Three AACR Undergraduate Research studnets presented posters of their work at the MSU 20th annual University Undergraduate Research and Arts Forum (UURAF) on April 13, 2018.
Hailey Cockerill presented Question Development Using Undergraduate Students’ Writing About the Origin of Genetic Variation
Tanner Foster presented Computer Model Development to Analyze Student Structure Function Understanding
John Knapp presented Student Conceptions of Structure-Function Relationships in Cell Membranes
Congratulations to each of them on their outstanding work!
Question Development Using Undergraduate Students’ Writing About the Origin of Genetic Variation
Mentor(s): - Mark Urban-Lurain
Abstract: Genetic variation is an important core concept for undergraduate biology education as highlighted in “Vision and Change In Undergraduate Biology Education: A Call to Action” (AAAS, 2011). How new genetic variants originate in a population is challenging for students to understand because of the common misconception that mutations are harmful. The Automated Analysis of Constructed Response research group (AACR) develops computer-automated tools to analyze students’ writing that use statistical analysis to predict human scoring. Our research group develops assessments focusing on “big ideas” in undergraduate STEM courses. To reveal student thinking on genetic variation, AACR uses constructed response assessments, in which students write short answers using their own words. In this study we developed a CR question about how a rapid increase in frequency of a new coat color in a population of buffaloes occurred. A qualitative analysis of responses using computerized tools (n=401) revealed several themes. However, many students restated phrases directly from the prompt, which lead us to revise the prompt. Qualitative analysis of most recent data responses (n=143) led to another prompt revision, which was distributed at two research universities. Analysis of data from University #1 revealed a strong reference to sexual mechanisms of the origin of genetic variation. This trend could indicate different instructor styles that may influence students’ writing; however, more data is needed. Variation in student writing is beneficial for model generation; therefore, the resulting computer model will be capable of handling a diversity of written student language.
Computer Model Development to Analyze Student Structure Function Understanding
Mentor(s): - Mark Urban-Lurain, John Merrill, Kevin Haudek
Abstract: “Structure and function” is one of five core concepts for biological literacy identified in Vision and Change for Undergraduate Biology Education. However, many students struggle to understand this topic. Compounding the problem, increasingly large class sizes make understanding what students think especially challenging. The Automated Analysis of Constructed Response (AACR) group is developing computer tools that analyze student writing, which can be applied in classrooms of any size. To do this, we are creating machine-learning models that can accurately predict expert scores for student constructed responses to a question about enzyme binding, a common example of structure and function relationship in introductory biology courses. We developed an analytic rubric that contained bins ranging from structure-focused responses, such as lock-and-key or induced fit, to function-focused responses discussing energetics or reaction rates. The classification used for these bins was developed through analysis of student responses, as well as instructor interviews in which learning goals were identified among various courses. Differences in terminology used between instructors from different disciplines were incorporated into rubric development in order to encompass a wide range of student responses. Reports generated from these models can be used to guide instruction and better allow instructors to address misconceptions in the classroom.
Student Conceptions of Structure-Function Relationships in Cell Membranes
Mentor(s): - Kamali Sripathi, Mark Urban-Lurain
Abstract: Student understanding of the structure-function relationship is an important core concept for undergraduate biology education as seen in “Vision and Change In Undergraduate Biology Education: A Call to Action” (AAAS, 2011). Structure-function specifically regarding cellular membranes is an important underlying concept for understanding most cellular processes. The goal of this research is to identify the most common student ideas about cellular membranes. Multiple Choice (MC) questions may not reveal the complete thought processes of students. Constructed response (CR) questions offer a more complex view of student thinking than MC questions. The Automated Analysis of Constructed Response (AACR) research group uses machine learning to provide lexical and statistical analysis of student constructed responses that predict expert human scoring. To understand student thinking about structure-function, we are developing a question that asks students to explain why individual phospholipid molecules in a membrane do not flip sides. We collected responses from 777 students from two universities. Consensus scoring between three scorers was done to develop consistent scoring and to uncover other emergent student conceptions that may be added to the scoring rubric. The most prevalent concepts about why membrane phospholipids will not flip include: unfavorable interactions between hydrophobic/hydrophilic parts of the lipids, unfavorable energy requirements, and integral proteins preventing movement. To further understand student thinking, eleven student interviews were conducted. Next steps include analyzing the student interviews, and obtain inter-rater reliability among human scorers. Our goal is to provide reports to instructors that represent the complexity of student thinking about this question.