COMEDY, CONTROVERSY, AND COMPUTERS - Evaluating the Validity and Usability of Machine Learning Classifiers

Aug 5, 2026

Entertainment media distribution has evolved alongside new research demands for methods that preserve validity while handling the scale and individuality of modern systems. Recent work highlights a persistent need to strengthen scholars’ toolkits to close this gap. Although political communication and journalism have adopted machine-learning approaches, entertainment research has not integrated them to the same extent due to concerns about accessibility and validity. This thesis evaluates NVivo’s experimental machine-learning implementation, testing precision and reliability, and examines training-set construction for future adoption.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages72

Reading Time3 h 1 min

AccessFree to download

CHAPTER ONE: INTRODUCTION

CHAPTER TWO: LITERATURE REVIEW

Old Theories; New Questions

Algorithmic Culture

Constrained Methods

Political Social Identity & Entertainment

Modern Family

Last Man Standing

The Middle

CHAPTER THREE: METHODS & MEASURES

Defining the Corpus

Conducting the Analyses

Human Coding

Automated Coding

Training Set Sampling

Horizontal Training

Vertical Training

Precision

COMEDY, CONTROVERSY, AND COMPUTERS: EVALUATING THE VALIDITY AND USABILITY OF MACHINE LEARNING CLASSIFIERS FOR ENTERTAINMENT MEDIA By GEOFFREY D. THATCHER A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF ARTS IN COMMUNICATION WASHINGTON STATE UNIVERSITY Edward R. Murrow College of Communication JULY 2023 © Copyright by GEOFFREY D. THATCHER, 2023 All Rights Reserved © Copyright by GEOFFREY D. THATCHER, 2023 All Rights Reserved To the Faculty of Washington State University: The members of the Committee appointed to examine the thesis of GEOFFREY D. THATCHER find it satisfactory and recommend that it be accepted. | Jordan Foley, Ph.D., Chair | | --- | | Stacey Hust, Ph.D. | | Alexis Tan, Ph.D. | Shannon C. McGregor, Ph.D. ACKNOWLEDGMENT I want to first and foremost thank my coders, who without their help I would not have been able to produce this thesis. Magda, Kai, and Andrew you were saviors and more than I could have ever hoped for as coding companions. Despite the challenges from time constraints and my illness, you all worked tirelessly to get the work done and done well. I would also like to thank Jordan Foley, my advisor. Without him I would be lost, still chasing my ambitions of doing everything all at once. His guidance has been indispensable in helping craft my thoughts into a contribution. Shannon McGregor, one of my committee members, has been mentoring me since our time together at the University of Utah. Despite moving across the country, I am so happy she agreed to join this journey and continue to support me through my academic career. My other two committee members, Alex Tan & Stacey Hust have been a source of wisdom and guidance. Several points during this project I felt stalled but long meetings with both individuals helped push the project forward. Their expertise is so deep I feel I have only been exposed to a fraction of their knowledge and wisdom. Finally, it truly did take a village. My friends and family scattered across the globe from the Swiss alps to the Boise bench have been there to assist me any way they could. Without their unyielding support I could not accomplished anything attributed to me to date. COMEDY, CONTROVERSY, AND COMPUTERS: EVALUATING THE VALIDITY AND USABILITY OF MACHINE LEARNING CLASSIFIERS FOR ENTERTAINMENT MEDIA Abstract by Geoffrey D. Thatcher, M.A. Washington State University July 2023 Chair: Jordan Foley As technology has changed the way in which entertainment media is distributed, researchers have had to adapt their methods to retain validity as well as match the new scale of these complex and individualized systems. Limitations sections of recent research continually cite the need to improve scholars' toolkits to address this growing gap. While scholars in political communication and journalism have used these techniques, scholars within entertainment have, comparatively, not adopted these methods to the same extent. Some scholars cite concerns that adoption of technology ignores potential problems within accessibility and validity. This thesis tests NVivo’s “experimental” implementation of machine learning; evaluating its precision and reliability. Results show potential for future adoption and validation of NVivo’s auto coder feature and gives insight on training set construction. TABLE OF CONTENTS Page ACKNOWLEDGMENT ................................................................................................................iii ABSTRACT ................................................................................................................................... iv LIST OF TABLES ........................................................................................................................vii CHAPTER CHAPTER ONE: INTRODUCTION ................................................................................. 1 CHAPTER TWO: LITERATURE REVIEW .....

FAQs

What problem does the thesis address in entertainment media research?

It addresses the need for methods that retain validity while matching the scale and complexity of modern entertainment-media systems and tools.

How does the thesis evaluate NVivo’s machine learning implementation?

It tests NVivo’s “experimental” machine-learning implementation by evaluating the precision and reliability of its auto-coder feature.

What do the results suggest for future use of NVivo’s auto coding?

They indicate potential for future adoption and validation of NVivo’s auto coder feature and provide insight into how training sets should be constructed.

Aug 5, 2026

Upload by Olivia Brown