Welcome to major work period 3 of the deliberation project, known to outsiders as Spring Term 2023! This is the top-level planning issue. 🙂

Team

@kepstein23 @Alan-Qiao @sky1113 @yiwei-ellen @JamesPHoughton

Schedule

Jan 9 - April 30 March 4-12: Spring Break

We will do one-week ~~sprints~~ iterations to keep up our momentum, and keep in better contact with each other. https://github.com/orgs/Watts-Lab/projects/12/settings/fields/2649143
We will meet twice per week. We will plan tasks for the week on Tuesdays at 9am on Zoom, and Thursdays 9 am in person (with food) to do some work together and get past any bugs or issues we need to address before the weekend. We have the conference room reserved from 8:30-12, so feel free to come and make a full working session of the morning.
We will try to run weekly pilots, nominally Monday afternoons. PRs will be due Sunday night, so there is time to merge and test before the pilot.

Goals

1. Demonstrate experiment design-space mapping

The premise of our high-throughput approach to studying deliberation (and our other high-throughput projects) is that (1) the effect of different interventions will vary with context and (2) this context can be formalized and effect sizes mapped. However, we haven't yet demonstrated that we can cover an adaptive experiment design space to allow us to show this context dependency. This goal is to establish a proof-of-concept using a small number of dimensions, that we can then build on to create a higher-dimensional map.

Completing this goal fulfills the objectives of our Phase 1 funding proposal.

To accomplish this goal, we need basic versions of all of the components of the adaptive experimentation loop to be in place. These components will of course be continually improved, and so our emphasis will be on setting up a system with a strong foundation of CI/CD and review processes that will make continual development possible.

2. Listening intervention study with Netta

In this study, we are testing a number of different interventions designed to elicit listening behavior from participants. This collaboration helps us to:

expand the range of experiments that our system is set up to do
work through issues of multi-institution data sharing and ethics approval
improve the relevance of the surveys
to build out our infrastructure for human and automated video coding

3. Discussion structuring (policing) study with Dean, Will, and co.

In this study, we are exploring interventions that structure the discussion by providing intermediate prompts for participants to discuss, and giving them instructions about who should be talking at different times. This collaboration helps us to:

expand the set of interventions we are able to support
establish patterns for data collection and integration between teams

Tasks

The red-shaded components of the architecture diagram below indicate the pieces to be created this term. The blue-shaded components are substantially in place as of Jan 1, 2023.

Major Project Areas

This term we will try to balance the cohort model that we used successfully over the summer with giving each student responsibility for a particular component or research arc that has a coherent set of tasks over the course of the term. My hope is that each RA will take primary responsibility for a particular section and seek to develop skills in that area, but that the team would continue to support one another's work in a collaborative fashion.

Watts-Lab / deliberation-empirica

🌯 Spring Term 2023 #380