Lake Forest College

This paper examines the effect of financial literacy on household investment decisions using data from the 2022 wave of the Survey of Consumer Finances. Using parental education as an instrument, the study estimates instrumental variables probit models for a broad set of financial assets and contributes new evidence on the effect of financial literacy on ownership of a range of asset classes. The results show that financial literacy significantly increases participation in complex, growth-oriented investments, such as individual retirement accounts, mutual funds, and stocks, while having no significant effect on simpler financial products. Despite large demographic differences in literacy and market participation, the study could not find evidence for heterogeneous effects of financial literacy based on sex, income level, and race. Once acquired, the benefits of financial literacy appear to be broadly similar across sex, income, and racial groups.

Gcd-graphs represent an interesting and historically important class of integral graphs. Since the pioneering work of Klotz and Sander, numerous incarnations of these graphs have been explored in the literature. In this paper, we define and establish some foundational properties of gcd-graphs defined over a general finite commutative ring. In particular, we investigate the connectivity and diameter of these graphs. Additionally, when the ring is a finite symmetric & Zopf;/n-algebra, we give an explicit description of their spectrum using the theory of Ramanujan sums that gives a unified treatment of various results in the literature.

The Roman Space Telescope Coronagraph Instrument will demonstrate a series of technologies and techniques to enable the direct detection of reflected-light planets with space-based observatories. To characterize and validate the performance of the Coronagraph Instrument, the Community Participation Program is developing corgidrp, an open-source Python-based data reduction pipeline. The pipeline can process data from the required and best-effort observing modes and their associated calibration sequences into calibrated science-ready data products. We present the software design and implementation of corgidrp and the motivation behind specific design decisions. We describe the software architecture, data flow, processing steps, automation tools, testing framework, and development philosophy. We also outline future development plans in preparation for on-sky data.

Large Language Models (LLMs) such as ChatGPT are transforming how scientists conduct and validate research, offering promise as tools to improve scientific reproducibility. However, computational reproducibility and error detection remain expensive and labor-intensive. We experimentally test how collaboration between researchers and LLM assistants influences the reproduction of quantitative social science findings across different levels of AI autonomy. We randomly assigned 288 researchers to 103 teams working under three conditions: human-only, AI-assisted (using ChatGPT as a collaborative tool), or AI-led (ChatGPT operating with minimal human oversight). Teams reproduced published results from leading social science journals, detected coding errors, and proposed robustness checks. Human-only and AI-assisted teams achieved comparable reproduction rates (94% vs. 91%) and performed similarly on most outcomes, except human-only teams identified significantly more major coding errors. Both substantially outperformed AI-led teams, which achieved only a 37% reproduction rate, detected fewer errors across all categories, proposed weaker robustness checks, and required more time. This autonomous approach, however, likely represents only a lower bound of AI capabilities. Despite rapid model advances, expert human judgment currently remains indispensable for reliable empirical verification. While AI assistance did not degrade most outcomes, it provided no measurable advantages and was associated with reduced detection of major errors. However, the 37% autonomous reproduction rate indicates that AI could provide value in settings where scale or cost constraints preclude human review of papers, even though general-purpose LLMs offer no immediate advantages for human-supervised verification.

A scoping review of empirical studies from 2000 to 2023 was conducted to explore the relations between teacher well-being and student well-being and learning. The search process was guided by a broad definition of well-being and learning, yielding 7,629 unique publications. Sixty-seven studies met the inclusion criteria. Findings indicated an increase in studies published since 2000, with a peak in 2020 and 2021. The greatest number of studies were conducted in the United States (34%) and with elementary school students (37%). To capture teacher well-being, most studies used a conceptualization of occupational stress and burnout (33%), while psychological distress was most common for student well-being (48%). Overall, teacher well-being was positively associated with both student well-being (93% of studies) and learning (84%). Studies used diverse research designs-cross-sectional designs were most common (54%), followed by qualitative (22%) and longitudinal approaches (22%). Noticeably absent were any experimental studies examining a causal link between teacher well-being and student well-being and learning, highlighting the need for more research.