Machine learning approaches to facial and text analysis: Discovering CEO oral communication styles

• Published date: 01 November 2019
• Author: Prithwiraj Choudhury,Tarun Khanna,Natalie A. Carlson,Dan Wang
• DOI: http://doi.org/10.1002/smj.3067
• Date: 01 November 2019

RESEARCH ARTICLE

Machine learning approaches to facial and text

analysis: Discovering CEO oral communication

styles

Prithwiraj Choudhury

1

| Dan Wang

2

| Natalie A. Carlson

2

|

Tarun Khanna

1

1

Harvard Business School, Boston,

Massachusetts

2

Columbia Business School, New York,

New York

Correspondence

Prithwiraj Choudhury, Harvard Business

School, Boston, MA.

Email: pchoudhury@hbs.edu

Abstract

Research Summary:We demonstrate how a novel synthe-

sis of three methods—(a) unsupervised topic modeling of

text data to generate new measures of textual variance,

(b) sentiment analysis of text data, and (c) supervised ML

coding of facial images with a cutting-edge convolutional

neural network algorithm—can shed light on questions

related to CEO oral communication. With videos and

corresponding transcripts of interviews with emerging

market CEOs, we use this synthesis of methods to dis-

cover five distinct communication styles that incorporate

both verbal and nonverbal aspects of communication. Our

data comprises interviews that represent unedited expres-

sions and content, making them especially suitable as data

sources for the measurement of an individual's communi-

cation style. We then perform a proof-of-concept analysis,

correlating CEO communication styles to M&A outcomes,

highlighting the value of combining text and videographic

data to define styles. We also discuss the benefits of using

our methods versus current research methods.

Managerial Summary:CEOs spend most of their time

communicating to investors, customers, and partners with

the aim of influencing these various stakeholders. To what

extent though does their effectiveness as leaders depend

on a mixture of what they say and how they say it? We

use cutting-edge machine learning approaches to measure

a CEO's communication style, which can give clues about

Received: 25 July 2018 Revised: 21 May 2019 Accepted: 25 May 2019 Published on: 6 August 2019

DOI: 10.1002/smj.3067

Strat Mgmt J. 2019;40:1705–1732. wileyonlinelibrary.com/journal/smj © 2019 John Wiley & Sons, Ltd. 1705

the major strategic decisions a CEO's firm must make.

With a collection of video interviews with 61 organiza-

tional leaders from emerging markets, we use textual anal-

ysis and facial image expression recognition to code

whether CEOs are “excitable,”“stern,”“dramatic,”

“rambling,”and “melancholy”in their communication

styles. As a proof-of-concept, we also show that CEOs

who were more dramatic in expressing themselves were

also less likely to oversee major acquisitions. Therefore,

not only can CEO communication styles help predict a

firm's ability to grow, adapt to change, and reallocate exis-

ting assets, styles can also be coded more intuitively by

using our new method, representing a vast improvement

over previous methods in both accessibility and

interpretability.

KEYWORDS

CEO oral communication, image analysis, machine learning,

managerial cognitive capability, topic modeling

1|INTRODUCTION

With the advent of empirical techniques based on machine learning (ML), research in social sciences

is arguably at an inflection point (Athey, 2019). Recent papers in economics—such as Mullainathan

and Spiess (2017) and Kleinberg, Lakkaraju, Leskovec, Ludwig, and Mullainathan (2017)—have

demonstrated the usefulness of empirical predictive techniques that build on ML concepts. Machine

learning techniques have been shown to be particularly helpful in analyzing new sources of “big

data”that previously have been underutilized for research, such as large textual archives

(Antweiler & Frank, 2004) and repositories of images (Glaeser, Kominers, Luca, & Naik, 2018).

More broadly, research across several fields within management has started to embrace big data and

text/image mining tools (e.g., Arts, Cassiman, & Gomez, 2018; Kaplan & Vakili, 2015; Menon,

Tabakovic, & Lee, 2018; Riedl et al., 2016). In this paper, we detail a novel synthesis of ML

methods for coding textual data and facial expressions to shine light on CEO oral communication. In

doing so, we attempt to advance the study of CEO oral communication by: (a) synthesizin g multiple

methods and data related to both verbal and nonverbal aspects of communication to generate mea-

sures for a CEO's communication style, and (b) demonstrating the benefit of using state-of-the-art

methods (e.g., a convolutional neural network method to code facial images) vis-à-vis methods cur-

rently being used in the literature (e.g., videometrics driven by human coding).

We study CEO oral communication in response to the call made by Helfat and Peteraf (2015) to

study verbal language and nonverbal communication, which they highlight as important components

of managerial cognitive capabilities. Communicating well is one of the most important skills in the

CEO toolkit. As Bandiera, Guiso, Prat, and Sadun (2018) argue, CEOs need to create organizational

1706 CHOUDHURY ET AL.