There's homework, and there's homework. When I was in school most of it was busy work. In college though, most of my problem sets were actually a good way to practice things.
I wonder if parents could just say to their kid that they didn't have to do any elementary school homework. Would a kid who otherwise performs well get away with not doing any homework? It would probably depend on the school.
Let's just continue with something of unknown utility. Why not?
Schools went overboard on homework for common core. It sucks. Most of it is poorly designed. It seems more designed to indoctrinate students to be complaint than to learn.
As a parent dealing with schools for almost 20 years, learning doesn't seem to matter anymore.
The purpose of homework isn’t learning. The purpose of homework is to sort those who will conscientiously put the new cover sheets on the TPS reports from those who won’t. It’s to filter for conscientious and agreeable personality traits as well as work ethic.
Kohn is not "[taking] the studies seriously anyway." He argues that the studies are, as you wrote, inconclusive at best and that one should recognize the demonstrable harms of homework and limit its use (seeing as it's not particularly compelling that it does much). It is easy to find compelling articles on the harms, including physical health, and it extends to even competitive environments where it actually negatively impacts learning (but helps with the educational rat race). e.g. https://www.tandfonline.com/doi/abs/10.1080/00220973.2012.745469
I would hazard a guess that homework vs proficiency would produce the exact same results in elementary school. If you throw more time at a skill the better you get, that seems pretty straight forward.
Practicing piano is another good example they should study. If you don’t practice every day, your progress would be limited. And the age works as well because the most amount of kids in piano lessons is likely elementary school. If they don’t have the drive to practice (such as myself) eventually they will quit.
The same goes for sports. Starting early and spending more time practicing will produce oversized results.
However if you graph enjoyment vs homework, I’m sure the graph would be inverted but that’s the whole point of parenting, isn’t it? To force your kids to learn skills they otherwise wouldn’t on their own?
One of my friends was essentially neglected as a child and spent most of the 80s and 90s watching TV. He’s in his 50s now and has no skills except for being a substitute teacher. I love him to death but I would never let him teach my child anything. He’s basically the poster child for forcing your kids to do homework.
Another potential confounder is that incompetent teachers almost certainly will either assign far too much or far too little homework. My own homework requirements were much more reasonable than those of my friend's kids, for example.
Speaking personally, I found it useful in some subjects, not useful in others. It depends on what you are given and how much of it there is to do. I think it made a lot more sense in the context of classes where they said "OK, we have this much work to do, and it's due tomorrow." If you finished in class, you finished in class - nothing wrong with that.
From a quick glance at the Nawaz and Welbourne study, it seems that they didn't look at performance in classes other than the algebra class.
I take the default model to be that with no required homework, students will nevertheless spend some time at home trying to keep up with the material in all of their classes. When a large amount of homework is required for one class, they will do more work for that class, and do better in that class, but will do less work for other classes (there's only so much time in a day), and do worse in those classes, for zero (or negative) net benefit. If you don't measure performance in other classes, you have no evidence against this default model.
Even if the total time spent at home on school work goes up as a result of requiring homework, and overall performance in all classes improves, it would not, of course, follow that this was a net gain. Kids have other important uses of their time, which may well include work on projects unrelated to the classes they are taking, but which would be considered "academic" work if it happened to have been assigned in class (eg, reading novels).
It’s so sad that we have almost no good data on this, despite the massive amount of time spent on it. My prior is that if it’s a skill (math, writing, analyzing arguments, probably reading), then everything else being equal time spent practicing it makes you better. The biggest confound is that a lot of homework is busywork, at least in elementary school.
It seems like edtech should help us get more data points like the algebra study more easily. Hope there’s more data out there that we just don’t know about yet.
Wait, this is really all the literature??? We have education departments at every university and this seems like a really easy thing to do randomized studies on.
I've had a teacher friend tell me that homework in elementary school is useless. But as someone who spends an hour a night helping a kindergartener and first grader with homework, I'll be annoyed if it doesn't help.
I don't doubt that the material is better ingrained in their minds having done homework. However, whether that has lasting importance is another question. They'll eventually learn reading, writing, and arithmetic with or without homework.
If I were doing this I might is a hot control or comparison with tutoring. Most tutoring and time spent with a tutor is essentially 'doing homework' and/or working through sets of problems. Especially if one were to look at tutoring centres vs scores...this could be a good approximation of how much 'practice makes perfect'. Though you'd be hard pressed to not be swamped with bias fo who goes to tutoring. But with a bit of hand waving statistics one could compare similar SES kids who get tutoring vs those who don't to see if practices helps.
Another way, which I'm surprised to see is missing, is to look at homework completion rates and scores. Few teachers bother to do this consistently, but there are some who collect and mark all homework as part of a homework score. This happens a lot more in Maths or Science classes than in something like english or history but can be found all over if homework is a formal part of their summative grade.
I found it was a good opener and I'd have a 'do now' problem at the start of class and/or I'd have them do a quick peer marking to swap their homework and mark each other based on answers I'd put on the board (with some QC from me). Then I'd collect that to have them do the marking for me and get the valuable experience of engaging in marking. This sort of actual homework submission is a decent proxy for 'practice' and better than time spent. If a kid spends 5 minutes per problem or 1 minute per problem because they are smarter/already know it etc. who cares? The main thing is the number of practice problems they completed, in my view.
And they would be a good method to compare if completing x% of practice sets with x% accuracy helps or not using their marked homework submissions. So yea I'd think to look at homework completion rates of a given band of high/middle/low achievers and their homework completion rates to show effort and the scores they get in that homework to show mastery.
e.g. If two students gets a 60/100 on their prelim test at the start of the school year, then one does 80% of their homework with 90% accuracy and the other does only 20% of their homework with 90% accuracy....what will their final exam results be? Will that extra practice by the 80% homework effort kid result in a higher score than the other kid who clearly didn't practice as much?
