The Public Interest Data Lab is intended to provide data science experience to students in service of the public interest. We have three goals:

• Make progress on a project that advances the public interest or social good, in service of a client's goals. And have a project to point to at the end, an example of your work.
• Practice working with real data to answer real questions. This includes finding data, exploring and cleaning data, analyzing and modeling data, and visualizing and communicating data.
• Develop experience working on a data science team, including processes for working collaboratively, openly, inclusively, and reproducibly.

Our client for this spring is the City of Charlottesville's Department of Social Services who wish to understand whether there is racial disproportionality or racially disparate outcomes in local child welfare services, and if so, to what extent and at what stages. There are a number of ways we might investigate this question, depending on the nature of the data the DSS records as well as other data we might use to supplement, merge with, or contextualize the data.

• code contains all of the exploratory data analysis, data cleaning scripts, and final data analysis and visualizations generated for the report.
  • code/dataanalysis contains all of the analysis and visualizations produced for the report.
  • code/dataprep contains finalized cleaning scripts that create useful data object for the analysis scripts.
  • code/exploratory contains exploratory data cleaning and data analysis scripts created by lab members in the early weeks of the class.
• images contains diagrams of git fetch demos.
• pages contains web pages for the project site, including our data confidentiality agreement, code review guidelines, course expectations, a collective problem statement, and the course schedule.
• resources contains materials from class discussions, including our research questions, analysis plan, literature & code review guidelines, a draft data confidentiality agreement, commentary on links in the data, draft problem statements, a discussion of the race variable, and individual weekly updates.

• Referral data: Clients referred to Charlottesville DSS from July 1, 2014 to June 30, 2017 (n=2706). Includes date of referral; age, race, ethnicity, gender, and census tract of referred children; nature of alleged abuse or neglect; relation of individual making the referral to the referred child; whether the case was accepted, the response priority, whether the case was assessed or investigated, the finding of the investigation and finding date. Data in the referral dataframe.

  • Cleaning script in prep_postreferral.R.
  • Analysis in analysis_postreferral.R.
  • Figures and visualizations of disproportionality in figures_postreferral.R.
  • Figures and visualizations of outcome predictions in figures_predictions.R.

• Ongoing client data: Active clients under CPS care between July 1, 2014 and June 30, 2017 (n=375). Includes race, ethnicity, age, and gender of child; date the child’s involvement with CPS began and child’s age at that time; and the number of face-to-face interactions between DSS and the child. Data in the active dataframe.

  • Analysis in analysis_postreferral.R.
  • Cleaning script in prep_postreferral.R.
  • Figures and visualizations of disproportionality in figures_postreferral.R.
  • Figures and visualizations of outcome predictions in figures_predictions.R.

• Foster care data: Foster care clients entering care between July 1, 2014 to June 30, 2017 (n=118). Includes race, age, and gender of child; age child entered custody, date child entered custody, date child existed custody, and reason for exit; child’s current/final placement type and the number of face-to-face interactions between DSS and the child. Data in the fc data frame.

  • Cleaning script in prep_fostercare.R.
  • Analysis in analysis_fostercare.R.
  • Figures and visualizations of disproportionality in figures_fostercare.R.
  • Figures and visualizations of outcome predictions in figures_fostercarepredictions.R.

• Foster care placement history data: Placement history of foster care clients entering care from July 1, 2014 to June 30, 2017. Includes date of entry for each new placement, type of placement, date of exist for each placement, and reason for exit. Data in the fcph data frame.

  • Cleaning script in prep_fostercare.R.
  • Analysis in analysis_fostercare.R.

• Charlottesville demographic data from the 2012-2016 5-year Sex by Age ACS estimates for Cville.

  • Data fetching and cleaning script in prep_basedisp.R.
  • Analysis in analysis_basedisp.

• Figure 1 created by analysis_basedisp.R
  • Top Panel: Population Proportions and Referral Proportions by Race, created in section 1
  • Bottom Panel: Racial Disproportionality Index in Referrals, created in section 2

• Figure 2 created by analysis_basedisp.R
  • Top Panel: Population Proportions and CPS Client Proportions by Race, created in section 3
  • Bottom Panel: Racial Disproportionality Index for CPS Clients, created in section 4
• Figure 3 created by prep_tractdata.R
  • Top Left Panel: Black Residents in Charlottesville by Census Tract, created in section 3
  • Top Right Panel: Charlottesville's Poverty Rates by Census Tract, created in section 3
  • Bottom Panel: Referrals to Charlottesville CPS by Census Tract, created in section 3

• Figure 4 created by figures_postreferral.R
  • Right Panel: Reporter Relation to Referred Children, created in section 1
• Figure 5 created by figures_postreferral.R
  • Top Panel: Children Accepted for Investigation/Assessment, created in section 2
  • Bottom Panel: Case Assigned to Investigation or Assessment, created in section 3
• Figure 6 created by figures_postreferral.R
  • Top Panel: Substantiation of Investigated Cases, created in section 4 of
• Figure 7 created by figures_predictions.R
  • Top Panel: Effect of Race on Post−Referral Outcomes, created in section 1

• Figure 8 created by figures_fostercare.R
  • Right Panel: Number of Foster Care Clients by Race, created in section 1
• Figure 9 created by figures_fostercare.R
  • Top Panel: Initial Placement Category by Race, created in section 2
  • Bottom Panel: Placement Category by Race, created in section 3
• Figure 10 created by figures_fostercare.R
  • Top Panel: Time in Each Placement By Race, created in section 6
  • Middle Panel: Number of Placements By Race, created in section 5
  • Bottom Panel: Duration of Care By Race (All Clients), created in section 7
• Figure 11 created by figures_fostercare.R
  • Top Panel: Nature of Exit from Foster Care by Race, created in section 4
• Figure 12 created by figures_fostercarepredictions.R
  • Top Panel: Effect of Race on Foster Care Placements, created in section 1
  • Bottom Panel: Effect of Race on Foster Care Outcomes, created in section 1