Reviewing the evidence on extending school time: the good, the bad and the pretty strong

It is no secret that Sir Kevan Collins’ recommendations for close to £15bn of education recovery spending were rejected by HM Treasury on the basis that the evidence for extending school time was weak and inconsistent. This ultimately led Sir Kevan to resign as Education Recovery Commissioner.

Following on from this, the government has established a review of the evidence on school time, which is expected to inform decisions in the upcoming spending review.

In this blog, I argue that the empirical evidence is by no means weak. In fact, there is strong evidence that extending school time can improve test scores if extra time is an extension of normal teaching practices and the curriculum, draws on school staff and is linked to schools existing activities in as many ways as possible.

Don’t trust international correlations

There are a range of international comparisons of the effects of extending school time, some of which are positive (Rivkin and Schiman, 2015) about the effects, whilst others downplay the effects of extra instructional time (OECD, 2018). However, there are important reasons to think international comparisons are at best, uninformative, and at worst, actively misleading.

First, one should totally discount any evidence based on international correlations between test scores and school time. This can only tell you about the effects of school time if we can hold everything else about different countries – including their socio-economic structures, education institutions and other relevant factors – constant, which is clearly not possible. Furthermore, such correlations could be biased downwards if policymakers extend school time in response to poor performance.

Second, and much more importantly, the effect of extended school time clearly depends on how the time is used and examples across countries include a whole host of different ways extra time was used. As a result, the average effect of extra time across different countries and studies often includes some very positive cases and some weak or negative examples. It is therefore more informative to look at individual examples to understand how extra time has been used well, and how it has been used badly.

The positive effects of extending school time

There are a range of studies across individual countries showing clear positive effects of extended school time:

• Italy – An after school programme led by pupils’ main teacher improved maths scores amongst disadvantaged students, but had little effect on language scores (Battistin and Meroni, 2016)
• Israel – A remedial after school programme for disadvantaged young people led by teachers improved high school graduation rates and was rapidly scaled up (Lavy and Schlosser, 2005)
• Denmark – New national regulations on primary school hours improved maths test scores, but there was no evidence of an effect on literacy scores (Jensen, 2013)
• Chile – A government-mandated longer school day led to improvements in maths and language scores (Bellei, 2009)

These positive examples share a remarkable number of similarities in their design and effects. All the programmes were led by pupils’ main teacher and/or integrated into existing school activities. The effects were also stronger for maths. This latter point is important as all the evidence on lost learning points to lower progress in maths than in literacy, which suggests that maths and numeracy skills are more sensitive to the effects of schools and teachers.

Learning from the weak and negative cases

There are also a range of examples where the effects of extended school time were relatively weak or close to zero. These are just as important to understand in order to avoid the pitfalls of past policies:

• Extended school services in the US – After school and summer school programmes in the US had little effect on achievement scores, with evidence suggesting this is because they were poorly attended, activities were rarely linked to empirical evidence and often unrelated to normal school classes (Slavin, 2021)¸ James-Burdumy et al, 2005, Henrich et al, 2010).
• After school teaching in Netherlands – Children attended after school classes at different sites with instruction based on the work of Robert Marzano. There were no significant effects on maths or language scores (Meyer and Kleveren (2013)
• Shine on Manchester – Saturday schools provided extra literacy and numeracy lessons based on music themes, but had not significant effects on pupil achievement (EEF ,2018)

These weak cases share a number of key features. Activities were often separate to existing classes and the curriculum, involved methods or activities that had not been properly trialed yet, were often run by different staff separate to the school and were sometimes run in totally different places.

This combination of positive and weak evidence therefore strongly suggests that the most successful examples of extended school time have ensured strong connections with existing school teaching, staff and the curriculum. This is confirmed by meta-analysis, which find positive effects of programmes led by teachers and based on traditional teaching methods and when they incorporate evidence-based tutoring programmes (Kidron and Lindsay, 2014, Lauer et al, 2006).

Linking to evidence on tutoring

There are also clear crossovers with the evidence on when tutoring works best, summarised in an excellent review of the evidence (Nickow et al, 2020).

• Who? – Effects are largest when led by teachers (0.5 standard deviations), though this is clearly an expensive approach and effects are still very large when tutoring is undertaking by well-trained teaching assistants and paraprofessional (0.4 standard deviations).
• When? – Effects are largest in literacy for children just starting school (0.5 standard deviations), but are consistently large for maths throughout primary school ages (0.4 standard deviations).
• How? – Larger effects are seen for programmes during the school day (0.4 standard deviations), but are still sizeable for after school programmes (0.2 standard deviations). There are slightly larger effects for longer and more frequent programmes.

Importantly, this tells us quite a lot about why tutoring might be effective. The stronger evidence on in-school programmes points to the importance of integration with existing classes and activities, and potential benefits from customisation and more immediate feedback for students. The stronger evidence on teacher-led approaches points to the importance of human capital and training. The larger effects on maths also mirror the evidence on lost learning and extended school time. The fact that extending the length of programmes only has a modest effect on outcomes suggest that creating extra time is not enough. To be effective, tutoring and extended school time needs to be delivered by staff with high levels of training and linked strongly to existing classes and teaching.

Enrichment and extra-curricular activities

Finally, there is also a body of evidence looking at the effects of extra-curricular and enrichment activities (Shlruf, 2010). Here, there is weak evidence of effects on student achievement. However, this is a somewhat unfair benchmark given that the goals of extra-curricular and enrichment activities are much wider, encompassing pupil well-being, happiness and mental health. This is one reason EPI has recommended any extra school time be used for both academic and enrichment activities, with the EEF toolkit also recommending that extended school time should incorporate both academic and enrichment activities to be successful.

The ultimate success of any extended school time will also be heavily shaped by attendance levels and framing extra time after normal school hours as being (at least partially) used for sport and the arts might end up being more appealing for pupils. There is also a clear risk that without extra time, such extra-curricular activities might get heavily squeezed out of normal school time if schools are more focused on catching up on literacy and numeracy levels.

In any case, the most effective approaches would need to learn from the successful example across the country where schools have already implemented modest extensions to the school day, including how it is framed, funded, organised and staffed.

Conclusions and summary

In conclusion, the evidence on extending school time is actually quite strong, both from the successful examples and from the less successful ones. Unsurprisingly, the effects of extending school time depends on how the time is used. It is most effective when it draws on existing and well-trained staff, integrated to existing classes and activities, based on sound empirical evidence and is more effective for maths. Providing this approach is followed, extra funding to enable extended school time is likely to yield consistent and strong returns.