Tongyi Jin
Thesis: Strategies for Spotting and Removing Bias in Financial News
@ AF FinTech Lab
[publication]
May 2019 - June 2020
Background and Goals
Inspired by my backgroup in finance and econlmics, I focued on the behavioral aspects of business and utilized insights from the field of psychology. In the context of Amazon's purchase of the Washington Post in October 2013 and media attention was the divorce of new owner with his former wife in 2019, I team up with two strategy consultant peers to examine how biased WP became in reporting Amazon related news after acuiqistionm. We also studied how CEOs use media ownership to influence their reciprocal relationships with financial journalists to get favorable forecasts.
This is an experimental research project I led at AF Tech Lab. The high level objective of the study is to understand stakeholser's intention of media acquisition as well as financial journists' motivation to report biased news, thus to inform the influence of misinformation to news audiences.
General Research Goals
-
Understand reporting difererence on tech-corporation-related events when a media ownership changed across targeted media platforms
-
Understand how the within differences of media ownership influenced the media bias
User-driven Research Goals
-
Understand how users (or audiences) are influenced by media bias in terms of the sentiment, length, and publication date of the articles
-
Undeerstand whether users' investment decision or performance are influenced by owership-related media bias
How might we help audiences understand their own investment decision by engaging user insights and experience on media ownership change and media manipulation?
Method
To achieve these goals, I was inspiared by research about sentiment anaylsis conducted by Google Research Team applying tools developed by Google Cloud AI & Machine Learning, we extract all needed information in the table format, containing the headline, author, publication date, length (words/sentences), first sentence, sentiment score, and magnitude. We collected all 613 articles, published by Washington Post and the New York Times, which mentioned Amazon 12 months before and after the acquisition. Then, we tracked their changes in sentiment, length, and time of publication and identified some data patterns. Additionally, we employed Python to analyze user's reaction to the media ownership change through their investment strategies data.
Figure 1. List of events reported by both WP and NYT
I converted 1226 headers and articles by hand but knew it needed to be less time consuming
Figure 2. Coding of integrated articles
Too Much Data!!!
Convert sentiment scores to binary
The score represents the individual score for a word within a text. The sentiment score ranges from -1.0 (negative) to +1.0 (positive), showing the overall emotional leaning of the text.
Magnitude is the sum of without considering signs, ranging from 0.0 to +infinity. Thus, if the score is "0.0" it can mean two things. (1) If the magnitude is close to 0, the text is neutral; (2) If the magnitude is large, then the text is mixed (positive and negative scores cancelled each other out). As such, a score closer to "+1.0" with high magnitude means that text is "clearly positive" and a score closer to "-1.0" with high magnitude means that text is "clearly negative".
My Learnings
-
Amazon's acquisition has posed a certain extent of influence on the publication time of the articles. Our findings highlight that ownership change limits the media's ability to fulfill its role as a watchdog without interference.
-
Motivation for psychology research:
-
The capstone project required me to investigate the literature that discusses the influence of media bias on the psychological and linguistic aspects of decision-making. This inspired me to pursue ways to research psychology outside of the business. At the end of 2019, I spent more than 600 hours learning theory and statistical analysis tools through online courses and then sent out over 30 emails to researchers with similar interests in Hong Kong. Thanks to this preparation, I obtained research opportunities in researching user experience of youth programs and broke into the world of psychology and UX research.
-
While transferring from market research, I realized that business questions lead to research questions, and research answers also guide business decisions and strategies. Business thinking would help me build up my research with a further understanding of stakeholders’ intentions in the business setting.
-
-
How to lead effective communication:
-
​Using a effective medium of communication: I was better at written communication with less edge than speech. Nevertheless, some message better delivered in written, while others in oral. My team consists of three girls from Hong Kong, mainland China and Kazakhstan. We usually use English to communicate with each other, though sometimes misunderstandings and miscommunication happened due to different native languages and cultural backgrounds. After several attempt to stay online at the beginning of pandemic, we realized the effectiveness of in-person meeting and eventually work together during a tensive working period. It turns out the close and open relationship between team member enables me to develop excellent communication skills and transfer untestable information with different audiences such as coordinators, advisors, and groupmates.
-
Importance of leadership: Different leadership styles also influence the communication process. At the beginning of the project, I failed to connect with teammates very well because of the lack of a clear set of values and plan in encouraging others working together (refer an example presenting the ineffective leadership style).
-
​
​
​
​
​
​
​
​
​
​
​
-
The development of social media undoubtedly has expanded the frame and functions of massmedia. Inspired by Amazon’s acquisition are also available for future collaborationsbetween enterprises and social media. With appropriate and flexible strategies, corporations can make good use of media as their communication mediator with their investors and consumers.
-
Though we utilizing qual+quant, did not look into users' insights on sentiment change
-
How might we help audiences understand their own investment decision by engaging user insights and experience on media ownership change and media manipulation?
For example, I prefer to work overnight and complete all the work together with little communication with others. The next day, when I woke up and surprisingly found people did not appreciate the way, I was aware that teammates are likely feeling confused or even unhappy for not interactively involved in the developing strategy in this circumstance. The consequence of playing behavioural leadership during communication requires my team to take adequate consideration to make interaction a successful one