Time spent is always going to be fraught with incredible inaccuracies akin to people self reporting their dietary/caloric intakes...only worse since it relies on children to report timespans.
Math teacher here. Homework is unquestionably good for math proficiency for many students if it is the right kind and there is the right amount of it, but the problem is that some students might need 10 problems while others might need 30 so there is a danger of assigning too much for some and too little for others. My proposal is to offer a large number of homework problems, not count homework performance for the students’ grades, but guarantee that X% of the problems on the test will literally be from the homework, where the ideal value of X may vary depending on the class and subject.
Seems obvious that practicing anything makes you better at it. That homework may be so poorly designed that doing it does not lead to improvement is possible but seems a different issue -- e.g. quality of teachers, of educational materials.
Personal anecdata: I did very well in almost every class that assigned homework (I have always been diligent about doing what was assigned). I did quite poorly in almost every class where homework was not assigned. This pattern repeated from middle school, through high school, undergrad and later grad school.
I know others whose results were completely uncorrelated with the amount of homework assigned and/or performed.
My guess is that the distribution is so wide, individual differences are drowned out by the inter-person variations. Or worse, the distribution might be multi-modal.
Homework is good if/when it gives children a chance to practice concepts they have not mastered. For a study to show homework to be helpful, it would need to take into account when students master material. I have four very different kids, and can not make grand sweeping statements about homework in general, but it has always been useful for each kid to practice skills until mastered.
I agree that the literature is a mess, and its messiness is frequently mentioned by researchers. I think I ended up feeling happier about the non-experimental studies than you did -- check out those Trautwein papers, I think they're starting to figure out where to look for this stuff.
Here was my conclusion: "For the youngest kids, homework matters a lot to parents but probably has less of an impact on achievement. Parents are often helpful, but sometimes it can turn into a negative experience at home. To handle that, teachers can assign higher quality assignments that won’t trigger negative interactions. Sounds tough, but OK.
For adolescents, classes that assign homework frequently are associated with more learning. But not classes that ask for more time on homework. And not classes that assign more homework that students think is more challenging. But yes classes where students think the homework is interesting or well-selected. And students reporting they put more effort into homework is associated with learning.
It sounds like teachers need to thread the needle for teens. They need to frequently assign homework that is purposeful – it can’t be busy work – but it shouldn’t be too hard. And there’s a possibility that homework isn’t really as much about practice as it is about taking the class seriously, feeling in control, taking responsibility for their own learning."
Basically: homework will be a net gain if kids actually do it. But when you're dealing with adolescents, realistically there are a lot of ways for them to get away with not actually doing it. (A lot.) Completion rates in most places aren't terrific. So the first question is, how do you get kids to do it and do it seriously? And the second is, what are the benefits of that sort of practice? I think research (along with common sense) supports the idea that there are on average benefits to kids when they do it.
For early elementary reading and math, Slavin & Madden at Johns Hopkins developed “Success for All” and.. they say there is research… maybe it’s hiding. Also it is more than a homework program, it’s class work interlocking with homework in a specific way and classrooms divided by level rather than grade. So it is not directly comparable to “homework only” but it might be something to look at.
Even if we were to accept that "homework," as some kind of undifferentiated treatment variable, has some net positive benefit, treating homework as an undifferentiated variable seems pretty useless from a pedagogical perspective. What kind of homework? For what kids? Going to what schools? Under whar kinds of conditions? As compared to what other type of intervention variable?
Between the ages of about 8-12 I had assigned homework that amounted to about an hour a week. I found it too easy and boring, just like most other schoolwork. But at that age I was messy, disorganised, terrible at time management, and good at coming up with excuses to get out of things I did not want to do. Because these things were challenges for me, it was good for me to have to a) remember to take it home, b) keep the pages clean and unwrinkled, c) initiate and complete a task I found boring and pointless, d) do this on a deadline, and e) remember to return it. My parents and teacher let me choose my own strategies to manage this, and then asked me to reflect on whether they worked - and this actually helped considerably with my organisational and task management skills, moreso than imposition of these strategies would have done. I also find it interesting that I can pinpoint things that are still issues for me (task initiation, leaving things to the last minute, losing items) back as far as early primary school.
From my memories of school, the quality and usefulness of homework varied wildly, even within the same subject ("math") and even with the same professor. My naive guess would be that repetition is differently-useful for different lessons in a way that resists sweeping generalizations, to say nothing of different students learning differently.
That said, most of my homework before the university level was crap.
On the plateau model: mathematics tends to be taught as a series of plateaus (plateaux?). You are introduced to a concept like "multiplication" or "sine function" or "Lebesgue integration" and then you work through a zillion examples until it's been thoroughly beaten into your head; you reach the plateau where you finally understand what a sine function is. Then you can move on to the next bit of mathematics, which assumes you understand the previous bit.
Unfortunately the number of examples you're obliged to work through is some constant, insufficient for the dumb kids and excessive for the smart kids. So the smart kids are bored doing the same thing over and over again while the dumb kids never quite reach the plateau and get left behind.
Some other subjects are unlike this. You don't need to understand the history of Ancient Greece before you can understand the history of Ancient Rome. Furthermore, while you can reach the "sine function" plateau pretty easily, no high school student will reach the "Ancient Greek History" plateau. So you can spend an unlimited number of hours studying ancient history, if you like, understanding it more and more deeply, although possibly not in any way that can show up on a high school exam.
