PROJECT 1
Project title:
Disparate Impact Analysis for the Service Effectiveness Model
Work in collaboration with:
Marc de Boer, Ministry of Social Development
Project description:
MSD is building an integrated view of the programmes and services it provides to help people receiving income support into employment. Part of this work is building ways in which to understand the profile of participants in these types of interventions. The project will involve with coming up with different ways in which to best to represent this information for business users, and then implementing these ideas in an R shiny dashboard tool.
Possible tasks for the internship:
- Data preparation
- Summary statistics describing the data
- Developing metrics in SAS or R
- Developing components for a Shiny R dashboard
Requirements:
- Technical/quantitative background: Mathematics, Statistics, Computer Science
- Strong in data visualization techniques
- Familiarity with R and libraries such as shiny, ggplot and plotly
- Good grasp of data processing techniques, knowledge of SAS would be an advantage
PROJECT 2
Project title:
Menu of Services to the Front Line
Work in collaboration with:
Tony Simmers, Principal Analyst, Ministry of Social Development
Project description:
Case managers at Work and Income Service Centres operate in a fast paced and complex environment. Part of their job is to match clients facing barriers to employment with appropriate training courses or other assistance. The assistance available to each region, and the process for making the link, varies between sites.
The project focusses on the operational deployment within a core enterprise system of a visualisation and user interface for services tailored to each client. The aim is to improve the way case managers access these data and in the process drive better outcomes for our clients.
As such the project is more about the IT development needed to move insights from discovery into production, than the use of mathematics to model processes or outcomes. The team is co-located with software engineers and is working within an agile framework.
Possible tasks for the internship:
- Assisting with the coding of the data preparation and visualisation, including ensuring code is of high quality and fully testable
- Assisting with the interpretation and analysis of test results
- Looking at data on services within the community that could expand the range of services presented as choices. Sourcing, describing, and cleaning these data would be required
Requirements:
- Technical/quantitative background: Computer Science, Mathematics, Statistics, or similar; ideally Masters level or higher
- Experience with data preparation and visualisation, particularly using one or more of Python, SAS, and SQL within a Linux environment
- Familiarity with website development, including CSS and JavaScript would be an advantage
- The nature of the project will not require significant statistical or modelling work
PROJECT 3
Project title:
Disparate Impact Analysis for the Service Effectiveness Model
Work in collaboration with:
Alexandra Chouldechova, Assistant Professor of Statistics and Public Policy, Carnegie Mellon University, USA
Project description:
It is not straightforward to understand the impacts of a model on sub-groups of a population. For example a model may affect one ethnic group differently even if ethnicity is not a predictor. As an initial working premise the term “disparate impact” has been chosen to summarise the potential for uneven effects of a model or decision process across sub-groups of a population.
The primary goal of this project is to develop a framework for assessing disparate impact of analytical models. As part of the assessment process we are going to identify the problems in this space and propose mitigation options.
As a case study, we are going to focus on the Service Effectiveness Model. This is a recent model developed by MSD to allocate clients into services.
Possible tasks for the internship:
- Data preparation
- Summary statistics describing the data
- Contributing to model re-implementation in R
- Analysis of the model focusing on different sub-groups identified in the data
- Writing a report of the main findings – can also be developed in the form of an interactive web application or dashboard
Requirements:
- Technical/quantitative background: Mathematics, Statistics, Computer Science or similar; ideally Masters level or higher
- Strong statistical background
- Coursework covering Machine Learning techniques (familiarity with Random Forests and other ensemble learning methods)
- Good grasp of data processing techniques, knowledge of SQL
- Strong scientific computing skills using R and libraries (e.g. dplyr, ggplot2, shiny)
- Familiarity with data visualization techniques
PROJECT 4
Project title:
Evidence syntheses for the social sector
Work in collaboration with:
Research and Evaluation Team, Insights and Investment, Ministry of Social Development
Project description:
There is a need for accessible and easily digestible ways of navigating the huge amount of research evidence out there about priority topics. The intern on this topic will gather, summarise and synthesise the evidence on these topics to construct an evidence ‘map’ which details what we know about what works, for whom and why. The first evidence map will be on interventions for family violence; subsequent topics may include housing, and mental health. This project is an excellent opportunity to develop or extend your skills in evidence synthesis and communicating research findings to a variety of audiences.
The primary goal of this project is to develop an evidence map of the effectiveness of interventions for family violence on a range of outcomes. Evidence maps are a layered summary of information about the evidence on a particular topic, depicting the general strength of evidence at a high level and enabling deeper exploration of different types of interventions, outcomes, and particular pieces of evidence (e.g. systematic reviews). Their purpose is to inform research priorities and facilitate the use of evidence in decision-making.
Possible tasks for the internship:
- Reviewing the literature
- Statistical meta-analysis
- Working with stakeholders in MSD and other agencies to identify their evidence needs
- Identifying evidence gaps and proposing research to fill them
- Writing evidence syntheses (such as reviews and evidence maps)
Requirements:
- Social/behavioural sciences background: third year undergraduate or higher
- Statistical knowledge
- Excellent communication skills
- Pro-active and enthusiastic approach
- Ability to integrate and synthesise research findings from a range of sources