The study compared three methods (Ridge, Elastic Net, and Lasso Regression) to see which one is best at predicting outcomes when there are multiple factors that are related to each other. They used computer simulations to create different scenarios and found that Elastic Net is the most accurate method when there is low, moderate, or high multicollinearity. However, when there is severe multicollinearity in the data with less than 10,000 observations, Lasso Regression is the most accurate.