My primary experience of homework is that it contributes to my performance in classes if it's one of two categories:
1. homework that strongly resembles material on the test
2. homework that covers information that won't come up in class
Type 1 homework is a waste of time for every purpose except improving my GPA.
Type 2 homework is actively useful (obviously, because I wouldn't have learned that information otherwise). This has been true for my entire life - I remember independent learning assignments as early as first or second grade, and I've always found them more useful than lecture-based instruction. My capacity for self-teaching has improved since I was six, and I can go longer without support/feedback now, but the basic principle is still the same. This is probably less useful if you have unmedicated ADHD or otehrwise aren't much of an independent learner.
(Unfortunately, type 2 homework tends to correlate with useless busywork during class time. The most notable example I've run into was high school AP bio, which was a "flipped class". 100% of my time outside of class - mostly reading the textbook - was useful; 90% of my time in class was spent doing repetitive worksheets or listening to the teacher repeat the textbook in smaller words. Then again, I probably only did 30% of the assigned homework. It was a weird year.)
I guess there's a third category, which is homework with the goal of improving fluency/speed at some task. Teachers love to give this kind of homework, but they're often remarkably bad at designing it. IMO, this sort of homework should include lots of very short problems, instant feedback, and enough surprises that making an excel spreadsheet/python program that solves the problem for you is more time and effort than doing it by hand. Also, my experience this literally only works for symbol manipulation problems (eg the kind of math that engineers do/basic proofs/sudoku). I have experienced many painful homework assignments that attempt to turn interesting real-world problems into fluency drills - this is both painful and probably actively detrimental to learning.
Is the batch of no homework students pulling down the average that much in the final study cited? I'll gladly take the side of "some math homework is useful, and more is technically better on average, but less still represents better than average progress" if it isn't. Less is definitely more of my time back.
Challenge: imagining arbitrary resources were available, but no more cooperation from teachers, students and parents than you'd normally expect, how would you design a study to effectively look at this? How would you control for teaching styles/techniques/skill? Circumstances in student homes? Quality of homework curriculum? Alignment of homework type with subject matter (Algebra isn't Geography isn't writing)? How do you isolate variables we can't agree how to measure?
We struggle to assess teacher quality in general, or genuine variation in student ability. In that context, figuring out how effective homework is seems like determining how good a chisel is without knowing the skill of the sculptor or the quality of the marble.
The post already mentions all of these considerations, but let me come out with an anecdote.
I was a somewhat gifted student, finishing comprehensive school (9th grade) with the highest possible grades (outside of P.E. and art classes, anyhow) even though the last time I did any work whatsoever beyond showing up was during the first grade - before I figured out I didn't need to do any. I continued on this route in high school and while I maintained top grades in some subjects, my grades in mathematics, physics and chemistry tanked (more specifically, calculus was my breaking point and upon being introduced to the topic, I received a failed grade), and I dropped out of school. I've since continued studying in university and I'm receiving good grades, but this time I'm doing the bare minimum and subjectively, I would expect to fail most courses if I didn't do that much.
What I gather from my experience is the following: Homework is useful, but only up to a point dependent on the individual and the topic. For me, at least as far as the granularity of school grading system goes, there was no use up until high school, but at high school level failure to do any work was sufficient to drop the math grades from excellent to failed. Presumably, other students might find homework to be the difference between success and failure at an earlier or later stage than I did. But perhaps more crucially, it has the additional effect of building up work ethic: to this day I find myself unable to cope with a workload of doing more than half of the number of courses expected of students, and I can only assume this is because I never did any work prior to university. This alone seems like a good justification to have mandatory homework, although it would be better still if homework wasn't seen by some students as arbitrary punishments handed out in child prisons, but gifted students could skip grades to the point where doing homework is actually required of them.
The way the problem is posed (like so many of these sorts of reviews) is problematic.
What matters is not just how well homework does (or does not work) cross some aggregate of all children, but what works optimally, ideally for each child and, when that is not practical, for whatever subgroups are easily isolatable.
I would say, for example, that my experience (unlike Scott's) is that I learned a lot from homework, more so than I would have from whatever the alternative on offer is (either "just listen in class" or "read the chapters of the textbook"). Some of that may be my idiosyncratic personality and "learning style", some may be my school, but *I* would be very reluctant to throw out homework until I'm presented with what looks (to me) like a better alternative.
Certainly some (not all, not most, but some) of that homework was a waste of time insofar as I learned things I didn't much care about at the time and have forgotten. But that's a different claim from the efficacy of homework; and I'm not sure how we resolve it given that kids (even at college age!) often are really not sure what they will ultimately find interesting and helpful (let alone secondary social goals like insisting on the teaching of the "second" language in a bilingual society).
It's interesting that everyone is uncertain how well academic homework works, including myself, but I don't think anybody doubts that if one afternoon you showed a group of ten-year-olds how to throw a baseball for 40 minutes, and that night one of them went home and practiced throwing a baseball for two hours, the next day that one kid would be the most likely to have improved at throwing a baseball.
Homework *sucks*, and I personally find the prospect of inflicting it on my future children abhorrent (they have one youth, and *that's* what they're going to spend it doing?), but on a purely competitive level, those who put in the hours at any given activity are in general those you can count on to rise to the top performance-wise.
