Pedagogy of teaching with large datasets: Designing and implementing effective data‐based activities

Abstract

Integrating the use of large datasets into our teaching provides critical and unique opportunities to build students' skills and conceptual knowledge. Here, we discuss the core components needed to develop effective activities based on large datasets, which align with the 5E learning cycle. Data-based activities should be structured around a relevant question, use authentic publicly accessible data, be scaffolded to include choice, and involve discussion of the results.It is important that the software that is used to manipulate, analyze and/or visualize the data is accessible for students. There are a range of strategies to reduce the barriers of working with large datasets through pre-organizing and pre-scripting code for analyses, using online cloud-based versions of software, and reducing opportunities for error in syntax. Resources exist for learning open-source software (e.g., Data Carpentry) as well as for support and professional development in teaching with large datasets (Project EDDIE).

Providing a Common Approach to Designing Dataset-Based Learning Activities Based on a Literature Review

“…Consideration should also be given to the time the students might need to learn to use the given software, and if necessary, the teacher can provide software, spreadsheets, or pre-written code to assist them. This is recommended when the level of the students is not adequate to the task and could help them focus on analyses and results without getting frustrated [74].…”

“…As students are unlikely to arrive at identical solutions in the final section of the activity, this presents a chance for discussion that could prompt students to contemplate their answers and the methods they used to arrive at them. This is expected to encourage students to reflect on their answer and their method for generating it [74]. During the self-reflection phase [76], individuals go through various internal responses, like emotions and their level of contentment, and may react by adjusting or defending themselves based on their learning experiences and results.…”

The National Health and Nutrition Examination Survey as a Tool to Teach Data Analysis to Public Health Students

“…However, the availability of datasets that are the appropriate size and that contain relevant information interesting to each student is often difficult for instructors to locate. Exposure to real world study data has been shown to build students’ research, quantitative and critical thinking skills, as well as improve their reflective practices (Atenas et al, 2015; Atenas & Havemann, 2015a, 2015b; O’Reilly et al, 2022). Therefore, human health study data may help students gain a deeper understanding of how data collection, data management, and data analysis relate to the health of the public, and how their findings can be translated to changes in health care and policy.…”

Cracking the code: An evidence-based approach to teaching Python in an undergraduate earth science setting

“…Yet formal scientific computing instruction is often absent in earth science curricula, including oceanography (Old, 2019), except for highly scaffolded modules that employ coding in courses where programming is not the primary focus (e.g., Rowe et al, 2021). Even in courses that more extensively utilize programming within activity modules, such as those distributed by Project EDDIE (Environmental Data-Driven Inquiry and Exploration), pre-written code is usually provided to students (O'Reilly et al, 2022). In this void, brief but intensive hands-on workshops like those offered by Software Carpentry (https://software-carpentry.org; Wilson, 2016), Data Carpentry (https://datacarpentry.org/; Irving, 2019), and scientific societies (e.g., Arms et al, 2020) have provided crucial training to young scientists.…”