Reading proficiency is essential for developing critical thinking and for pupils' academic success across all subjects. Globally, millions of pupils struggle with basic reading skills, with international assessments revealing persistent gaps in reading achievement (OECD, 2019). In China, although pupils perform strongly in assessments such as PISA and PIRLS, these tests primarily assess reading in Chinese and are administered only in economically advanced regions like Shanghai (OECD, 2020), limiting generalisability to broader populations. For English reading, which has been compulsory in Chinese primary education for over two decades, little robust evidence exists regarding pupils' proficiency or effective instructional approaches.

Following China's 2022 English Curriculum Standards, greater emphasis has been placed on reading proficiency and pupil autonomy. However, persistent challenges remain, including limited access to suitable reading materials, ineffective pedagogical approaches, and regional disparities in resources. To address these issues, substantial government investment through initiatives such as the Education Informatisation 2.0 Action Plan (Ministry of Education, 2018) has led schools to increasingly adopt computerised reading programmes. The COVID-19 pandemic further accelerated this digital adoption. Yet many programmes lack independent rigorous evaluation despite strong claims by developers (See & Gorard, 2020), posing risks when schools and parents adopt them uncritically.

One widely used programme is Accelerated Reader (AR), deployed in over 50,000 schools across 90 countries, including more than 800 schools in China (Renaissance, 2026). AR is a computerised supplementary reading programme providing guided reading instruction and formative feedback through comprehension quizzes. The theoretical framework underpinning AR is Vygotsky's (1978) zone of proximal development (ZPD), which posits that learning occurs most effectively when instruction is pitched slightly above a learner's current level but within reach with appropriate support. AR operationalises this theory by matching pupils to books at their individual reading levels and providing immediate feedback through quizzes, enabling pupils to progress through increasingly challenging texts. Cambridge Reading Adventures (CRA), a freely accessible alternative offering comparable functions with digital levelled reading materials, serves as a cost-effective comparison in this study.

Despite the large number of schools using AR in China, no independent research has been conducted to test its effectiveness among Chinese pupils. Most existing studies were conducted in the US and UK (Gorard et al., 2015; What Works Clearinghouse, 2016), with no rigorous evaluations in non-English speaking countries, limiting generalisability to contexts such as China where English is taught as a foreign language. The purpose of this research is to investigate the impact of AR and CRA on Chinese primary school pupils' English reading achievement and reading behaviour (defined as reading habits and reading attitudes). The main research questions are: (1) What impact do AR and CRA have on Chinese school pupils' reading performance? (2) What is the impact of AR and CRA on Chinese school pupils' reading behaviour (reading attitudes and habits)? (3) How much of the variation in reading performance can be explained by reading behaviours? 

This research contributes to the international evidence base on computerised reading programmes in non-English speaking contexts. By comparing a commercial programme with a freely available alternative, the findings offer insights relevant to educational policymakers across Europe and beyond who face decisions about technology investments in linguistically diverse classrooms. The study has particular relevance for contexts where English is taught as a foreign language and where resource constraints necessitate cost-effective solutions.