Sometimes the point of repetition in the form of homework is not to be able to answer the question at all, but to be able to answer the question quickly and easily without having to think about it. It's more like training than education; the point is to develop unconscious competence so that you can just do the task automatically. I've been a great math student, but I tend to do mental arithmetic relatively slowly; I always came up with tricks to compensate for a lack of immediate recall, so I never got to the point where I could multiply any two single-digit numbers instantly. For example, if you ask me to multiply 8 by 7, I might not remember immediately that the answer is 56, so I'll remember that 7*7 is 49 and add 7 to that to get 56. In second grade I discovered that I could calculate 13 - 7 by calculating 10 - (7 - 3) instead of by knowing that the answer was 6. When school math did become more complicated than arithmetic, being clever with numbers and manipulations was a big help, but at the level in which memorization is the fastest algorithm I didn't quite keep up.
I think it also depends on what you are targeting. Homework is good for practicing a particular action so it becomes easier, and memorizing things. I'm less convinced that it is useful for understanding concepts.
50 years ago, the ability to memorize facts and be good at executing certain repetitive action was useful in some cases (math, writing) but I would argue it is far less useful now. I think in modern society, computers/internet/automation have removed a lot of the value in memorization and mastery over repetitive tasks. The way I'm able to be a high functioning member of society is specifically by *not* memorizing things (wastes time and mental resources) or being good at repetitive tasks and instead I'm good at looking stuff up on the internet and knowing how to ask WolframAlpha good questions.
If we want children to be good at taking closed book tests, which is completely non-analogous to the modern real world, then sure, homework may help. This feels like Goodhart's Law in action though as test taking just doesn't represent reality at all.
Did they include hours watching TV while doing homework? Homework took me a long time because I did it while watching my favorite after-school TV shows.
I'm going to conjecture that homework in elementrary school has some value if it's set to "overlearn" - what non-educationalists call "practice until you can do it in your sleep" or something - basic skills such as reading, writing and arithmetic (particularly number bonds, times tables etc.).
When you're trying to do 3 x 14 or something like that, not having to "context switch" to work out 3 x 4 really helps, and this is consistent with cognitive load theory in psychology, which as far as I know does have good studies behind it. How do you get 3 x 4 without having to stop and think? Lots and lots of practice. Homework is one way to find extra time outside classroom hours to do that practice.
A school with no homework but a longer school day would probably do exactly the same, there's no reason why the practice has to be done at home. But it does have to be done.
From my own school experience, homework was a mixture of practicing things we'd already done in class, reading new things in preparation for next class, and busywork. The latter of course doesn't have much learning value.
Back in the dear departed days of psychedelic experiment, there was a phrase to describe the most important predictors of the experience of a proposed 'trip': Set, and setting. Set, the person's emotions and attitudes to such a possible experience. Setting, the surroundings - kind of surroundings, comfort, trust in anyone there to support you. 'Bad trips' (unpleasant or disturbing ones) could be largely avoided by personal confidence, and pleasant surroundings with trusted friends.
I suggest this is a useful way of thinking of investigating the experience ('outcome') of homework. It becomes evident that in particular 'setting' is crucial - a noisy, chaotic small apartment is not conducive to calm 'homeworking' - no matter how long spent on it. I'll bet one factor not even thought of by the investigators was hunger. I'll bet not one of them had ever feared for security of food, or indeed been hungry beyond a half hour beyond usual mealtime.
In other words, cliched experimental designs give cliched results - useless for educational technique guidance.
If I understand the study, "the students with more homework" did *not* do better in an absolute sense. They improved their baseline scores more.
The public policy issues with homework are almost never studied. Instead, people diddle with this nonsense as if it matters, and oh, lord, the elementary school parents talking about it! Jesus, people. Homework at that level is irrelevant, because nothing in elementary school matters, gradewise.
But the public policy issues with homework dwarf the piddling possible improvement it may or may not reveal. Some of you have mentioned this. At the high school level, the only relevant issue with homework is that it substitutes compliance for ability. It's very common for kids with high abilities to get lower grades than kids with lower abilities (and also similar abilities, which is just as idiotic).
Grades in America are a fraud, and we are now in a world where these fraudulent grades are the only legal tender for college. It's revolting. Meanwhile, people spend time and energy wondering whether a kid who counts on his fingers will get one more problem right out of fifty if he spends 30 minutes a night laboriously practicing.
Crazy. Michael, I thought your summary was fine and very clear.
Parents of elementary school children: homework is irrelevant and you shouldn't waste a single second worrying about it. If your kid doesn't understand classwork, find a shrink to see what LD he or she has.
Parents of high school kids: do whatever you can, including bribery, to get your kid to do homework. Do not make it a moral issue or a learning issue. Make it clear that it's bullshit, asinine hoops they need to jump through to keep their grades up.
If, on the other hand, your kid copies homework just to get it done, judge them severely and wonder where the hell you went wrong. The first kid is far superior.
On the third hand, if you have a kid who just does the homework even though he or she knows it's bullshit and genuinely enjoys ticking off tasks, well, they're weird but it makes life easier for you.
I don't give homework. I turn anything important into classwork. I write it about it a lot, but here's The Rules that Matter:
I. Teachers assign homework or they don't assign homework.
A. If they assign homework,
then they grade homework or they don't grade homework.
1. If they grade homework,
homework is essential to understanding or it's not.
a) If homework is directly essential to understanding,
then the homework grade is the same as the assessment grade.
b) If homework is not essential to understanding,
then the homework grade is corrupting the assessment grade.
(1) If the homework grade is corrupting the assessment grade, then the school's administration will prevent this or accept this.
(a) If the school's administration prevents grade corruption then the
public will scoff.
(b) If the schools administration accepts grade corruptions, the district will
prevent this or accept it.
(i) If the district prevents grade corruption, see I.A.1-b-(1)-(a) (ii) If the district accepts grade corruption, this is a typical school. 2. If they don't grade homework, then students won't do homework.
