Abstract The “leaky pipeline” is the oft-used metaphor to characterize the loss of STEM students in the progression from high school through college and into the labor force. This view of STEM education—which posits a unidirectional flow out of STEM fields—is both pervasive and influential in K-12, higher education, and workforce development policy. But there has been limited empirical support for this claim that “leaving STEM” is unidirectional and reduces the size of the STEM cohort. This study analyzes two cohorts in nationally representative, longitudinal surveys of four-year college students to examine freshman-to-bachelor's degree pathways. We find that although a third of STEM-intending freshmen leave STEM fields and graduate in non-STEM majors, the size of the STEM graduating class is larger than the freshmen STEM class in the two cohorts of bachelor's degree graduates. STEM graduates in both cohorts are comprised of large shares of students who did not initially declare a STEM major: just under 20% of graduates first began as non-STEM majors and were “late entrants” in STEM. Thus, rather than a leaky pipeline, STEM disciplines appear to be a two-way street, with many students attracted into STEM fields during college. The findings suggest that the dominant focus on STEM retention should be broadened to consider STEM “attraction” during college, since a large share of STEM graduates begin college as non-STEM and undeclared majors.

Intelligent tutoring systems are known for providing customized learning opportunities for thousands of users. One feature of many systems is differentiating the amount of practice users receive. To do this, some systems rely on a threshold of consecutive correct responses. For instance, Khan Academy used to use ten correct in a row and now uses five correct in a row as the mastery threshold. The present research uses a series of randomized control trials, conducted in an online learning platform (eg., ASSISTments.org), to explore the effects of different thresholds of consecutive correct responses on learning. Results indicate that despite spending significantly more time practicing there is no significant difference on learning between two, three, four, or five consecutive correct responses. This suggests that systems, and MOOCS, can employ the simple rule of two or three consecutive correct responses when determining the amount of practice provided to users.

Dilated cardiomyopathy (DCM) is the third most common cause of heart failure and the primary reason for heart transplantation; upward of 70% of DCM cases are considered idiopathic. Our in-vitro experiments showed that reduced hybrid/complex N-glycosylation in mouse cardiomyocytes is linked with DCM. Further, we observed direct effects of reduced N-glycosylation on K v gating. However, it is difficult to rigorously determine the effects of glycosylation on K v activity, because there are multiple K v isoforms in cardiomyocytes contributing to the cardiac excitation. Due to complex functions of K v isoforms, only the sum of K + currents (I Ksum ) can be recorded experimentally and decomposed later using exponential fitting to estimate component currents, such as I Kto , I Kslow , and I Kss . However, such estimation cannot adequately describe glycosylation effects and K v mechanisms. Here, we propose a framework of simulation modeling of K v kinetics in mouse ventricular myocytes and model calibration using the in-vitro data under normal and reduced glycosylation conditions through ablation of the Mgat1 gene (i.e., Mgat1KO). Calibrated models facilitate the prediction of K v characteristics at different voltages that are not directly observed in the in-vitro experiments. A model calibration procedure is developed based on the genetic algorithm. Experimental results show that, in the Mgat1KO group, both I Kto and I Kslow densities are shown to be significantly reduced and the rate of I Kslow inactivation is much slower. The proposed approach has strong potential to couple simulation models with experimental data for gaining a better understanding of glycosylation effects on K v kinetics.

Mental models—internal, dynamic, incomplete representations of the external world that people use to guide cognitive processes such as reasoning, decision making, and language comprehension—have practical implications for predicting attitudes and behaviors across various domains. This study examines how mental models of the human–nature relationship predict pro-environmental behavioral intentions directly and indirectly as mediated through anthropocentric and biocentric environmental attitudes. To address these aims, participants were asked about mental model components of the human–nature relationship (human exceptionalism, beliefs about human impact on nature, and beliefs about nature's impact on humans), pro-environmental attitudes (biocentric and anthropocentric), and their pro-environmental behavioral intentions (protection and investment). We found that protection intentions were (1) directly predicted by human exceptionalism beliefs (negatively) and perceived human impact on nature (positively) and (2) indirectly predicted by mental model components via biocentric attitudes. Investment intentions were directly predicted by nature's perceived impact on humans, and were similarly indirectly predicted by mental model components via biocentric attitudes. The results suggest that mental models of the human–nature relationship provide a cognitive foundation for environmental behavioral intentions both directly and through their association with environmental attitudes. These findings have implications for pro-environmental interventions that deal with conceptual and attitudinal change.