This study employed a cluster randomised controlled trial conducted in Chinese public primary schools across four provinces. The main study involved 1,059 pupils in Years 4 to 6 from 21 classes. Ten classes with 469 pupils were randomly allocated to receive the intervention, while 11 classes with 590 pupils served as business as usual controls. An additional non-randomised class with 54 pupils was included to strengthen sample size. Random allocation was conducted at the class level to minimise contamination between treatment and control conditions. The intervention lasted 12 to 13 weeks, during which treatment pupils used either AR or CRA during designated reading sessions. Reading performance was assessed using pre and post tests adapted from the Cambridge English A1 Movers examination (Cambridge English Qualifications, 2023) and the Australian National Assessment Program Literacy and Numeracy reading tests (National Assessment Program, 2023). Using standardised instruments independent of AR's own assessment tools addressed a key methodological limitation identified in previous studies, where measurement bias from programme specific tests compromised validity (Gorard et al., 2015). Reading behaviours were measured through validated questionnaire items: reading habits were assessed using items adapted from PISA 2018 (OECD, 2019) and PIRLS 2021 (Mullis & Martin, 2019), while reading attitudes were measured using the Heathington Attitude Scale (Heathington, 1975). Alongside the impact evaluation, a process evaluation was conducted through classroom observations and semi-structured interviews with teachers and pupils. This component assessed implementation fidelity, identified conditions essential for programme success, and explored mechanisms of change. Intent to treat analysis was employed, with effect sizes calculated using Hedges' g to account for small sample bias. Subgroup analyses examined differential effects by prior attainment, school type, ethnicity, and family socioeconomic status. Compliance analysis assessed the relationship between quiz completion and outcomes. Path analysis explored direct and indirect effects of AR on reading performance via reading behaviours.

The findings indicate that both AR and CRA improved Chinese primary pupils' English reading performance compared to business as usual instruction. AR produced an effect size of +0.22, while CRA showed a larger effect of +0.40, though the latter should be interpreted cautiously due to baseline imbalances between groups. These results add to the existing evidence that AR has a positive impact on students' English reading outcomes in English speaking countries (Gorard et al., 2015; What Works Clearinghouse, 2016), while extending this evidence to non-English speaking contexts. Both programmes enhanced pupils' reading habits. However, results for reading attitudes were mixed: AR showed positive effects, while CRA appeared to negatively affect attitudes despite improving reading enjoyment and willingness to read during free time. Compliance analysis demonstrated that programme fidelity matters: pupils completing more quizzes achieved higher reading scores. Path analysis suggested that AR had a direct positive association with reading performance, independent of behavioural pathways, implying that structured comprehension activities may directly enhance attainment. The process evaluation identified essential implementation conditions including adequate teacher preparation, appropriate reading levels, protected reading time, sufficient book availability, and reliable technical infrastructure. These findings offer cautious optimism for computerised reading programmes in EFL contexts whilst suggesting that core pedagogical elements, including appropriate reading levels, pupil choice of reading materials, sufficient independent reading time, and timely teacher feedback, may play a more central role in programme effectiveness than the delivery format itself. For resource constrained contexts common across Europe and beyond, programmes like CRA present a viable free alternative. Future research should examine longer term effects and conduct component analysis to identify which mechanisms drive programme effectiveness.

Cambridge English Qualifications. (2023). A1 Movers preparation: Sample tests. https://www.cambridgeenglish.org/exams-and-tests/movers/preparation/ Gorard, S., Siddiqui, N., & See, B. H. (2015). Accelerated Reader: Evaluation report and executive summary. Education Endowment Foundation. Heathington, B. S. (1975). The development of scales to measure attitudes towards reading [Unpublished doctoral dissertation]. University of Tennessee. Ministry of Education of the People's Republic of China. (2018). Education Informatisation 2.0 Action Plan. http://www.moe.gov.cn/srcsite/A16/s3342/201804/t20180425_334188.html Mullis, I. V. S., & Martin, M. O. (Eds.). (2019). PIRLS 2021 assessment frameworks. TIMSS & PIRLS International Study Center. National Assessment Program. (2023). NAPLAN practice tests. https://www.nap.edu.au/ OECD. (2019). PISA 2018: Insights and interpretations. OECD Publishing. OECD. (2020). Benchmarking the performance of China's education system. OECD Publishing. https://doi.org/10.1787/4ab33702-en Renaissance Learning. (2026). About us. https://renaissance.cn/ See, B. H., & Gorard, S. (2020). Effective classroom instructions for primary literacy: A critical review of the causal evidence. International Journal of Educational Research, 102, 101577. https://doi.org/10.1016/j.ijer.2020.101577 Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press. What Works Clearinghouse. (2016). Accelerated Reader intervention report. Institute of Education Sciences.