Studying User Behavior in Digital Platforms Restricted; Files Only
Mousavi, Seyyedehnasim (Summer 2023)
Published
Abstract
Studying user behavior in digital platforms holds significant importance in today’s
interconnected world. As our lives increasingly revolve around digital interactions,
understanding how users engage, react, and make decisions online becomes crucial.
By analyzing user behavior, we gain valuable insights into their preferences, motivations,
and decision-making processes, which can inform the design and optimization
of digital platforms. This knowledge empowers businesses to enhance user experiences,
tailor personalized content, and improve the effectiveness of their digital transformation
strategies. Additionally, studying user behavior helps identify patterns,
trends, and emerging behaviors, enabling researchers and practitioners to anticipate
and adapt to evolving user needs. Ultimately, a deep understanding of user behavior
in digital platforms fosters innovation, drives user satisfaction, and enables organizations
to thrive in the dynamic digital landscape.
This dissertation focused on studying user behavior within two specific digital platforms:
recommendation systems and online learning. First, I explore the decoy effect
in recommendation systems and how adding a decoy in a set of recommended items
impacts the systems’ effectiveness. I specifically differentiate between personalized
and non-personalized recommendations and show that including a decoy minimizes
the demand for the target option when personalized recommendations are presented,
which deviates from the traditional decoy effect. However, a decoy increases the
target’s demand when non-personalized recommendations are shown, following the
conventional decoy effect. Then, I investigate the interplay between personalized
recommendation systems and sponsored content to explore how users behave toward
sponsored content in a set of personalized recommendations. The results of this study
indicate that users react negatively to sponsored content, even when it aligns with
their preferences. Surprisingly, the adverse reaction is even stronger when sponsored
content is shown to be a high match for the users’ preferences. In the last chapter, I
examine user behavior in a Massive Open Online Course (MOOC) and investigate the
impact of interpersonal relationships on user behavior. My findings demonstrate that
different types of discussions could significantly impact user engagement and performance.
Specifically, I show that both off-topic and on-topic discussions can benefit
users in online learning platforms by strengthening their social identity. However, the
effectiveness of each discussion type varies depending on the user’s characteristics.
Table of Contents
Introduction
1.1 Research Problem and Context
1.2 Overview of the Three Essays
1.2.1 Overview of Essay1: The Decoy Effect and Recommendation
Systems
1.2.2 Overview of Essay2: Sponsored Content and Personalization
Systems
1.2.3 Overview of Essay3: Discussion Types and User Behavior in
Online Learning Platforms
2 The Decoy Effect and Recommendation Systems
2.1 Introduction
2.1.1 Research Motivation
2.1.2 Research Questions and Contributions
2.2 Literature Review
2.2.1 Decoy Effect
2.2.2 Contextual Factors in Recommendation Systems
2.3 Theoretical Framework and Developed Hypotheses
2.4 Experiment Design and Procedure
2.4.1 Online Platform and Recommendation System Ratings Seed
2.4.2 Experimental Design and Procedure
2.5 Analysis and Results
2.5.1 Manipulation Check
2.5.2 Model-Free Evidence
2.5.3 Empirical Analysis
2.5.4 The Decoy Effect on the Target Item
2.5.5 The Decoy Effect on the No-Choice Option
2.6 The Relative Changes of All Options
2.7 Underlying Mechanism Test
2.8 Replication Study
2.9 Robustness Check
2.9.1 Between-Subject Analysis
2.9.2 Role of Item Quality
2.9.3 The Saliency of the Decoy
2.10 Discussion
2.11 Conclusion
3 Personalization Systems and Sponsored Content
3.1 Introduction
3.2 Related Literature
3.3 Theoretical Framework and Research Questions
3.4 Experimental Design and Procedure
3.5 Analysis and Results
3.5.1 Manipulation Check
3.5.2 Model-Free Evidence
3.5.3 Empirical Analysis
3.5.4 Robustness Check and Further Exploration of Prediction Scores
3.6 Underlying Mechanism
3.7 Discussion and Conclusion
4 Should Online Learning Platforms Facilitate Off-Topic Discussions?
Randomized Field Experiment on a Massive Open Online Course
4.1 Introduction
4.1.1 Research Motivation
4.1.2 Research Questions and Contributions
4.2 Theory and Background
4.2.1 Shared Social Identity
4.2.2 User Interactions in Learning Environments
4.3 Experimental Setup and Procedure
4.4 Empirical Model and Initial Analysis
4.4.1 Intention to Treat
4.4.2 Causal Treatment Effect
4.5 Empirical Results
4.5.1 Main Effect
4.5.2 Testing the Underlying Mechanism
4.5.3 Heterogeneous Effects
4.6 Robustness Check
4.7 Post-Hoc Analysis
4.7.1 Discussion Types
4.8 Conclusion & Discussion
5 Conclusion
Appendix A The Decoy Effect and Recommendation Systems 116
Appendix B Personalization Systems and Sponsored Content 127
Appendix C Should Online Learning Platforms Facilitate Off-Topic
Discussions? Randomized Field Experiment on a Massive Open Online
Course 134
About this Dissertation
School | |
---|---|
Department | |
Subfield / Discipline | |
Degree | |
Submission | |
Language |
|
Research Field | |
Palabra Clave | |
Committee Chair / Thesis Advisor | |
Committee Members |
![Preview image embargoed](/assets/No_Preview-c99a369db0c8f31173bf3fdc06f581cabae85282e8a747dd66f99b731b54aec0.png)
Primary PDF
Thumbnail | Title | Date Uploaded | Actions |
---|---|---|---|
![]() |
File download under embargo until 16 August 2029 | 2023-06-21 15:00:33 -0400 | File download under embargo until 16 August 2029 |
Supplemental Files
Thumbnail | Title | Date Uploaded | Actions |
---|