Via an article in The Guardian I found an interesting research paper by Richard Murphy and Felix Weinhardt for which they asked 15,000 pupils to rate themselves in English, maths and science, and compared their rankings with their exam results. They did this at  stage 2, when the pupils finish their primary education at age 11, and at key stage 3, when the pupils have turned 14 and are in secondary education. In their research Murphy and Weinhardt took into account factors such as parental involvement and concluded that “non-cognitive skills such as confidence, perseverance and resilience have large effects on achievement”.

From the Guardian:

It has long been one of education’s central tenets: pupils benefit from being around children who are brighter than them. Competition in the classroom raises standards across the board, so the argument goes, because the example of the brightest pupils will rub off on the rest of their class.

But a new study, which tracks the exam performance of more than two million pupils in England, suggests this view fails to tell the whole story. Coming up against higher-achieving peers could have a damaging impact on the confidence of those pupils who believe they are in the bottom half of their class, the study concludes.

Abstract of the research paper that you can download here:

We find an individual’s rank within their reference group has effects on later objective outcomes. To evaluate the impact of local rank, we use a large administrative dataset tracking over two million students in England from primary through to secondary school. Academic rank within primary school has sizable, robust and significant effects on later achievement in secondary school, conditional on national test scores. Moreover we find boys gain four times more in later test scores from being top compared to girls. We provide evidence for a mechanism using matched survey data, which shows that rank affects an individual’s self-concept. The paper discusses other potential channels but concludes that malleable non-cognitive skills such as confidence and belief in own ability are most likely to generate these results. We put forward a basic model where rank effects costs and effort allocation when faced with multiple tasks. We believe this is the first large-scale study to show large and robust effects of rank position on objective outcomes of that have consequences in the labour market.