Search

Student persistence in science, technology, engineering, and mathematics (STEM) has long been a focus of educational research, with both quantitative and qualitative methods being used to investigate patterns and mechanisms of attrition. Some studies have used machine learning to predict a student's likelihood to persist given measurable classroom factors and institutional data, while others have framed persistence as a function of a student's social integration in the classroom. While these methods have provided insight into broader underlying patterns of attrition in STEM, they have not investigated class structures or teaching methods that promote persistence. In this study we explore how a research-based instructional format for an introductory calculus-based physics class using real world problem-solving (RPS) was correlated with higher persistence for students at a large research-intensive university. We found that the one-year persistence rates for the RPS course were 74% (fall semester) and 90% (spring semester), while the lecture-based class had a persistence rate of 64% and 78%, respectively. In spring, the RPS persistence rate was significantly higher (p=0.037). The RPS also had higher final grades and larger learning gains than the lecture-based class despite lower scores on a physics diagnostic test. We also note that the higher rates of persistence were not completely explained by higher final grades. This study motivates future work to understand the structural mechanisms that promote student persistence in introductory physics courses.

This paper describes measurement systems required to forecast, measure, and control activities for large, complex software development and support programs. Initial software cost and quality analysis provides the foundation for meaningful management decisions as a project evolves. In modeling the cost and quality of software systems, the relationship between the functionality, quality, cost, and schedule of the product must be considered. This explicit relationship is dictated by the criticality of the software being developed. This balance between cost and quality is a viable software engineering trade-off throughout the life cycle. Therefore, the ability to accurately estimate the cost and quality of software systems is essential to providing reliable software on time and within budget. Software cost models relate the product error rate to the percent of the project labor that is required for independent verification and validation. The criticality of the software determines which cost model is used to estimate the labor required to develop the software. Software quality models yield an expected error discovery rate based on the software size, criticality, software development environment, and the level of competence of the project and developers with respect to the processes being employed.