We present a curricular approach to support student learning and data sense-making in a course-based undergraduate research experience (CURE). The CUREs focus was the socio-ecological drivers of Swift Fox (Vulpes velox) population decline and conceptual modeling underpinned instruction. Students (n=21) created 4 individual models throughout the course to help them integrate primary social and ecological literature with classroom lecture and their own research results. To assess integration of primary literature with CURE data, the individual models were coded for accuracy against an expert-generated model and a one-way repeated-measures ANOVA was used to investigate changes in model accuracy. Additionally, student group (n=4) discussions during a modeling exercise was audio-recorded, transcribed, and coded using a modified coding scheme (see Sins et al. 2005) to explore student use of the data they collected themselves. Student use of CURE data during the modeling process was characterized descriptively as the proportion of modeling time and the cognitive modeling process in which the students were employing these data.
Results/Conclusions
For model accuracy, there were significant differences across the 4 models (F(3,79)=51.15, p<0.001). Post hoc comparisons (paired t-tests and Bonferroni correction (α =0.008)) between each modeling time-point indicated that Model 1 (M=9.04, SD=4.80) was significantly different than Models 2 (M=26.81, SD=8.58, p<0.001), 3 (M=26.85, SD=11.66, p<0.001), and 4 (M=33, SD=10.31, p<0.001); but no significant difference between any other model pairs (p>0.009). Students were introduced to the primary literature between Model 1 and Model 2, which likely explains this significant increase. Students completed their CURE research between Model 3 and Model 4 and we do see an increase in student model accuracy, though not significant. This suggests that the combination of engaging with primary literature and first-hand data collection experiences helps students develop more expert-like understanding. In terms of student use of CURE data in their modeling, we saw 3 of the 4 groups actively used the CURE data, though less than 10% of the conversation time was spent modeling with these data. Additionally, when groups did talk about the CURE data, they most often did so in less sophisticated modeling processes (surface cognitive processes), such as identifying components to be included in the model, rather than reasoning about their model. Taken together, this work suggests that students benefit from the inclusion of scientific data opportunities to develop greater nuance in their understanding of systems, but may need more scaffolding to support integrating and contextualizing these data.