Using large public datasets in the undergraduate ecology classroom

Le résumé fourni par la source

Ecology and environmental sciences are in the midst of an "ecoinformatics" revolution, including real-time data streams from sensor networks, remote-sensing program data, and dataset archives (Michener and Jones 2012). These data are creating unprecedented opportunities to investigate environmental questions at regional, continental, and global scales (Peters 2010). However, many instructors and students are ill-prepared to engage in data-intensive, large-scale research (Hernandez et al. 2012; Strasser and Hampton 2012). The skills needed to engage in such research include how to (1) formulate and test hypotheses involving multiple factors and at large spatial or temporal scales; (2) use efficient methods for locating and retrieving data; (3) synthesize diverse data sources; (4) critically evaluate data quality; (5) manage and manipulate large datasets; and (6) use ethical practices in acquiring data and crediting data authors (Michener and Jones 2012). In 2009, the Ecological Society of America (ESA), the National Center for Ecological Analysis and Synthesis (NCEAS), and the National Ecological Observatory Network (NEON) sponsored a distributed seminar entitled Engaging Undergraduate Students in Ecological Investigations Using Large, Public Datasets (https://groups.nceas.ucsb.edu/big-data/front-page). Over a 2-year period our team – faculty from five colleges (including historically black, tribal, and primarily undergraduate institutions) and colleagues from the three sponsoring organizations –worked together to develop, implement, assess, and disseminate classroom exercises using publicly available ecological datasets. Our objectives were to create exercises that, by working with real data, enhance students' ecological knowledge and critical thinking skills, engage students in the synthesis of ecological knowledge, and instill in students an understanding of the value of publicly available archived data. We also intended these exercises to introduce 21st-century "ecoinformatics" skills in data mining, processing, and analysis, and to be of practical use for undergraduate institutions and with any population of undergraduate ecology or introductory biology students. Six exercises were developed and tested on undergraduate classes by members of the working group: four are published in Volume 8 of Teaching Issues and Experiments in Ecology (WebPanel 2). These exercises facilitated student exploration of publicly available national or global-scale data, including datasets on water quality, avian species richness, land use and land cover, invasive plant species distributions, wildfires, spread of West Nile virus in North America, ice cover in the Arctic Ocean, and species richness in conservation protected areas. Assessment. Pre and post surveys – measuring (1) learning gains related to the ecological concepts that the six classroom exercises were intended to teach, (2) attitudes about those exercises, and (3) general perceptions about using publicly available data in ecology – were administered to 130 undergraduate students at five colleges. We also brought two students from each participating institution to a meeting of the working group at NCEAS, to provide their perspectives on the exercises we developed and to share and help synthesize the distributed seminar's results. Critical thinking and synthesis. Qualitative responses to the surveys and interviews with students indicated that students valued the opportunity to explore real data about ecological issues and appreciated the tangible connections with other course content. They felt motivated to complete these exercises despite the challenging nature of working with big datasets, because they "really wanted to know". For the students, completing the exercises was more similar to real research, more open-ended, with the possibility of making an authentic discovery, than they normally felt was true in ecology exercises. Many thought the concepts and skills acquired by doing the exercises would be valuable in their intended careers. However, they also found the exercises to be frustrating at times. These exercises were collaborative projects and students experienced the typical challenges of working in groups. For example, for one exercise at one institution (see www.esa.org/tiee/vol/v8/experiments/langen/abstract.html) the post surveys indicated that students were convinced that, by performing a big data exercise, they learned ecological concepts (88% agreed or strongly agreed) and improved skills involved in locating and acquiring data from the internet (83% agreed), conducting research (83% agreed), and communicating research (75% agreed). Moreover, the students demonstrated that they had learned three concept objectives related to spatial scale, land-cover/land-use patterns, and local-to-continental patterns of species richness. The outcomes were assessed by two questions that required a student to apply the concepts to novel questions about patterns of biodiversity and land use around college campuses; students demonstrated substantial gains in the accuracy and sophistication of their responses between the pre and post tests. Data from surveys of other exercises at other institutions indicated qualitatively similar results, but also that in some cases all learning objectives may not have been met, necessitating further refinement of the exercises. Building ecoinformatic skills. The exercises were time-consuming. The students were most challenged by the number of conceptual and process skills that had to be mastered: generating hypotheses that were testable with the available data, locating and navigating databases to aggregate the appropriate data, interpreting metadata, downloading and processing data, organizing and coordinating data management tasks, performing statistical analyses, and generating and interpreting data visualizations. These were skills that they had not acquired in their previous coursework. Value of publicly available data. There was a diversity of student attitudes toward using publicly available ecological data, mirroring the breadth of attitudes across ecology at large (Zimmerman 2008). A few students had attitudes similar to the faculty involved in the seminar: these data provide an invaluable resource for investigating questions at larger spatial and temporal scales, and with a greater variety of data, than can be done by individual investigators and their classes. Surprisingly, while some students considered the online data to be "authoritative", others felt that these data were less reliable than self-collected data because of lack of confidence in the quality of data collected by others, and uncertainties about methods used to collect the original data (this was unexpected given that the datasets included very detailed metadata). Some students also expressed concern that archival data might be obsolete, and that self-collected data would be "fresher" and therefore more reliable. Many of these same students felt that using other researchers' data was not really "doing science", and some worried that it was a form of intellectual property theft. Our working group was surprised at how difficult it was to find the specific data needed for an exercise online (if they existed at all). Creating adequate data templates and documents to guide students was also very challenging; we found that students required considerably more technical help than is typical for an ecological exercise. We needed to carefully "scaffold" concepts and tasks to make these exercises doable by our students. Additionally, at some institutions, we had to deal with imposed cyber-security blocks on downloading computer applications and data onto instructional computers. Select datasets that help students grapple with a compelling environmental issue or intriguing ecological concept in a way that supports authentic inquiry. Create adequate data templates and technical help documents to guide students.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.