Oliver Hauser: Research Paper Competition Winner | 2018 Wharton People Analytics Conference

Summary

Oliver Houser from Harvard University discusses a collaborative research project with Harvard, New Mexico, and Deloitte, focusing on using data science and behavioral analytics to predict and reduce unethical behavior. The study employs machine learning to identify individuals at high risk of fraud and then tests various behavioral interventions to deter such actions. The presentation highlights the importance of combining predictive analytics with causal field experiments for effective intervention strategies, suggesting potential applications in public policy, health, and education.

Takeaways

• 🏆 The speaker, Oliver Houser, expresses gratitude for the award and acknowledges the collaborative effort with Harvard University, New Mexico, and Deloitte in conducting research on reducing unethical behavior.
• 🔍 The research combines data science with field experiments to predict and reduce unethical behavior, instead of the traditional reactive approach of dealing with issues after they occur.
• 🎥 A reference to the movie 'Minority Report' is used to illustrate the idea of predicting criminal behavior, but the speaker clarifies that their approach is ethically different and focuses on prevention.
• 🤖 Machine learning algorithms are utilized to identify individuals at high risk of unethical behavior, such as fraud, by analyzing various features including job history and online behavior.
• 🧪 A randomized control trial, or field experiment, is conducted to test different interventions and determine what methods are effective in reducing unethical behavior among the high-risk group.
• 📊 The study focuses on the unemployment population in New Mexico, using the context of unemployment benefits and the requirement to report work status honestly to study fraud.
• 📈 The machine learning model assigns a risk score to individuals, which is then compared with actual fraud rates to validate the model's accuracy.
• 📝 Behavioral insights from textbooks and literature are used to inform the design of messages aimed at reducing unethical behavior, such as reminders and social norm messages.
• 📈 The field experiment showed that certain messages, particularly those leveraging social norms, significantly increased the likelihood of individuals reporting work honestly.
• 🔄 The speaker suggests a two-step approach: first, using predictive analytics to identify at-risk individuals, and second, applying causal field experiments to tailor interventions for maximum impact.

Q & A

What is the main focus of Oliver Houser's research presentation?

Oliver Houser's research presentation focuses on combining data science and people analytics tools with field experiments and causal evidence to reduce unethical behavior in organizations.

Why is the traditional approach of reacting to unethical behavior not ideal according to the presentation?

The traditional approach of waiting until unethical behavior occurs before acting is not ideal because it is reactive rather than proactive. It does not prevent unethical behavior and can be inefficient and costly.

What is the concept of 'predictive analytics' mentioned in the script?

Predictive analytics refers to the use of data, statistical algorithms, and machine learning techniques to identify the likelihood of future outcomes based on historical data.

How does the Minority Report movie relate to the discussion on predicting unethical behavior?

The Minority Report movie is used as a thought piece to illustrate the concept of predicting future criminal behavior. However, the presentation clarifies that they do not intend to use such predictive methods due to ethical concerns and the potential for false positives.

What is the significance of collaborating with researchers from Harvard University, New Mexico, and Deloitte in this research?

Collaborating with researchers from these institutions and organizations provides a multidisciplinary approach, combining expertise in data science, behavioral science, and practical knowledge from industry partners, which is crucial for the comprehensive study of unethical behavior.

What is the two-step methodology used in the research to address unethical behavior?

The two-step methodology involves first using machine learning to identify individuals likely to behave unethically, and then conducting a field experiment to test interventions aimed at reducing such behavior.

What is the role of behavioral science in this research?

Behavioral science plays a role by providing insights and strategies that can be tested through field experiments to determine the most effective ways to deter unethical behavior.

How does the unemployment population in New Mexico serve as a context for the field experiment?

The unemployment population in New Mexico serves as a context because they are recipients of unemployment benefits and are required to report their work status, which can be an area prone to fraudulent behavior if they fail to report earned income.

What are some of the behavioral insights used in the field experiment to reduce unethical behavior?

Behavioral insights used in the field experiment include reducing ambiguity through clear messaging, using social norm messages to encourage honesty, and testing various interventions to see which ones are most effective.

What was the outcome of the field experiment in terms of reducing unethical behavior?

The field experiment showed a significant increase in the disclosure of work status among high-risk claimants, indicating a reduction in unethical behavior due to the tested interventions.

How does the presentation suggest combining predictive analytics with causal field experiments?

The presentation suggests that predictive analytics should be used to identify potential unethical behavior, and then causal field experiments should be conducted to test and implement interventions that effectively deter such behavior.