B. If they don't assign homework, the public will scoff, and policy makers will scowl, but that's why god created tenure.
I've always hated homework and frequently didn't do it when it was assigned. The particular details of when I would or wouldn't do it are complex.
In high school I ended up doing all the assigned math homework anyway, and in college I did much less than that. And it seems pretty clear to me that this directly impacted how much material I ended up retaining; the quality of my learning is much lower for the college material.
So I would say that yes, practice matters in math. On the other hand, I would estimate the amount of homework that's necessary to fully learn and retain "9th grade algebra" at zero. It's not at all complex and doing work further along in the curriculum will provide more than enough drill in the basics of algebra.[1]
Moving back to the topic of the post more generally, I would bring up the result that student performance in math generally tracks "exposure to math in school" pretty closely, whereas performance in other subjects (and most notably reading) does not. I find the argument persuasive that this is mostly due to the fact that students rarely do more math than they are assigned in school, whereas many students do a lot more reading than they are assigned in school.
[1] Another thing that seems obvious to me is that different people will require different amounts of practice to learn and to retain the same material.
I think that homework is important, because to me it's practice; you wouldn't say to a kid "did you learn that music piece/recipe/sport?" and have them say "Oh yeah, we did it in class" and then "well have you played the piece/cooked the recipe/played the sport?" "No need, we did it in class, I remember it!"
We see the value of practicing/trying out practical skills, I don't think mental ones are much different. For primary school, leave homework until they're older (7+) and only a little, because at that age it's (a) getting them into the habit of having a routine and doing it and (b) physical practice like writing so they get the skills.
I don't know how much homework secondary students get in America (or indeed what it's like in modern Ireland); in my time coming up to the Leaving Certificate (final year national exam for ages 17-18 school leavers) we were recommended by our teachers to do an hour per subject study every night. You could be doing 6-8 subjects, so I don't know how many people did spend 6-8 hours every night studying. But you do need *some* time studying and not just relying on "we did it in class and I remember it".
I agree that "amount of time spent on homework" does *not* correlate with "results on tests" except very crudely; good students may spend more time on homework because they are studying, and good students will probably do better on tests. But you have, as pointed out, smart students who skip or spend little time on homework, who will do well on tests anyway, and poor students who spend a lot of time on homework because it's hard for them and will do not so well.
(Just for clarification, I am German so that is the "your native language"-subject.)
I spent less time on homework in German (the subject, I mean) than in math but the reason was not that the German homework was less, it was actually more. I just didn't do it because it felt like it was too much and very not fun to me while math was always fun (either it was easy or it was like solving riddles designed to be easily solvable). I think another reason was that you get a lot of "I finished part of it" rewards in your head, since Math homework was usually a bunch of different tasks while German homework was usually one big essay kind of thing. I did English and French homework more regularly too and that was usually a bunch of little tasks and not one big essay.
I’m a professor, and I research (among other things) learning and the development of expertise. I think that much of the (very interesting) discussion seems to ignore or downplay comments about the quality of studies in this area. The lack of good studies on HW specifically should motivate us to rely more on theory and on related literature (and improve studies).
We have lots of studies that are more theory driven, and we do have first principles. When we ask a question like “is homework effective?” What is the assumption about when students will practice the concepts that they have been exposed to in class, if not via HW? Just theoretically, if we had a complete dataset on the exact amount of time that every student spent doing HW, is anyone hypothesizing that in elementary mathematics (say), there is no relationship between the amount of time students spend practicing a task such as adding fractions with unlike denominators and (a) their accuracy on a test of that material, and (b) their facility / speed given a consistent level of accuracy (i.e., expertise) with which they would be able to execute concepts such as these? (Controlling for other relevant factors such as ability.) In other words, is the conclusion of the “anti-HW”/“HW is useless” crowd that, given a perfect dataset with full controls (or random assignment and perfect measurement in terms of time and attention), deliberate practice on tasks like learning times tables, adding fractions, working with percentages, word problems, etc. would have *zero relationship* with expertise as measured by tests?!?
If that were true, it would be very surprising, to say the least. In laboratory experiments where people are required to learn things, practice (and attention - “deliberate practice”) and time on task are strong predictors of learning. I used mathematics as an example, but if the concept were “knowledge of history”, or “knowledge of geography”, it does not seem possible that this relationship would be zero. Learning to write well involves writing, then seeing feedback on that writing (and preferably, a chance to fix whatever errors were made, without punitive grading), etc. All of these practice opportunities cannot be fit into a class day without sacrificing instructional time.
What many commenters seem to be getting at is that much elementary school HW is a waste of time in comparison to the hypothetical HW that I have postulated (one person mentioned coloring assignments, but there are so many “silly” things, many of which are really “parent assignments”). I do agree with other commenters that some of those comments are too strong, because it implies that academic skills are the only skills that elementary school is cultivating, which is untrue.
As a final point, the academic literature in Education does not tend to be the strongest in terms of statistical or analytical rigor. As just one example, the analyses in most of the HW literature fail to account for attenuation of the focal variable due to the (gigantic) measurement error with which HW time is measured (see https://en.wikipedia.org/wiki/Regression_dilution if this is unfamiliar to you). This is one of many similar issues that plague this kind of work, and which may or may not be affecting our conclusions.
Scott, your post is about about the need for meaningful, systematic data and the woeful, pervasive lack of that data. Therefore, I'm going offer an anecdote from my own personal experience, which of course will totally, definitely resolve the issue once and for all (for reference, I was born in the early 1970s and had the standard range of k-12 public schooling, along with college and grad school):
1. In elementary school, most homework seems to have been busywork to do so the teacher could say they assigned homework.
