ABSTRACT
Implementation fidelity or integrity is a key variable in evidence-based interventions, but is rarely assessed by student response or over a long period. This study related implementation fidelity in independent book reading to tested reading achievement over a year. The sample of 852,295 students in 3243 primary and secondary schools was reduced by schools providing incomplete data and the discarding of mid-year data. Achievement was measured pre-post by STAR Reading, a computerised item-banked adaptive test of reading accuracy and comprehension. Implementation fidelity was measured by variables from the computerised Accelerated Reader software, which analyses comprehension of a real book the student has chosen by a quiz. Results compared key variables for the four regions of the UK, local authorities and best performing schools. Computerised methods offer an interesting alternative to teacher behaviour for investigating implementation and outcomes. This study showed stronger links between implementation and outcomes at an individual level than when all data are aggregated, and that student response is at least as good an implementation index as teacher behaviour. The implications for practice, policy and future research were outlined.