For decades, politicians, policy makers, and industry leaders have told a disturbing narrative: An inferior educational system, this account goes, prevents many young Americans from entering STEM (science, technology, engineering, and mathematics) fields, weakening the nation's innovation and scientific enterprise. The resulting talent shortage, it reasons, can only be overcome by admitting larger numbers of foreign nationals who have the required skills. But this narrative misrepresents what actually drives large numbers of capable, ambitious, well-educated Americans away from STEM fields and toward careers in business, law, medicine, and other nonacademic endeavors: a university labor system that exploits would-be scientists.

As a cold war with China heats up, the U.S. and other members of the G7 need new approaches to their science and technology innovation (STI) policies. Dominance on the innovation frontier is no longer possible through traditional techno‐nationalist policies that view nations as ‘competing’ through exclusive STI development. Instead, we must recognise that talent and intellectual property are globally distributed, and thus build global collaborations that draw on the world's greatest talent while providing benefits equitably in a global STI commons. We need to recognise this new reality, not only for the benefits this would confer on humankind, but also to contend with China's growing STI capabilities and, eventually perhaps, integrating China into a system of global collaboration. Additionally, and importantly, national policies must recognise the geographically untethered operations of multinational enterprises that are the developers and/or repositories of STI but have weak ties to any one nation, thus blunting policies that try to contain STI within a country's borders. In this paper, we suggest approaches to advance these goals for global STI based on theories and cases of collective action.

STEM acronymically refers to four areas of inquiry – Science, Technology, Engineering, and Mathematics. But as its use has become ubiquitous, STEM has taken on social and political meaning far beyond the sum of its component parts. In this paper, we take a first step in clarifying the analytic categories of STEM in education. This, we propose, is a necessary first building block for STEM analysis – to understand what constitutes STEM coursework, the constituent element of a STEM education. We first review the STEM definitional problems we have identified in the process of examining two sets of NCES nationally-representative data, provide analysis of the extent of potential mismeasurement, and estimates of impact. We then outline an approach to resolving the mismeasurement problems in nationally-representative postsecondary student surveys.

Research experiences provide educational and intellectual benefits to undergraduate students alongside traditional coursework. However, little prior work has directly examined the relationship between a student's participation in research and their later employment outcomes. Drawing on years of transaction-level administrative records from nine universities, we identify students with paid research experiences on grants from NIH, NSF, and other federal funders alongside a comparison group of students with paid federal work-study employment. Record linkage to later employment information allows us to follow students beyond their university years, measuring estimated salary (n = 15,528) and employment churn (n = 16,628). Using generalized linear models, we estimate differences between research and work-study experiences associated with these two outcomes. Those with federally funded paid research work as undergraduates, our findings demonstrate, have higher salaries and fewer job changes in their early careers. These results support the vision of the university as a site of integrated research and teaching, where students' engagement with research contributes to personal and professional development.

In several author name disambiguation studies, some ethnic name groups such as East Asian names are reported to be more difficult to disambiguate than others. This implies that disambiguation approaches might be improved if ethnic name groups are distinguished before disambiguation. We explore the potential of ethnic name partitioning by comparing performance of four machine learning algorithms trained and tested on the entire data or specifically on individual name groups. Results show that ethnicity-based name partitioning can substantially improve disambiguation performance because the individual models are better suited for their respective name group. The improvements occur across all ethnic name groups with different magnitudes. Performance gains in predicting matched name pairs outweigh losses in predicting nonmatched pairs. Feature (e.g., coauthor name) similarities of name pairs vary across ethnic name groups. Such differences may enable the development of ethnicity-specific feature weights to improve prediction for specific ethic name categories. These findings are observed for three labeled data with a natural distribution of problem sizes as well as one in which all ethnic name groups are controlled for the same sizes of ambiguous names. This study is expected to motive scholars to group author names based on ethnicity prior to disambiguation.

How can we evaluate the performance of a disambiguation method implemented on big bibliographic data? This study suggests that the open researcher profile system, ORCID, can be used as an authority source to label name instances at scale. This study demonstrates the potential by evaluating the disambiguation performances of Author-ity2009 (which algorithmically disambiguates author names in MEDLINE) using 3 million name instances that are automatically labeled through linkage to 5 million ORCID researcher profiles. Results show that although ORCID-linked labeled data do not effectively represent the population of name instances in Author-ity2009, they do effectively capture the 'high precision over high recall' performances of Author-ity2009. In addition, ORCID-linked labeled data can provide nuanced details about the Author-ity2009's performance when name instances are evaluated within and across ethnicity categories. As ORCID continues to be expanded to include more researchers, labeled data via ORCID-linkage can be improved in representing the population of a whole disambiguated data and updated on a regular basis. This can benefit author name disambiguation researchers and practitioners who need large-scale labeled data but lack resources for manual labeling or access to other authority sources for linkage-based labeling. The ORCID-linked labeled data for Author-ity2009 are publicly available for validation and reuse.