Classroom response systems (CRS), or clickers, are used in many undergraduate introductory science classes to increase student engagement and learning. CRS also provide an opportunity to collect data on students' conceptions of geologic processes. Most research on students' conceptions uses qualitative methods, such as interviews, drawings, and sorting tasks. Research on classroom pedagogy often documents change using multiple-choice concept inventories. Smart device technology and the shift of CRS systems to web-based platforms has enabled most popular content management companies to offer click-on-diagram (COD) questions as a response option format. COD questions are open-ended, providing rich insight into students' understanding of diagrams and spatial thinking, yet more efficient than many qualitative approaches. This is a tremendous opportunity for the geosciences, which depend on diagrams to teach students about structures and processes that occur at scales greater than can be observed directly. In this proof-of-concept study, we developed COD questions that targeted a collection of well-documented and novel geologic misconceptions to evaluate the usefulness of COD questions as a tool for conceptions research. The COD questions were implemented in an introductory geology course for non-majors and administered prior to and directly after instruction, as well as at the end of the course. Students' responses to COD questions about the Earth's internal structure and geologic time confirmed misconceptions that are documented in the literature. COD questions about erosion to base level and hot spots identified previously undocumented conceptual challenges related to spatial processes. Plotting the COD data using Geographic Information Systems (GIS) software directly revealed the impact of instruction on the targeted conceptions – outcomes depended on the intensity of the instructional intervention and the complexity of the problem. The findings suggest that COD questions are an efficient way to probe for and document students' developing spatial conceptions of geologic structures and processes. Although the main focus of this study was to evaluate the usefulness of COD questions for research purposes, applications to classroom practice will be discussed.

Geoscience diagrams that illustrate 3D relationships can be challenging for students, but important because scientific practice requires comprehending and generating diagrammatic representations. Students' understanding of 3D diagrams in geology can be improved if given the opportunity to make predictive sketches (Gagnier et al., 2016), which allows them to externalize and spatially correct inaccurate mental models. Research in math education found that working with incorrect examples can be as effective as learning from correct examples (Booth et al., 2013). The current study combined these lines of research and tested which method was more effective for improving 3D geologic block diagram understanding where common spatial errors in mental models can be anticipated. Students were randomly assigned to generate predictive sketches, grade incorrect examples, or copy correct sketches. It was hypothesized that generating sketches or grading incorrect examples would improve diagram understanding compared to copying the correct answer.

A 2x3 repeated-measures ANOVA looking at the effect of activity condition on the Geologic Block Cross-Sectioning task demonstrated that performance was better at posttest than at pre-test, but there was no effect for activity condition,. There was a marginal interaction, and follow-ups revealed no difference between pre and posttest performance in the copying condition, but participants who sketched, or graded, improved from pre to post by half of a standard deviation.

A one-way ANOVA looking at number of sketches completed as a function of activity condition, revealed that participants who graded completed more sketches than participants who sketched or copied, which did not differ. A one-way ANOVA looking at time on task revealed that sketching took more time to complete than grading or copying.

This study replicated work showing sketching can improve block diagram comprehension. A new finding is that identifying and explaining common errors in incorrect examples can be as effective as generating sketches. Further, participants completed more items in less time in the incorrect examples condition. Although showing students errors may seem counter-intuitive, presenting them as known errors is an efficient and effective way to improve 3D diagram comprehension.

Spatial reasoning difficulties may impact student learning, but instruction can help overcome such difficulties. These studies explored the impact of two different forms of instruction on a particular spatial reasoning task. Study 1 involved high school students enrolled in Earth science (N = 102). Participants completed a problem-based task asking them to sketch a subsurface cross-section after receiving feedback data via written and teacher instructions (draw1: geographical map data; draw2: map and a core sample data; and draw3: map, core sample, and strike-dip data). Spatial reasoning reflected in participants' drawings were scored as: 2-dimensional (2D) reasoning when drawings only reflected surface features, 3-dimensional (3D) straight-in when drawings had vertically oriented stratified layers, and 3D horizontal for horizontal stratified layers (i.e., scientifically correct drawing). Overall change in participants' reasoning was scored as no change, weak change, or strong change. A Friedman nonparametric test revealed significant increases in participants' drawing scores between draw1, 2, and 3; χ2(2) = 26.2, p < .001. Follow up Wilcoxon signed rank tests showed significant increases in all score categories from draw1 to 2, all zs > 3.64, ps < .001; but non-significant change from draw2 to 3 in all categories.

In study 2, undergraduate students enrolled in an introductory geology course completed the same subsurface reasoning task (N = 60), but changes were instead investigated over a semester-long introductory geology course. Spatial reasoning and change in reasoning were scored in the same way as Study 1. A repeated measures MANOVA revealed a significant score increase from draw1 to draw2, Wilks' λ = .457, F(2,58) = 24.5, p < .001, η2 = .45.

Results suggest feedback using data shifted high school students spatial reasoning and an introductory geology course provided similar shifts for undergraduates. This work identifies some key aspects of learning in the context of traditional geoscience courses that can improve students' spatial reasoning.