Jong-Min Kim’s Post

📚 New research co-authored by Policy Research staff members has been published in “Criminal Justice and Behavior!” Drs. Zottola, Desmarais, and Stewart examined whether the presentation of pretrial risk assessment results and the race of the person charged influenced pretrial court actors’ release decisions. The authors found that the presentation of pretrial risk assessment results (framed in terms of success or failure or presented as a frequency or a probability) did not have an impact on pretrial court actors’ release decisions. However, the race of the person being charged did impact release decisions. 📖 Learn more about what they found: http://pra.tw/TouL50OXrpq 🎧 Listen to our Data Points episode highlighting this research: http://pra.tw/A70A50OXrpo

Read all about our new research in the published article or listen to the accompanying podcast episode! Let me know your thoughts or questions.

Are you struggling to find time to get those technical reports and trials written up? LWT can help with everything from statistics to reporting and full journal papers. As experienced researchers, statisticians and journal editors and reviewers, our staff can guide you to assist in getting your research work in the public domain. Contact me with your needs.....and for a competitive quote!

The Abstraction and Reasoning Corpus (ARC): Where do we stand today? An article I wrote, featuring the perspectives and work of some of the people working on the benchmark: https://lnkd.in/eGTtWJFD

In this semester’s final blog post, MJLST Staffer David Lee discusses how new amendments to Federal Rule of Evidence 702 clarify the role of judges as arbiters of the admissibility of expert opinions, and why science literacy in the judiciary is imperative to properly and consistently serve this gatekeeping function “in an era increasingly dominated by complex scientific evidence.” Check it out here: https://lnkd.in/gxeYwmiA

Super interesting piece of #causalML research by my colleague Björn Bokelmann. We consider the evaluation of uplift models. Assessing causal uplift models is challenging and requires large amounts of test data, which is costly to acquire. Studying the variance of uplift evaluation metrics such as the Qini, we derive easy-to-compute adjustment methods for variance reduction. Theoretical and empirical analysis confirms the effectiveness of our adjustment methods. Simply put, they facilitate reducing the amount of costly test data w/o sacrificing evaluation quality; or raising the quality of evaluation using the same amount of test data. Sounds like a free lunch? Indeed, we expect our propositions to improve uplift model evaluation practices in a wide variety of settings, which the paper defines formally. 50-days free access to the paper is available via: https://lnkd.in/eTfHKbSV

Unlocking the mysteries of statistical analysis! In my latest blog gain valuable insights on statistical concept of Confidence Interval. Discover how confidence intervals empower decision-making and research and enhance your statistical literacy. #statisticalanalysis #confidenceinterval #research #datascience #analysis

In 2019 I participated in training complexity theory and communication at the Taiwan National University of On-going Adult Education with Shiro Cyril. This is his response: "We have used the draft of this paper and completed 10 sessions of 3 hours workshop in the last 3 months with about 30+ philosophy practitioners, and we have become quite familiar with this model in real life applications!"

What is The Honest Broker about, anyway? Data, analyses and commentary missing from public discussions of science, policy and politics Honest brokering, as it turns out, is a group effort, which is exactly how I view this publication, a group effort https://lnkd.in/giQaGMkq