Optimising the impact of a multi-intervention outreach programme on progression to higher education: recommendations for future practice and research

Abstract

Despite substantial financial commitment to widening participation activities internationally, robust evidence demonstrating 'what works' in facilitating disadvantaged learners to access Higher Education (HE) is remarkably sparse. Much effort has been directed at measuring immediate post-intervention changes in the aspirations, attitudes and behaviours thought to drive access to HE, rather than actual access itself. Here, we present an innovative quasi-experimental study of a multi-intervention outreach programme (UniConnect) consisting of 1,386 learners from the Aimhigher West Midlands database whose HE application results were known, while controlling for multiple variables, including estimates of deprivation. The results showed that any engagement with UniConnect, no matter how limited, was associated with an improved chance of achieving a place in HE, but the type of engagement, the extent of engagement and the combination of types of engagement all mattered. The more learners engaged with UniConnect, the greater were their chances of HE acceptance, but the benefit of each additional engagement beyond five or six engagements was small. To our knowledge, these findings are the first to indicate the number, type and combinations of interventions that are most effective in supporting progression to HE. These results therefore have important implications for future practice, enabling funding for such work to be used for optimal impact. Furthermore, we found large differences in success between schools, even when controlling for several other variables; a finding which has important implications for future evaluation research.

Keywords: Access and participation; Evaluation; Higher education; Multi-intervention; UniConnect; Widening participation.

Figure 1

Venn diagram showing the percentage of co-engagements of the five most common types of UniConnect activity. Percentage values below 1% have been suppressed for data protection purposes.

Figure 2

Showing the relative risk of UCAS acceptance (±standard error) by the number of UniConnect engagements and the best-fitting growth curve (±95% confidence intervals) Showing the relationship between Number of UniConnect activities, the relative Risk Score of Engagement and the expected probability of HE progression.

Figure 3

Panel a) Shows the relative risk of UCAS success for each type of UniConnect activity compared to no engagement. So, for example, the relative risk for Mentoring here refers to the overall risk associated with Mentoring regardless of whether it was engaged with alone or in combination with other activities. In contrast, Panel b) Shows the relative risk of UCAS acceptance for each of the twelve most common combinations of UniConnect activities where each relative risk compares the risk in the specified group to the risk of all other combinations. In this case, the relative risk for Mentoring refers to the risk of engaging with Mentoring and only Mentoring. Additional combinations including all combinations involving ‘Summer School’ and a miscellaneous group of combinations not otherwise included are also shown.

Figure 4

Showing the relative risk of UCAS success for learners attending each of the 36 largest schools plus a miscellaneous group of schools with fewer than ten learners. Each relative risk compares the risk of UCAS success in the specified group to the risk of all other schools combined.