****NEED ASSIGNMENT COMPLETELY REWRITTEN. I HAVE INCLUDED MY ORIGINAL ASSIGNMENT AS WELL AS THE FEEDBACK FROM INSTRUCTOR. PLEASE RECOMPLETE ASSIGNMENT WITH INSTRUCTOR’S FEEDBACK. I HAVE ATTACHED ALL RELEVANT DOCUMENTS TO ASSIST YOU WITH THIS MEMO**** Before beginning assignment, watch these videos:
1: Predictive Algorithms: https://vimeo.com/961568543?p=1l
2: Review of Sample Policy Memo: https://vimeo.com/962581752?p=1l
Before you review this assignment and download the attached documents, you must have reviewed the above videos, along with the sample memo, in proper sequence so that you have a foundation to understand the expectations of the assignment.
This assignment requires you to generate a policy analysis using a stakeholder analysis of policy options to address bias in predictive algorithms that rates a defendant’s risk of a future crime. Known as “risk assessments,” these algorithms are used to inform judges about who can be set free and what level of bond amounts can be set. Predictive algorithms use historical data to calculate a probability of future events, similar to how a sports bookie determines odds for a game or pollsters forecast an election result. Independent studies have found that when these algorithms forecast who would re-offend, the algorithm makes mistakes among black and white defendants. For instance, the formula can falsely flag black defendants as future criminals wrongly labelling them this way at almost twice the rate as white defendants. On the other hand, white defendants can be mislabelled as low risk more often than black defendants.
Given this backdrop, the Sentencing Commission for the Commonwealth of Pennsylvania is considering adopting a predictive algorithm since the Pennsylvania legislature mandated that the commission consider an algorithm to help courts decide the sentence after someone is convicted. During a recent hearing by the commission to adopt a predictive algorithm, the American Civil Liberties Union protested the adoption on the grounds that the program will exacerbate rather than reduce racial bias in sentencing.
Matthew Kleiman, Executive Director of the Pennsylvania Commission on Sentencing requests a policy analysis to contain bias when a predictive algorithm is used. Kleinman’s question for the policy analyst is: How can public-sector algorithmic systems be vetted and evaluated to contain bias?
This assignment requires you write a five-to-six-page policy analysis memo (single spaced) that includes an introduction with the client’s policy question, an assessment of stakeholder’s positions regarding the policy question raised by the client, an options specification that examines each stakeholder’s level of support for a specific policy option, an assessment of options that examine the level of feasibility of adopting a particular policy, and a recommendation based on the option assessment.
Download the attached document that include the stakeholders and the policy options. You will use this information to write the memo and fill in the tables to perform your policy analysis. Also, attached are links and sources that explain the use predictive algorithms in government for your reference.
NOTES:
IMPORTANT ITEMS FOR THE POLICY MEMO: PLEASE READ AND UNDERSTAND
ITEM 1: The Status Quo Means Keeping Things as they Currently Stand. At the current time, the commission is not using predictive algorithms. So, what that option means is the commission would continue to use human decision-making and not opt to use predictive algorithms without or without safeguards. Nonetheless, this option does leave the opportunity to re-visit the potential use of algorithms in the future.
ITEM 2: Use of Predictive Algorithms Without Safeguards is NOT the Status Quo. See my comments in item #1. However, the Use of Predictive Algorithms Without Safeguards could be a separate policy option in addition to the options provided in the assignment.