hIt is a complaint I sometimes make on this blog or on stage: quite a lot has already been written and researched about homework. Often, the discussion revolves around a fairly simple question: does homework work? Or, sometimes, a slightly more concrete one: how much homework should we assign, from what age does it make sense, and how much time should students spend on it?
A new meta-analysis by Jianzhong Xu and colleagues in Educational Research Review takes a different approach. Rather than looking primarily at how much homework students receive, the researchers examined what teachers do with it. How well do teachers design the assignments they give their students? How are students supported? And how good is the feedback? And how does all of this relate to motivation, behaviour, and academic performance?
29 studies and more than 35,000 students
The researchers brought together 29 independent studies involving a total of 35,702 students. Because individual studies often examined multiple relationships, this meta-analysis included no fewer than 506 effect sizes. The researchers used a so-called three-level meta-analysis, which accounts for the fact that different effect sizes from the same study are not independent of one another.
The team distinguished six ways in which teachers can be involved with homework:
- the quality of the homework itself,
- autonomy support,
- content support,
- the quality of feedback,
- the amount of feedback,
- and, finally, a combination of different forms of support.
Overall, they found a small-to-moderate positive association between teacher involvement and various student outcomes (r = .246). Moreover, this result appears to be quite robust. When the researchers reran their analyses in different ways or removed one study at a time, the result hardly changed. But there is more!
Not all involvement is equal
There are considerable differences in how teachers handle homework. The researchers found the strongest association for the quality of the homework itself (r = .357). This was followed by feedback quality (r = .321), combined support (r = .298), and autonomy support (r = .290). Content support came in at r = .217. The amount of feedback was lower, at r = .157.
In the robustness analyses, the differences with the amount of feedback were clearest for homework quality, feedback quality, and autonomy support. The evidence was weaker for some of the other comparisons. This is in part because only a small number of independent studies were available.
So we cannot conclude from this study that simply being more involved with homework is better by definition. The researchers describe it rather in terms of purpose. What a teacher does should be both pedagogically meaningful and instructionally coherent. Assigning homework is one thing. Assigning good homework is apparently something else. This may not sound that earth-shattering, but wait…
So what does ‘good homework’ mean?
The researchers describe high-quality homework as assignments that:
- are purposeful,
- clearly connect to what happens in the classroom,
- and focus on understanding rather than simply completing a task.
In this way, well-designed homework can give students opportunities to further practise, apply, or extend what they have learned in class. This is in line with much previous research on homework.
We see something similar with feedback. The amount of feedback was less strongly associated with outcomes; what seems to matter more is whether that feedback is useful and informative. This is also consistent with what we know from broader research on feedback. More feedback is not necessarily better feedback.
More than academic performance
The researchers did not look solely at academic performance. They also included motivation, interest, effort, engagement, self-regulation, homework completion, and procrastination. The association between teacher involvement and academic achievement was r = .169. Positive, in other words, but relatively modest.
For autonomous motivation, on the other hand, the association was r = .433. For interest in homework, it was r = .312. The associations were also higher for engagement (r = .279) and effort (r = .273) than for academic achievement. Greater teacher involvement also correlated with less procrastination (r = −.162).
The robustness analyses made the pattern even clearer: teacher involvement was more strongly associated with autonomous motivation and homework interest than with academic achievement. The evidence for other outcomes is more limited. So how teachers design and follow up on homework may shape how students experience and approach it even more than how well they perform academically.
And, just to reiterate for anyone who has missed it so far: these are correlations! The relationship can also work in the other direction. Students who are more motivated or perform better may, for example, elicit different behaviour from their teachers. The authors themselves explicitly point to the possibility of such reciprocal relationships. They therefore call for more longitudinal and experimental research to give us a better understanding of cause and effect.