In this paper we propose an ideal point model to classify the decision direction (conservative vs. liberal) of the outcome of Supreme Court cases. We test various models for this classification problem, finding that a random forest model trained on our dataset of opinions plus facts about the case, performs the best with an F1 score of 0.7. We also test these same models on just the facts dataset and just the opinions dataset, which gives us lower F1 scores. We outline possible reasons for this discrepancy, as well as areas for further research based on these results.
https://docs.google.com/document/d/1-hrJJ6_CvkysV75cim2et-YyqAVRhq6pAq6X2BX_esg/edit?usp=sharing
https://drive.google.com/file/d/1eplfMSEovVOaoKDFrm97j9dfhpEpmBIU/view?usp=sharing