How Data Analytics Can Be Misleading

This workshop will provide the participants with a macro view on evaluating the quality of data analytics from the perspective of context, method, and validity. Rather than just demonstrating the power of data analytics, this workshop offers participants a unique perspective on how data analytics can be erroneously conducted. With a series of intriguing examples, participants will experience how easily data analytics can deliver inconclusive and even misleading findings. Those examples include:

• Common stereotypes, such as the idea that students with excellent academic performance are not good at athletic sports, and that leaders of organizations are less intelligent than their subordinates.
• Controversial debates, such as if females are discriminated against in university admission, and if the death penalty could curb crime.
• Technical discussions, such as when adding more data to your model could be harmful, and when applying theoretical predictions from one context to another context can be problematic.

Those examples come with both R and Python codes for the participants to simulate the data generation process and “see” how and why subsequent data analytics can be wrong.

In sum, the workshop aims to help participants to establish a system to evaluate data analytics procedures and outcomes critically, and in turn, make data analytics a force for good.

• Clarify the connection between data analytics and coding
  
• Appreciate the importance of obtaining contextual knowledge;
  
• Identify common threats to the validity of data analytics;
  
• Understand the limits of analytics methods (using regression as an example)

• Beginners who are looking for a comprehensive view of valid data analytics practices
  
• “Consumers” of business analytics who need to use business analytics outcomes in their decision-making.
  

• Discuss the essence of data analytics and coding
  
• Understand how data analytics can be misleading
  
• Learn to simulate the data generation process to illustrate misleading data analytics outcomes
  
• Establish a system to evaluate data analytics procedures and results critically
  

Zhen “Richard” Tang is originally from Huaiyuan, a beautiful one-million-population small town in eastern China. Though Richard learned BASIC programming on his own and wanted to be a software engineer when he was in high school, he graduated from East China University of Science and Technology with B.A. and M.S. in Business Administration. Richard then earned his Ph.D. in Marketing with a minor in economics from the University of Arizona. He is now serving as an assistant professor of marketing at Loyola Marymount University (LMU). At LMU, Richard teaches marketing analytics, natural language processing, and mentors students in various data science competitions.

Richard’s training on quantitative research methods consists of econometrics and machine learning (with a focus on natural language processing). He is interested in applying those quantitative methods to generate constructive insights for businesses and society. Topics of his current research include quantifying business environments with geographical location information, extracting consumer insights from user-generated-content, assessing the effectiveness of AI-based service robots, and redesigning organizational structure to unleash the power of business analytics.

Note

This is one training session of 2021 IM DATA Annual Conference held by RMDS Lab. You can find out more about other fascinating training sessions at https://grmds.org/im-data-2021

Event Venue

Online