Data and code for Uncertainty Displays Using Quantile Dotplots or CDFs Improve Transit Decision-Making (CHI 2017)

Michael Fernandes (mfern@uw.edu)
Logan Walls (logan.w.gm@gmail.com)
Sean Munson (smunson@uw.edu)
Jessica Hullman (jhullman@uw.edu)
Matthew Kay (mjskay@umich.edu)

This repository contains data and analysis code for the following paper:

Michael Fernandes, Logan Walls, Sean Munson, Jessica Hullman, and Matthew Kay. "Uncertainty Displays Using Quantile Dotplots or CDFs Improve Transit Decision-Making", Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems - CHI 2018. DOI: 10.1145/3173574.3173718

Final analysis

The final analysis is outlined in final_analysis.md (also available in html). This analysis was compiled from an RMarkdown notebook, final_analysis.Rmd.

Additional materials from the final analysis are available here:

• data/final_trials.csv: data from the final study

• models/final_model.rds: fitted model object from final Beta regression. This is a brmsfit R object.

• experiment_screen_shots.pdf: Screenshots of the major end points encountered by subjects in our online experiment. Includes an example tutorial that walks a subject through the important details of how to use one of the uncertaintiy visualizations (probability density plots).

Pre-registration

Our pre-registered analysis plan is in pre-registration.pdf. It is also available for verification on AsPredicted.

Pilot analyses

Some (rougher) pilot analysis notebooks are also included in this repository:

• valuation_analysis.pdf: Valuation analysis that helped us set payoffs.

• pilot_exploration.ipynb / pilot_exploration.html: An iPython notebook with some preliminary analysis of pilot data.