SIS INTERNATIONAL RESEARCH: NYC FOCUS GROUP FACILITY
Our new GreenBook Directory site is live!
Your #1 strategic guide to consumer insights.
Qualtrics X4 The Experience Mgmt Summit
Brand & Retailer tickets for all IIeX events now start at just $99! Get or give one today!

Codifying Political Speech: Using Biometric Data to Identify Demographics

Ever wonder why some folks are naturally successful as politicians, while others are not? Through the study of communication and similarities amongst different politicians, we see how research can help us identify a candidate's appeal to the masses.

Editor’s Note: This post is part of our Big Ideas series, a column highlighting the innovative thinking and thought leadership at IIeX events around the world. Josipa Majic will be speaking at IIeX Europe 2020 in Amsterdam, Feb. 25-26. If you liked this article, you’ll LOVE IIeX Europe. Click here to learn more. 


Affective Computing has emerged as an important field of study that aims to develop systems that can automatically recognize emotions. We have decided to self-fund studies where we will be analyzing the most interesting digital habits, politics, and culture.

In this piece, we have explored the communication styles, similarities, and differences among several political leaders: Trump, Macron, May, Merkel, Trudeau, and Abe.

Our experimental protocol was designed to acquire the physiological responses of subjects during talks of all 6 different politicians, and the stimuli presentation was done via our proprietary AfC system and mobile app, to ensure full synchronization between stimuli shown and biometric data collected.

In this study, we have collected EEG, ECG, facial coding, speech sentiment analysis, and psychometrics to analyze emotional responses. We have analyzed longer political speeches, but have decided to highlight only snippets for the purpose of this blog post. We synchronize the data using the Lab streaming layer (LSL) and then use computer vision on recorded video footage in order to codify which stimulus elicits what reactions.

Participants

A group of 50 healthy volunteers students, suffering neither from cardiovascular nor evident mental pathologies, were recruited to participate in the experiment. They were balanced in terms of age (22 to 25 years old) and gender (50% male, 50% female). Inclusion criteria were as follows: age between 22 and 25 years; Croatian nationality; having no formal education in politics; having no previous experience of analyzing political speeches or communication styles.

Methodology

  • 50 students at the University of Zagreb Men and women aged 22-25 years, SEC(socio-economic classification) A/B.
  • Using digital products and mobile apps daily
  • Location: Zagreb, Croatia
  • Testing design
  • Prior to the first testing situation:
    • Respondents answered a predetermined set of psychological assessments and questionnaires to create a psychological profile and capture baseline data. The data baselining period for EEG was 60 sec.
  • In every testing situation: 
    • Respondents answer a question about their general mood
    • Respondents go through the testing experience following instructions on the mobile app.
    • Randomized and balanced stimuli order
  • Raw data collected:
    • EEG, EKG, facial coding, speech sentiment analysis, psychometrics
    • The respondents go through the testing of every stimulus on 2 separate occasions, where we collect their implicit and explicit feedback when they are evaluating the video stimuli features. This is done to eliminate any general mood biases on a given day.
    • Through the entire experience, the user is guided with the instructions on the mobile app that leaves timestamps on the affective feedback results and divides the testing into most relevant testing phases.
  • Testing order:
    • Using shuffle function Mersenne Twister Random Number Generator – our software randomly offered the respondent visual stimuli to test (the one that was not tested previously).
  • Key metrics:
    • We have used to data gathered to calculate the following metrics, Valence, Arousal, cognitive workload, stress levels.

Key Insights Gathered

What we have discovered is the following: each political leader we analyzed has a distinctive communication style that can be broken down into key metrics we analyzed. Those key metrics are emotional classification based on arousal and valence, cognitive workload and stress levels.

Trump: Keeping it Simple and Intense

Formula: high in arousal + moderate to high valence + low to mild stress levels + low cognitive workload.

Trump’s style can be easily defined as very emotionally charged throughout the duration of the talk, but combined with very low levels of cognitive workload. That essentially means his speeches are high in arousal, moderate to high valence, low to mild stress levels and low cognitive workload. This is ideal for the digital age as elements of his speeches remain very emotionally intense and memorable, but easy to understand.

Macron: Intellectually Curious and Complex

Formula: Low/moderate arousal +moderate valence +moderate stress + high cognitive workload

Macron’s communication style sparks curiosity and is intellectually intriguing, but it comes at a cost of high cognitive workload, meaning it can be difficult to follow or to share his messages in the digital world. His vocal cadence is able to rise arousal to moderate levels, while the pleasantness or valence is kept at moderate levels throughout this talks.

Merkel: Stable and Cerebral

Formula: Low arousal + low to moderate valence + high stress and high cognitive workload

Merkel’s style is best described by keeping both emotional elements, arousal and valence, relatively low. Not very exciting nor pleasant, but her speaking demands lots of attention due to high cognitive workload and moderate stress are detected.

May: Engaging but Nervous

Formula: High arousal + low/moderate valence +high stress levels + low cognitive workload

May has a very engaging communication style; arousal is kept high and paired with moderate valence. Although the cognitive workload isn’t high compared to the rest of the speakers, moderate to high stress is detected due to the choice of words and tone of voice.

Trudeau: Relaxed and Likeable

Formula: low arousal + high valence + moderate stress + moderate cognitive workload

Trudeau has very high valence during speeches, meaning he evokes positive emotions but due to low arousal they aren’t very strong nor memorable. His choice of topics and words may seldom evoke peaks in stress and cognitive workload metrics, but those are otherwise kept low.

Abe: Slow-Paced but Powerful

Formula: high arousal + high valence +moderate stress + moderate cognitive workload

Abe’s style can be described as relatively slow-paced, but resulting in very high arousal and valence, so being both intense and likable at the same time. At times his talks may evoke stress peaks, but otherwise, stress is kept at moderate levels – just like cognitive workload.

Please share...

Join the conversation

Josipa Majic

Josipa Majic

CEO & Founder, Tacit