         Philip Davies Conservative
9     Jonathan Djanogly Conservative
10   Jackie Doyle-Price Conservative
11        Mark Francois Conservative
12          Marcus Fysh Conservative
13          Chris Green Conservative
14          Mark Harper Conservative
15       Mark Jenkinson Conservative
16       Pauline Latham Conservative
17         Andrew Lewer Conservative
18 Ian Liddell-Grainger Conservative
19       Karl McCartney Conservative
20         Esther McVey Conservative
21         Huw Merriman Conservative
22         John Redwood Conservative
23     Andrew Rosindell Conservative
24          Henry Smith Conservative
25           Greg Smith Conservative
26        Julian Sturdy Conservative
27       Desmond Swayne Conservative
28          Robert Syms Conservative
29         Craig Tracey Conservative
30       Charles Walker Conservative
31        William Wragg Conservative
32        Jeremy Wright Conservative

Rebels: Covid Recovery Group

The Covid Recovery Group (CRG) is thought to comprise of around 70 backbench Tory MPs, and they claim to have the support of many more. With prominent members including Steve Baker and chair of the 1922 Committee Sir Graham Brady, the group is styled on the once highly influential European Research Group (ERG).

(https://www.politicshome.com/news/article/tory-crg-coronavirus-lockdown-lift-rebels-boris-johnson)

How can we study this?

bogey

How can we study this?

bogey

What are our hypotheses?

bogey

Hypotheses

  1. Rebels are more likely to speak

  2. Rebels are going to make longer speeches

  3. Rebels express different emotions

  4. Rebels use different language (pronouns)

  5. Rebels use less complex language

Hypothesis 1: Rebels are more likely to speak

Likelihood of speaking: raw data

No Speech Speech
Voted No 24 11
Voted Yes 418 16
  • What do you think? Do we have support for hypothesis 1?

Likelihood of speaking: visual inspection

plot of chunk unnamed-chunk-2

Likelihood of speaking: Statistical test

  • It's called a \( \chi^2 \) test

  • It looks something like this:

plot of chunk unnamed-chunk-3

Likelihood of speaking: Statistical test

  • It basically compares our data with data that implies that voting and speaking to choose are independent

  • See that value up there? If our statistical test gives a higher number, the differences are really big

  • We have 40.97

  • It's off the chart. Rebels want to speak more!

Hypothesis 2: Rebels will make longer speeches

Length of speeches: Raw data

  • How do we test that?
mps vote nwords
Maggie Throup Aye 3349
Rachael Maskell Aye 89
Jim Shannon Aye 87
Andrew Murrison Aye 2187
Vicky Foxcroft Aye 69
Clive Betts Aye 240
Daisy Cooper Aye 1862
Alex Norris Aye 2082
Gary Sambrook Aye 82
Tom Hunt Aye 51
Luke Evans Aye 20
Huw Merriman Aye 2878
Ben Spencer Aye 971
Alec Shelbrooke Aye 326
Simon Hoare Aye 83
Bob Seely Aye 1393
Desmond Swayne No 118
Mark Harper No 2334
Steven Baker No 1426
William Wragg No 57
Karl McCartney No 98
Mark Jenkinson No 253
Graham Brady No 1235
Steve Brine No 2128
Sammy Wilson No 245
Christopher Chope No 1557

Length of speeches: Comparing means across groups

plot of chunk unnamed-chunk-5

Length of speeches: Comparing medians across groups

  • Rebels definitely talk more! We have support for hypothesis 2

plot of chunk unnamed-chunk-6

Hypothesis 3: Rebels express different emotions

Emotions: raw data

“I am very pleased that compulsory mask wearing has not been extended to the hospitality sector, as that would have been deeply damaging to businesses. What is the position of Her Majesty's Opposition on this issue, because I think I heard something slightly different just now from what I heard yesterday?”

Emotions: sentiment analysis and polarity

When we do sentiment analysis, we:

  • classify the polarity of a given text

  • We will classify each speech, based on the share of positive/negative words in sentences

  • Can be positive/negative/neutral

Emotions: sentiment analysis and polarity

  • Easy: “I like your dog”

  • Not so easy: “I don't like your dog”

  • Need to take care of valence shifters

Emotions: comparing polarity by group

There's no difference!!!

plot of chunk unnamed-chunk-7

Emotions: comparing polarity by MP

There's no difference!!!

plot of chunk unnamed-chunk-8

Problem: Short speeches

  • Most negative speech

“I am very pleased that compulsory mask wearing has not been extended to the hospitality sector, as that would have been deeply damaging to businesses. What is the position of Her Majesty's Opposition on this issue, because I think I heard something slightly different just now from what I heard yesterday?”

  • Most positive speech

Will the Minister give way? Will the Minister give way? My hon. Friend is the kindest of Ministers. On the question of the new variant's severity, I wonder if she has data to hand about whether any of the new cases of omicron in this country that she mentioned have been detected in patients hospitalised with covid.“

plot of chunk unnamed-chunk-9

No Support for hypothesis 3

That is what happens sometimes in science

bogey

Wait... let's look at specific feelings!

plot of chunk unnamed-chunk-10

plot of chunk unnamed-chunk-11

  • Not sure: leave for further research

Hypothesis 4: Rebels will use pronouns differently

Use of pronouns: Raw Data

  • How do we test that?
key freq pronoun
i 176 First
we 130 First
it 87 Third
they 66 Third
my 48 First
us 32 First
them 30 Third
she 29 Third
their 25 Third
our 23 First
her 12 Third
he 7 Third
his 6 Third
themselves 5 Third
its 2 Third
herself 1 Third
itself 1 Third
myself 1 First
him 1 Third

Pronouns: raw data by group

plot of chunk unnamed-chunk-13

plot of chunk unnamed-chunk-14

Pronouns: percentages of pronouns usage

plot of chunk unnamed-chunk-15

No Support for hypothesis 4

Welcome to the social scientist's life!

bogey

Hypothesis 5: Rebels will use simpler language

Readability: Speeches given a Flesch readability score

bogey

Readability of speeches

plot of chunk unnamed-chunk-16

Most complex/simple

  • Complex

“Have the Opposition done any analysis of how much it would cost to implement ventilation en masse across educational settings?”

  • Simple

“I wish to return to the point I made earlier, and I wonder whether my hon. Friend shares my genuine concern. As he said, we are here today because of a variant that has been discovered.”

Let's compare the groups

plot of chunk unnamed-chunk-17

No Support for hypothesis 5

Obligatory cat

bogey

Summary: what did we learn

  • That we can use the web to gather lots of data (email me if interested)

  • That rebel behavior (in our case) it significantly different in terms of likelihood of speaking and length of speeches

    • This is also a general, robust result

  • That we cannot conclude whether they are different in other regards (further research)

  • That we can use social scientific tools to study parliaments

  • That you don't always get what you want (as a social scientist)