sfbrigade/CA_DOJ_OpenJustice

Name: CA_DOJ_OpenJustice

Owner: SFBrigade

Description: Statistical hypothesis tests and predictive modeling for the CA DOJ's OpenJustice project.

Created: 2016-08-20 17:45:32.0

Updated: 2017-11-30 04:34:23.0

Pushed: 2017-02-23 04:47:33.0

Homepage: null

Size: 1820

Language: Jupyter Notebook

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

CA Dept. of Justice OpenJustice Project

Modeling & Hypothesis Testing

Below, members of the Data Science Working Group have been charged with answering, via inferential statistics, some of the California Department of Justice's inquiries around criminal justice. These more pointed inquiries were inspired by the OpenJustice project's exploratory analyses at OpenJustice.org.

Responsible DSWG Members:

• Catherine Zhang
• Matthew Mollison, Ph.D.
• John Huynh
• Saniya Jesupaul
• Brian Smith
• Biljana Rolih, Ph.D.
• Jude Calvillo
• Alex Novet
• Nitya Subramanian

Status, as of January 25, 2017:

• We were recently informed of some very interesting data coming down the pipeline (gun sales and justice department transactions, post-arrest), and the team has already formulated some great hypotheses to test upon that data.
• We will be having another call with OpenJustice's head of research, Eric Giannella, Ph.D., in the coming weeks, to see how realistic it would be to execute our latest batch of hypothesis tests, given state regulations/policy, as well as to get his insights into the sociological issues involved.
• We're continuing to move toward integrating some of our findings into the DoJ's data stories and getting additional statistical -and visualization- prompts from the DoJ team.
• Prompt #1 pretty much complete (Anonymous Analyst) »
• Prompt #2 hypothesis tests being extended to all years and top BC codes (Biljana Rolih, Ph.D.)
  
  • Initial exploration and hypothesis tests (Catherine Zhang) »
    
• Prompt #3 in progress but with important questions (Matt Mollison) »
• Analysis/support of Prompt #5 has been deprecated, as it doesn't seem we'll be able to improve upon the state of the art.
  • However, if one is interested in our initial modeling attempts, by John Huynh and Saniya Jesupaul, they can see those here »

Past Prompts (new prompts coming)

1. Which counties/agencies arrest African American juveniles at a statistically significantly higher rate than that of other counties/agencies?

   • Extending analysis to each ethnic group represented
   • Drilling down to felonies vs. misdemeanors

2. For the same criminal offense, are particular ethnic juvenile groups more likely to be treated with harsher consequences by law enforcement?

3. Statewide, what contextual and ethnic factors best predict the arrest of juveniles for felonies?

4. Statewide, what contextual and ethnic factors best predict the arrest of juveniles for battery, specifically?

5. For resource allocation prompt: ~. predictors to statewide crime rate (i.e. not necessarily optimization; just a first, exploratory step, probably via LM)