2. Exception to #1: Beginning around 5th or 6th grade (maybe as early as 4th grade?), we were assigned short stories to read and book reports, etc. There wasn't enough time in the class day to read in class, so we had to read those at home.
3. Another exception to #1: Well, it's only a partial exception. Even when it was busywork, it taught me a measure of self-responsibility which ultimately was, for me, empowering. I learned to schedule and to plan. I probably overlearned those skills and became a little obsessive about it. Or a lot obsessive. It also made it harder for me in college and, especially grad school, when I had to learn to put things aside for a while and focus on other things because there was too much to do all in one night.
4. In middle school and high school, homework was sometimes busywork, but just as often necessary to learn what I was assigned. With some subjects (history, French, English lit), I really looked forward to it. With other subjects (math, physics, chemistry), I struggled through the homework and didn't learn anything.....until I had a trig and calculus teacher, in my junior year in high school who had a different approach. She assigned the problems first and had us teach ourselves how to do them. The next day, we went over the problems and what we didn't understand from the textbook. I, for one, loved this approach, because it taught me to take control over my own learning and not depend on the teacher's lectures. (Of course, that approach didn't work for everyone.)
5. In college and grad school, homework was most of what I had to do when it came to learning. The lectures helped. I wouldn't've been able to get by by teaching myself and reading. But (ugh!) graduate seminars (where we all talked as if we knew what we were talking about, having pretended to read the first paragraph of the introductions of any given book) did very little and it was the homework. And, as I said above, the "responsibility" and work ethic I learned from my early years of homework made it difficult to triage.
Before I read this post, I pre-registered my thoughts about homework. Updated thoughts are at the bottom.
How important is homework?
I am sure tat if you look at large aggregate data, it will show low effectivness for homework with respect to learning an future earnings.
So where should you look?
First a definition.
Homework is type of practice of the concepts, processes, and skills one learned through instruction. It’s self-guided exploration in the zone of proximal development.
Second, some comparisons.
Singapore, Finland, South Korea all have homework; they are the three of the best education systems in the world, maybe they know something you don’t.
Third, the best interventions we know of in schooling are ability grouping and direct instruction. Both of these interventions would change the nature of homework. Turning inconsequential “practice” flailing about completing worksheets with deliberate practice that is up to the level of the student.
Fourth, I run two schools which only give homework two days a week and have three days a week of in-person instruction. We consistently achieve above average on standardized tests and one consistently rank in the top 20 of schools. So, I both am prone to belief that practice is important, and that the quantity of time spent in school is gratuitous.
Given these facts, I have the following hypotheses:
1. Homework will be as effective as one’s practice regimen.
2. Homework will be effective if it employs the right quantity of a student’s understanding.
3. Homework is necessary, because learning happens in the space between one’s ears. If one is not doing something with the brain, it is not learning.
Maybe these are uncontroversial and nearly tautologically true?
---
Update:
Scott brings up a question I haven't thought much about. How does homework's importance change with time? At the same time time, he admits that's practicing algebra problems probably makes one better at algebra. To the question about hand, I don't think the answer changes much. When I watch the curriculum closely, I notice that fourth grade homework is especially focused on memorization and dozens of 'microskills' like placement of punctuation, building up 3 digit multiplication, and the "states and capitals" world context stuff.
While we don't know the value of homework in the RCT Sense of 'know', I will still put my confidence levels in the value of homework as high as my confidence levels in the value of unguided deliberate practice.
About halfway through my engineering degree in college, I realized that if I wanted to get good grades, I either needed to attend class or do the homework. This realization served me well till I graduated.
Regardless if having homework for elementary schoolers is conducive of their learning, it does seem useful to have it at all if doing homework is useful for high schoolers, right? I had barely any homework in elementary/middle school and remember having a culture shock early in high school when I realized I couldn't finish all my work in ten minutes anymore.
I know it's only anecdotal evidence, but when studying partial differential equations in college, it was not enough for me to attend class and do the assignments; i had to spend significant extra effort doing all the problems at the back of each chapter to really learn the material well enough to pass the course. I was motivated to do that and ended up with a B in a class I was initially failing. So at least in this one case, 'homework' made a big difference.
Public school teacher here. I’ve observed the current trend to be moving away from assigning homework. In my master’s program for teaching, we were told homework was not equitable because not all students have the privilege of free time or the proper environment at home. Honestly, I’m not sure what to make of that. Many of my high school students are not motivated by failing grades. Like others have said, if they don’t want to do it, they won’t. Granted, some subjects do require more practice and time, so I don’t see a way out of some homework. Like others, I developed my own intrinsic motivation to learn in university. In education research, the studies are quite pitiful. Over time, I’ve just settled on the belief that you can’t force anyone to care about anything- especially youth existing in the current state of public education.
1 - Even if younger students don't benefit as much from HW, you'd be hard-pressed to start instilling the habit starting in their teens. That's a losing battle for parents, it's like asking your kid to do chores starting at age 14. Extensive amounts are probably redundant, but I'd foresee a benefit to their assignment in the long-run.
2 - Some have shared anecdotes that they performed well in school despite ignoring HW wherever they could. This doesn't account for study time ahead of tests and exams, which, for classes like math, amounts to the exact same thing as HW. I've seldom encountered people make the claim that they excelled in school despite doing no HW *and* not studying. Those that have likely possess superior focus and memory skills.
