Describing students’ learning about evolution through the lens of digital concept mapping

Background

Evolution education is a central tenet of biology education in school, yet the topic is conceptually complex and students’ understanding is fraught with misconceptions. Learning Progression Analytics (LPA) aims to trace students’ conceptual development along established learning progressions. For this purpose, data from students’ interactions with tasks in digital learning environments are analysed. This is done with the intention to make conceptual change and knowledge integration processes accessible to teachers for formative assessment and feedback. One assessment strategy that can mirror such processes in conceptually complex topics is concept mapping. This study presents an initial attempt to analyse concept maps that students created over the course of a digital teaching unit on factors of evolution (i.e., mutation, natural and sexual selection, genetic drift, gene flow). Our aim was to determine which metrics could be suitable for the use in LPA.

Method

We collected data from 250 high school students who participated in a hybrid teaching unit on five factors of evolution in the school year 2022/23. Students completed a pre- and posttest and created a total of five concept maps over the course of the unit, repeatedly revising and reworking their previous maps. We split the students into three groups based on their gain from pre- to posttest and analysed their maps for differences (1) between the different measurement points and (2) between the groups at each measurement point regarding (a) their similarity to expert concept maps, (b) concept scores, and (c) different network metrics.

Results

We found significant differences between most of the consecutive measurement points for all calculated metrics (e.g., number of nodes and links, concept scores) across the sample. We found significant differences between the three groups for the average degree and number of edges at two measurement points.

Conclusions

From our results, we conclude that the most promising metrics from our study for the use of concept maps in LPA are those focusing on the connections (i.e., average degree and number of edges). Further research is needed to refine these assessments in controlled environments and determine their value for automated assessment and feedback more definitely.