Fall Research Expo 2024

Making the Most of Large Datasets: Meaningful Exploratory Analysis in Observational Studies

Reflecting on this research project has highlighted the complexities of studying the long-term effects of parental alcoholism, as well as allowing me to appreciate the growth that I have experienced as a researcher. Working with the Wisconsin Longitudinal Study (WLS) dataset, which tracks the lives of 1957 Wisconsin high school graduates, has provided a unique opportunity to explore how parental alcoholism impacts the adult lives of their children. The broad scope of our research question required a focused approach, which led us to concentrate on five key areas: economic outcomes, physical health, mental health, substance use, and interpersonal relationships. 

During the project, the team made the decision to leverage the methodological approach of splitting the dataset into two subsets, which would be analyzed separately. This enhancement presented a challenge that required careful planning. It underscored the importance of rigorous statistical methods, especially in observational studies where controlling for bias is crucial. Working on this project has sharpened my understanding of observational studies and the nuances of interpreting data in which controlled experiments are not possible. By maintaining a familywise error rate of 0.025, we designed the project to ensure the reliability of our findings. 

Our deep analysis of the WLS dataset also highlighted its limitations, specifically its demographic homogeneity. The dataset consists primarily of white, middle-class individuals, limiting our findings' generalizability. Acknowledging this limitation has been a critical learning point, emphasizing the need to consider the broader context of any dataset used in research. Understanding these limitations has made me more critical and reflective as a researcher, particularly in how findings are interpreted and applied.

PRESENTED BY
PURM - Penn Undergraduate Research Mentoring Program
Wharton 2027
Advised By
Dylan Small
Professor of Statistics and Data Science
William Bekerman
PhD Student, Statistics and Data Science
PRESENTED BY
PURM - Penn Undergraduate Research Mentoring Program
Wharton 2027
Advised By
Dylan Small
Professor of Statistics and Data Science
William Bekerman
PhD Student, Statistics and Data Science

Comments