Evidence for the Success of a Quantitative Assessment Instrument for Teaching Evolution in Primary Schools in England

Teaching evolution in primary schools is important, not only to form the foundation of a planned spiral curriculum but also to address alternative conceptions before they have a chance to become entrenched. However, there is little research into how best to teach evolution in primary schools, with some doubting that abstract concepts of evolution can be understood in this age group. While there are several untested proposed lessons, there is little evidence that evolution understanding in this age group can be fairly assessed in a quantitative framework. We present an assessment instrument with items selected and modified from the AAAS Project 2061 Evolution and Natural Selection test base and show that it is fit for purpose. We also report initial results, employing this tool, from a large trial (n > 1000) in 17 primary and middle schools. We show that a four-lesson scheme of work (variation, natural selection, geological timelines and homology/common ancestry) leads to increased understanding (mean increase in score ~16%) in students of all abilities. We additionally show evidence for both longer-term retention and waning of understanding. Using the successful intervention, we sought to identify parameters that might explain variation in response. We examined three student-level predictors of improvement and, after control for pre-test score, found that ability and gender (females > males) have predictive ability, while age does not. In an exploratory analysis, we considered numerous class-/school-level predictors and reported that teacher acceptance of evolution was the only factor that significantly correlated with student improvement in score (but not after Bonferroni correction). We conclude that the assessment method is fit for purpose and the intervention shows evidence of working well, including for longer-term retention. We aim to replicate these results with a second tranche of independent data