Understanding the concepts is one thing, but retention (let alone of the details) can be fickle, and benefits enormously from repetition. It's an important foundation for understanding and applying more advanced concepts in layers. I think it's a mistake to blow up the importance of long-term retention such that performance in tests is then thought inconsequential. The skill instilled from study or hw is probably more important than retaining some abstract concepts in hs-level STEM. If most of us don't use calculus in the long-run, difference in learning outcomes (if outcome is measured well after the fact) in that domain is negligible; we barely remember how to put it to practice. Use it or lose it. Basically everyone keeps writing and speaking their native language beyond school, that doesn't usually erode. If I pick up my guitar now after having it collect dust for several years, I can recall scales and a few songs, but what I can muster out of it is tantamount to doing basic arithmetic. Goodbye Paganini and Eruption solo. What are the concepts doing for me now, in isolation? If you only remember coles-notes versions of everything you're still basically impotent in those respective disciplines.
(a propos writing: many novelists will tell you that writing is a craft, and there's no magic behind it. If you hack away at it for hours a day, you improve)
I haven't looked at all of the comments, so someone may have already gotten here and I'm late to the party. But it seems to me that one solution to both the homework problem and the diminishing-connection-to-our-IRL-communities problem is to make homework an in-class team task (a type of flipped classroom) that has to be completed by everyone in the class or credit is not given to anyone. The students would have to be given support from their teacher in how to troubleshoot issues like a kid not being able to do it, so they would learn a lot about different capacities and the ability to support others on a team IRL. Yes, it requires a lot of teacher training, but it's what they are missing the rest of the day (on their phones) and enforcing it in the classroom may be the only way we heal that.
Tangentially, regardless of the inconclusive evidence (or trusting of the mediocre evidence?) my teacher training program emphasized that it *doesn't* help and urged us to give homework sparingly.
Even the algebra study is based off a very specific counterfactual of homework as it's generally done in algebra classes, where the teacher lectures the whole class and students only get practice doing it on their own. It would seem much more logical to completely invert that, where students watch a lecture on their own time with the ability to fast forward through or rewatch parts as they deem necessary then go through exercises during class time where they can get help from the teacher and classmates when they get confused.
Art of Problem Solving (https://artofproblemsolving.com/) changed my feelings about homework -- they run these amazing math (and two programming) classes where you are in class for only 1.5 hours a week, but mostly you learn through doing the excellent problem sets, which both build skills in the material and give practice in doing challenging problems, and are fun problems to boot. If all homework were like this, I'd be in favor of it. And it also clarified to me that in the best case, homework can act as a practical way to build skills, just as people have brought up music or sports above.
Though as it is, I'm glad my kids' school doesn't give homework so I can have them do AoPS and Beast Academy (for lower grades, run by the same people) instead.
This is actually something I've thought about quite a bit while I was teaching introductory chemistry and physics at a private high school (7 years before transitioning to software).
First, I have yet to see a *non* garbage educational study (coming from a hard science background). Small sample sizes, buckets of confounding variables, ad hoc "controls" or no controls at all, studies that either don't replicate or (most commonly) don't scale at all beyond the small focused lab schools, etc.
"Time spent" is a horrible measure of anything. First, because it's horribly unreliable. It was a running joke that students, teachers, and parents would all estimate completely different amounts of time for the same set of homework. Part of that was that students would include things like "I have the book open, but am snapchatting with my buddies". For tests, I generally used a 5-6x modifier--if I could do it in 10 minutes, I'd expect it would take the median student 50-60 minutes.
Beyond that, I actually ran an experiment (because I was lucky enough to have an administration that gave me carte blanche). For the last few years, I didn't assign "home"work. Instead, I did a few things:
1. Practice time and practice sets in class, with me walking around helping people. If you didn't finish, you'd have to do so at home. But they were scaled to take up the class time.
2. Those practice sets were *irregularly* graded. But rarely. Instead, I did a combination of two assessment methods between major summative assessments, both fairly low stakes (~5% of the total grade, with the median effect being ~1% either way).
2a. Pop quizzes on the previous-day's work. 3-4 questions, 10 minutes. Basically taken from the problem set the previous day with numbers changed. Graded right there in front of everyone, usually by a peer as we'd walk through the problems in detail.
2b. "Online Practice Sets", which involved a tool I wrote to generate a pocket static website that presented questions (multiple choice and fill in the blank, mostly) and then auto-graded them (including some allowed variance for numeric questions). They could do them as many times as they wanted (including later on to prepare for the tests), and the system would only report the highest score. The problems and answers were randomized (in order at least). If there was a pop quiz, it would often come from those problems.
I changed because I realized (both from surveying students and my own observations) that traditional homework ended up in one of two buckets--
1. Common enough to be useful...but then the grading load was too much so they didn't actually get feedback and it ended up being graded on completion. This was utter garbage for everyone--no one took them seriously, including myself.
2. Rare enough to have time to give meaningful feedback and be scored properly. These, in turn, ended up being basically just another summative assessment. And weren't frequent enough to actually provide any reinforcement.
So switching to a basically "no homework, but lots of practice embedded throughout" system was win-win. Can I say that it had substantial effects on performance? No, because we also were revamping the curriculum fairly constantly (one of the reasons I left was that we were just always chasing the new hot thing instead of actually measuring whether the changes were doing anything useful). But it was way more honest and the kids liked it more and were more engaged. Or had fewer people to blame when they weren't. And allowed me to target more practice for some things than others and let them redo their practice. Plus, the panic on their faces when they'd have a pop quiz (even when it was announced) was beautiful to my jaded, evil soul.
Say I wanted to improve some practical skill, e.g. batting in baseball (a relatively intricate motion required coordination, full disclosure I don't play baseball). I expect it would be hard to ask a question like if I swing faster will I bat better in a single experiment/study.
1) A sudden shock to the learned motion most likely will make things worse before they get better (at least, this is my personal experience).
