Prior Setting In Practice: Strategies and rationales used inchoosing prior distributions for Bayesian analysis
This github repository contains supplementary materials for the CHI 2020 submission, Prior Setting In Practice: Strategies and rationales used inchoosing prior distributions for Bayesian analysis
This repository is intended for use by reviewers of the paper only.
The repository is structured as follows:
-
The directory labelled Survey contains materials used in the survey. The survey had three separate conditions, the materials for them can be found in separate directories within the Survey directory:
- Condition 1 - Parameter scale density vis contain the first two pages for the
Parameter scale density visualization
condition. They survey can be viewed online here. - Condition 2 - Response scale density vis contain the first two pages for the
Response scale density visualization
condition. They survey can be viewed online here. - Condition 3 - Prior predictive density vis contain the first two pages for the
Prior predictive density visualization
condition. They survey can be viewed online here. - The page 3 of the survey, which asks participants some open-ended questions can also be found within the Survey directory.
- Condition 1 - Parameter scale density vis contain the first two pages for the
-
The directory labelled data analysis contains the two R scripts detailing our analysis of the data, and our calculation of predictive probability mass (depcited in Figures 4 and 5). It also contains the deidentified responses to our survey.
-
The directory labeled user-study-html contains the materials used during the follow-up interviews. Although this was different for each interviewee, an example can be found here. This document is for a participant who was in Condition 2 - Response scale density vis in the survey.
-
The directory labelled Transcripts contain the transcripts from the follow-up interviews.
-
The semi-structured interview protocol we used for our follow up interviews can be found in the root directory.
If you have any questions regarding this study or our paper, please contact Abhraneel at abhraneel@u.northwestern.edu