Article Critique Sample: Critique of Research on Homework and Attainment

Model Answer

Overview of the Article

This critique evaluates Marsden and Iqbal (2020), a quasi-experimental study titled ‘Homework Frequency and Attainment in Key Stage 4 Mathematics: A Quasi-Experimental Study’, published in the British Journal of Educational Practice. The researchers compared GCSE mathematics attainment between pupils in six secondary schools assigned either a high-frequency (three sessions weekly) or low-frequency (one session weekly) homework policy over one academic year, while holding teaching staff and curriculum content constant within each school. Attainment was measured using standardised end-of-year assessment scores, and prior attainment (Key Stage 2 results) was used as a covariate. The authors reported that pupils in the high-frequency homework group achieved significantly higher standardised scores than those in the low-frequency group, after controlling for prior attainment, and concluded that increasing homework frequency is an effective, low-cost strategy for raising mathematics attainment at Key Stage 4.

Aims and Rationale

The research aim — to test whether homework frequency affects Key Stage 4 mathematics attainment — is clearly stated and educationally significant, given ongoing policy debate about homework’s value and the workload implications for both pupils and teachers. The rationale is reasonably well grounded in existing literature on distributed practice and self-regulated learning (Radcliffe and Osei, 2019), and the authors appropriately note inconsistent findings from earlier correlational studies as justification for a more controlled comparison. The use of a quasi-experimental design, exploiting naturally occurring variation in school homework policy rather than random allocation, is a pragmatic and ethically sensible choice, since randomly assigning pupils to receive less homework would raise practical and ethical concerns in a real school setting. However, the introduction does not fully engage with confounding factors that commonly differ between schools with different homework policies, such as socioeconomic intake or overall school effectiveness, which weakens the strength of the causal framing adopted later in the paper.

Methodology

The study used a quasi-experimental, between-schools design comparing six secondary schools: three with a high-frequency homework policy and three with a low-frequency policy, involving 612 Year 11 pupils in total. Mathematics attainment was measured using a standardised end-of-year assessment common across all six schools, and Key Stage 2 prior attainment was included as a covariate in an analysis of covariance (ANCOVA), an appropriate statistical approach for adjusting for pre-existing differences between groups that were not randomly assigned (Field, 2018). Using an established, moderated national assessment as the outcome measure, rather than a bespoke test, strengthens the validity and comparability of results across schools. Ethical approval was obtained from the researchers’ university ethics committee, and schools opted in voluntarily with parental information provided, consistent with British Educational Research Association (2018) guidelines. A significant methodological limitation is that homework policy was determined at school level rather than pupil level, meaning pupils were not individually randomised, and school-level clustering (for example, differences in teaching quality or school ethos between the six schools) is not modelled using multilevel or hierarchical techniques (Pemberton and Hollingworth, 2020), despite these being well-established methods for handling nested data of this kind. A structured appraisal using a framework such as the Education Endowment Foundation’s (2021) padlock rating system would likely flag this clustering issue as a notable security-of-evidence concern, even though the covariate adjustment for prior attainment is a genuine methodological strength relative to simple pre-post comparisons. This omission means unmeasured school-level differences remain a plausible alternative explanation for the attainment gap observed between the two homework-frequency groups.

Results and Analysis

The ANCOVA revealed a statistically significant difference in adjusted mean attainment scores between the high-frequency and low-frequency homework groups, favouring the high-frequency group, with a small-to-moderate effect size (partial eta squared = .06) after controlling for prior attainment. The statistical approach is appropriate for the covariate-adjusted comparison intended, and effect sizes are helpfully reported alongside significance values, supporting transparent interpretation of the practical, not just statistical, significance of the finding. However, because homework frequency was assigned at school level rather than pupil level, the individual pupil is not truly the unit of random assignment, and analysing 612 individual pupils without accounting for their clustering within just six schools risks underestimating the true standard error, potentially inflating the apparent statistical significance of the group difference. The authors use causal language (‘increasing homework frequency raises attainment’) that a quasi-experimental, non-randomised, school-level design cannot fully support, since any number of unmeasured school-level factors correlated with homework policy could equally explain the observed difference.

Discussion and Conclusions

The authors interpret their findings as evidence that increasing homework frequency is a practical, low-cost lever for improving Key Stage 4 mathematics attainment, and they connect this appropriately to distributed-practice theory (Whitcombe, 2017), which predicts that more frequent, shorter practice sessions support stronger retention than infrequent, longer ones. This theoretical linkage is a strength of the discussion. However, the conclusion’s confident policy recommendation — that schools should increase homework frequency to raise attainment — sits awkwardly against a design that cannot rule out school-level confounding, and a more cautious conclusion acknowledging this as an association pending further, individually randomised or multilevel research would have better matched the evidence’s actual strength. The discussion also does not address potential negative consequences of increased homework frequency, such as pupil wellbeing or motivation, which a balanced evidence-informed policy discussion in education research would typically be expected to weigh alongside attainment gains.

Strengths and Limitations

This study’s strengths include its use of a standardised, externally moderated attainment measure, inclusion of prior attainment as a covariate, a reasonably large combined pupil sample, and transparent reporting of effect sizes alongside significance values. Its quasi-experimental design is also a pragmatic and ethically appropriate response to the practical impossibility of randomising individual pupils to different homework regimes within the same school. Set against these strengths, the school-level, rather than pupil-level, assignment of homework policy is the study’s most significant limitation, since it leaves school-level confounders such as socioeconomic intake, teaching quality, and general school effectiveness unaddressed, any of which could independently explain differences in attainment between the two groups of schools. The absence of multilevel modelling to account for pupils being nested within schools further weakens confidence in the reported significance level. It is also worth noting that homework completion rates are not reported for either group, so it remains unclear whether the low-frequency group actually completed less homework in practice, or whether compliance varied enough within groups to blur the intended policy distinction. The study is also limited to a single subject, mathematics, and a single year group, meaning findings should not be generalised to other subjects or key stages without further research specifically examining those contexts.

Overall Judgement

Overall, this is a useful and practically motivated study that adds to the evidence base on homework frequency, using a sensible and ethically appropriate quasi-experimental design given the constraints of a real school setting. Its principal weakness is the mismatch between the strength of its causal conclusions and what a non-randomised, school-level design can support, given the plausible influence of unmeasured school-level confounders. For education practitioners and policymakers, the findings are best treated as promising but preliminary evidence for the value of frequent homework in mathematics, warranting replication using individually randomised or multilevel designs before informing whole-school homework policy with confidence.

References

British Educational Research Association (2018) Ethical Guidelines for Educational Research. 4th edn. London: BERA.

Education Endowment Foundation (2021) Homework: Teaching and Learning Toolkit. London: EEF.

Field, A. (2018) Discovering Statistics Using IBM SPSS Statistics. 5th edn. London: Sage.

Marsden, K. and Iqbal, S. (2020) ‘Homework frequency and attainment in Key Stage 4 mathematics: a quasi-experimental study’, British Journal of Educational Practice, 46(3), pp. 289–307.

Pemberton, R. and Hollingworth, D. (2020) ‘Multilevel modelling in school effectiveness research: a practical guide’, Journal of Educational Measurement and Methods, 8(2), pp. 55–71.

Radcliffe, S. and Osei, B. (2019) ‘Homework and academic achievement: a research synthesis’, Review of Education Practice, 15(1), pp. 1–24.

Whitcombe, R. (2017) ‘Distributed practice and self-regulated learning in secondary education’, Educational Psychology Review Digest, 10(4), pp. 201–218.