Project Does That Affect Everyone The Same? Determining How Universal Experimental Treatment Effects Are Program Margo Katherine Wilke Undergraduate Research Internship Program Term Fall 2023 Status Accepted Research Area Sociology; Experimental Methods; Statistics Description Experimental methods are seen as the gold standard for answering causal questions. For example, to see if a new medicine works a group of people are divided into two groups: control and experimental. The two groups' outcomes at the end of the study (e.g., are they still sick?) are then compared to determine whether the treatment had an effect. However, these types "average treatment effect" analyses simply look at averages across each group -- not at outcomes for individual people. For example, even if many people benefitted from the treatment -- some probably did not. From an individual standpoint, it is important to know not just whether the treatment worked for some people -- but whether it will be effective for you. In this study, we (Dr. Mize and Dr. Scott Feld) propose a new method for determining "heterogeneous treatment effects" or how a given treatment will affect different people differently. To do so, we plan to gather previously collected public data from experiments and then re-analyze the data using our new method. Supervisor Trenton D Mize Mentor Type of work required The student would need to help find, download, document, and clean the public data (which has been archived here: http://www.tessexperiments.org/paststudies). The student would then need to recreate the key analyses done in the past studies; i.e., the traditional way of analyzing experimental data. Then, we would re-analyze the data using our new method. Websites https://www.trentonmize.com/about Qualifications Experience with Stata is ideal (experience with R is an acceptable alternative). The student would need to be able to download and "clean" the data -- including manipulating and labeling variables for analysis. The student would then need to be able to recreate basic experimental analyses (e.g., regression analyses). Experience with applied statistics and statistical software is needed. Credit Hours 3 Weekly Hours 6 (estimated)