Publications
Empowering Educational Leaders for AI Integration in Rural STEM Education: Challenges and Strategies
This study explored the role of educational leaders in integrating artificial intelligence (AI) into rural K-12 STEM education. The research focused on the perceptions, strategies, and barriers that educational leaders face in navigating AI integration, particularly in the unique context of rural schools. Drawing on survey and focus group data from rural school leaders, this study examined how rural educational administrators perceive AI's potential and the challenges associated with its adoption. It provided insights into leadership strategies for addressing infrastructural limitations, fostering professional development, and ensuring equitable AI adoption. Ultimately, this research aimed to empower rural educational leaders to harness AI technologies to enhance STEM education.
Completing advanced high school math coursework relates to better adulthood outcomes. Our understanding of why youth with learning disabilities (LDs) and/or ADHD have less access to high math course attainment is limited. Using data on around 20,000 adolescents from the High School Longitudinal Study of 2009, results indicate that, regardless of disability status, structural inequities in family social position are more salient for youth's math course attainment than formal disability programming, universal supports, or structural inequities in how students are sorted across schools. Among youth with the same disability status, youth from higher SES families, or whose parents have a STEM degree, have heightened access to high math course attainment even after accounting for prior achievement. Disparities in access to high math course attainment that persist net of controls for both youth with an LD and youth with ADHD present the possibility of disability-related stratification and stigma during high school.
ABSTRACT The PISA assessment 2022 of creative thinking was a moonshot effort that introduced significant advancements over existing creativity tests, including a broad range of domains (written, visual, social, and scientific), implementation in many languages, and sophisticated scoring methods. PISA 2022 demonstrated the general feasibility of assessing creative thinking ability comprehensively at an international scale. However, the complexity of its assessment approach—such as time‐consuming scoring requiring human raters—implies the risk that it may not be easily applied by the scientific community and practitioners. In this commentary, we outline important next steps building on the PISA assessment to further enhance future assessments of creative thinking. Crucial future directions include 1) determining what tasks and scorings ensure high psychometric quality including content validity, 2) enabling efficient, objective scoring by applying AI methods such as Large Language Models (LLMs), 3) ensuring high language accessibility via multilingual tests, 4) targeting a broader age group, and 5) facilitating standardized, reproducible assessments via an open online testing platform. In sum, these developments would lead to an efficient, validated multilingual test of creative thinking, which enhances the accessibility of effective creative thinking assessments and thereby supports the democratization and reproducibility of creativity research.
A student's science identity has important implications for their persistence in science classes and success in STEM careers. Unfortunately, many marginalized students in the science field do not feel like scientists; thus, we need to understand ways in which we can help foster science identities in elementary education. Researchers surveyed 148 third- and fourth-grade students and their teachers across 10 public schools in the Mid-Atlantic region of the U.S. to understand students' science identity. A mixed methods approach, using a multi-level model and thematic coding of open-ended responses revealed that science curiosity, science self-efficacy, and teacher practices may be important factors to consider when examining students' science identity development. Hands-on experiences that match students' perceptions of what scientists do in the real world, positive affirmations like telling students they are scientists, and encouraging the process of doing science rather than needing to know all answers were common themes amongst students that felt high perceptions of science identity. Classroom practices and implications were explored.
Generative AI systems, especially Large Language Models (LLMs) such as ChatGPT, have recently emerged as significant contributors to creative processes. While LLMs can produce creative content that might be as good as or even better than human-created content, their widespread use risks reducing creative diversity across groups of people. In the present research, we aimed to quantify this homogenizing effect of LLMs on creative diversity, not only at the individual level but also at the collective level. Across three preregistered studies, we analyzed 2,200 college admissions essays. Using a novel measure—the diversity growth rate—we showed that each additional human-written essay contributed more new ideas than did each additional GPT-4 essay. Notably, this difference became more pronounced as more essays were included in the analysis and persisted despite efforts to enhance AI-generated content through both prompt and parameter modifications. Overall, our findings suggest that, despite their potential to enhance individual creativity, the widespread use of LLMs could diminish the collective diversity of creative ideas.
The authors report the results of an intervention designed to trigger cultural change in science, technology, engineering, and mathematics academia. Drawing on Dobbin and Kalev's managerial engagement approach, the authors mobilized faculty members as change agents by teaching them to deliver a prosocial bystander intervention curriculum to their students. Observations and in-depth interviews with faculty members reveal how opportunities to commit the content to memory, share examples from their own lives, and articulate its value to students increased accountability to the material. This led faculty members to expand their definition of bystander intervention beyond direct, in-the-moment confrontation and recognize past instances in which they had been prosocial bystanders, which boosted their confidence to intervene in the future. Interviews with nonfacilitating faculty members and a small, randomized experiment suggest that leading an activity creates greater readiness than merely attending a training. With universities currently under government pressure to eliminate programs that historically signaled compliance with civil rights mandates (e.g., diversity and sexual harassment trainings), it is important to identify more sustainable strategies for cultural change. The present findings chart potential pathways for building prosocial bystander readiness when the risks of intervening and the costs of not intervening may be increasing.
In the rapidly evolving world of hardware security, developing metrics for evaluating the security improvements of hardware designs is important. This work examines the prevailing threat model for secure analog-to-digital converter (ADC) architectures and explains how signal-to-noise ratio (SNR), root-mean-square error (RMSE), and bit-wise accuracy (BWA) are used to evaluate security improvements. The existing metrics are mathematically related through the proposed Proxy ADC framework. The proposed SNR-RMSE and BWA-RMSE relationships are validated using a power side-channel attack on a commercial ADC. The SNR-RMSE relationship achieves an average percent error of 1.69% across four trials, while the BWA-RMSE relationship achieves an average of 7.97%. Using results from past secure ADC works allows for additional demonstrations of the relationships. These relationships can estimate accuracy in a realistic attack scenario where ADC outputs cannot be measured to verify the evaluation, and recontextualize the metrics of standard ADC design for hardware security. Furthermore, the Proxy ADC framework allows for comparison of tradeoffs between designs' security and efficiency, revealing trends to leverage for future secure architectures.
Most deaf and hard-of-hearing children are born to hearing parents, often delaying exposure to their first language. This negatively influences development of not only language, but also many other aspects of cognition, including exact representations of large quantities. The core knowledge view of numeracy predicts that delays in language exposure should not affect nonverbal representations of small quantities (1-3). This study is the first to investigate effects of language modality (spoken vs. signed) and timing of language experience (early, from birth vs. later) on the representation of small quantities of objects. We adapted the "Mr. Elephant" task (Shusterman et al., 2017) and examined whether children (age 3 to 7 years) succeeded on trials involving quantities 2 and 3. A logistic regression found that Timing and Socioeconomic Status significantly predicted Mr. Elephant performance, while Modality and Age did not. Early-exposed children were more likely to succeed on the task than Later-exposed children. For an exploratory follow-up, two measures of language were added into the analysis: Highest Count, which records children's recitation of the count list, and Give-a-Number ('Give-N'), which assesses children's understanding of the cardinal principle (CP). This logistic regression found that Timing and Give-N performance significantly and independently predicted Mr. Elephant performance, but Socioeconomic Status and Highest Count did not. Children who were CP-knowers were more likely to succeed on Mr. Elephant than non-CP-knowers. These results suggest that the representation of small quantity representations is associated with the timing of children's language exposure and their knowledge of the cardinal principle.
What is the origin of exact numerical reasoning in humans? Previous studies report that innumerate humans are unable to recognize that two sets placed in one-to-one correspondence are exactly equal, and argue that this ability emerges when humans acquire symbolic numbers. Here we tested blind children and adults (n = 140), and found that despite having symbols to label and count large quantities, these participants did not use one-to-one correspondence to match large sets. We also found that blind individuals who had a gradient of visual sensitivity were more likely to match sets exactly, but participants who only used haptic exploration did not. Thus, numerate individuals (who can sometimes match sets exactly) are nevertheless unable to express their understanding of exact equality in a commonly used measure of this ability. We conclude that expressing knowledge of exact equality via set-matching requires not only conceptual understanding, but also experience with sophisticated sensori-motor procedures, suggesting a possible re-analysis of studies with innumerate and semi-numerate humans.
This study investigated the structure of the home math environment (HME) of preschoolers. Examinations include the relation between parental communal beliefs, home math engagement, children's math skill development, and whether communal beliefs mediated these relations. Parents reported data ( N = 652, 49% female, mean age = 4.25; parental demographics: 87.4% White, 9% Latine and Hispanic, 6% African American, 58% 4-year degree or higher, mean income = $70,000–$79,999). The HME structure diverged from prior work. Significant differences were found in home math engagement based on children's age. Communal beliefs were significantly different based on parental ethnicity and education. While children's HME and communal beliefs were significantly associated with their math skill development, communal beliefs were not a significant mediator. Overall, this study provided important progress toward understanding the role of these constructs in the development of children's early math skills.