2) To accomplish this goal I will need to adjust other elements of my swing, it is hard to isolate one element at a time...
To assess whether batting harder is better I would instead consider the mechanism (yes, hitting things harder make them go further) and some empiricism (yes, athlete that bat further hit harder).
This strikes me as similar to the case for homework and teaching in general, the reason that we believe that homework improves performance is that there is a clear mechanisms and successful teachers seem to have settled on a certain balance of homework. The type of study Scott seems to be looking for seems to be inherently fraught, like asking a someone to suddenly change their swing and concluding something from the experiment. Perhaps (speculating here) this is why there are not so many studies of this flavor.
Has anyone asked whether the student is motivated to learn? I suspect that homework might be of value to a student wanted to learn and didn't get the concept or have the vocabulary mastered yet.
When I was teaching myself foreign languages, I assigned myself brutal amounts of homework and forced myself to do it because I wanted to learn. I don't think I could have otherwise developed whatever capabilities I did in fact develop.
When I was an undermotivated student, which was most of the time in any sort of coursework, homework was just a hoop I had to jump through, and cats hate jumping through hoops.
"The utility of homework and its impact on academic achievement has been an ongoing debate for more than 100 years. To date, there is no rigorous experimental research to show whether, and the extent to which, variations in the amount of homework assigned can impact the academic performance of elementary school students. In this study, 440 second grade students were randomly distributed in 3 groups within the classes they attended. Each group received different amounts of homework in writing and math for 20 days. The results showed a significant immediate effect of homework quantity on writing competency (but not on math competency). The writing homework effects were sustained 4 months later, but only for the group that had allocated a moderate amount of homework to writing skills practice. This study shows that the additional opportunities for practice offered by homework can have differential short-term and medium-term effects on the academic achievement of elementary school students and supports the assignment of a moderate amount of writing homework at this age"
b. Commentary
I can understand not including this one in your review.
There's not much to like about the design of the study, or the figures within the writeup (both are overly complicated, particularly the figures).
Still I did their findings (and underpinning theory) plausible for 2nd grade kids, and so for all completists, note this RCT and its takeaways.
2. "The Role of Homework in Student Learning Outcomes: Evidence from a Field Experiment"
"In this article, the authors describe a field experiment in the classroom where principles of micro- economics students are randomly assigned into homework-required and not-required groups. The authors find that homework plays an important role in student learning, especially so for students who initially perform poorly in the course. Students in the homework-required group have higher retention rates, higher test scores (5 to 6 percent), more good grades (Bs), and lower failure rates. The authors also study the relationship between endogenous homework submission and test performance using instrumental variable estimation and find that homework submission has a large positive effect on test performance."
b. Commentary
The gains appear real, and while modest also meaningful, across outcomes.
It was also striking how much homework students in the “no-homework” group opted out of; making it a requirement changes behavior to a much larger extent than I would have expected in this setting (college).
3. "Online Mathematics Homework Increases Student Achievement" (Michael Pershan mentioned this in an earlier comment.)
"In a randomized field trial with 2,850 seventh-grade mathematics students, we evaluated whether an educational technology intervention increased mathematics learning. Assigning homework is common yet sometimes controversial. Building on prior research on formative assessment and adaptive teaching, we predicted that combining an online homework tool with teacher training could increase learning. The online tool ASSISTments (a) provides timely feedback and hints to students as they do homework and (b) gives teachers timely, organized information about students’ work. To test this prediction, we analyzed data from 43 schools that participated in a random assignment experiment in Maine, a state that provides every seventh-grade student with a laptop to take home. Results showed that the intervention significantly increased student scores on an end-of-the-year standardized mathematics assessment as compared with a control group that continued with existing homework."
b. Commentary
I can also understand not including this one in your review, as it’s NOT a "to homework, or not to homework, that is the question" study in the strictest sense. But an interesting comment on what *kind* of homework works.
One notable finding is that it helped low-achieving kids more (same as study 2). Effects were ~0.29 for kids below the median, ~0.12 for kids above it.
In contrast to study 1, and I know this is a matter of taste in some respects, I really liked the study. No deception or tricky maneuvering, as far as I can tell.
4. "Disrupting Education? Experimental Evidence on Technology-Aided Instruction in India"
"We study the impact of a personalized technology-aided after-school instruction program in middle-school grades in urban India using a lottery that provided winners free access to the program. Lottery winners scored 0.37σ higher in math and 0.23σ higher in Hindi over just a 4.5-month period. IV estimates suggest that attending the program for 90 days would increase math and Hindi test scores by 0.6σ and 0.39σ respectively. We find similar absolute test score gains for all students, but much greater relative gains for academically-weaker students. Our results suggest that well-designed technology-aided instruction programs can sharply improve productivity in delivering education."
b. Commentary
Whether this is "homework" or "an after-school program" is a matter of debate. I think 85% former and 15% latter in my interpretation. The software just happened to be really good, targeted homework.
Case in point in this study: kids in the bottom tercile in the comparison group literally learned nothing at all between baseline and endline:
"Students in the treatment group see positive value-added in all terciles whereas we cannot reject the null of no academic progress for students in the bottom tercile in the control group.."
So homework, in this case, might be one (of many potential?) piece of low-hanging fruit that could get kids to learn more than they otherwise would in a low-productivity setting.
*
Summary
-These four studies, plus the algebra one Scott cites, tend to support that homework *can* work, under some conditions.
-The key variables, by study, appear to be (i) for the younger kids in Romania, giving them novel contexts to practice already-learned skills, (ii) for college kids at East Carolina State, making them do the homework (and make it relevant to success on exams), (iii) for the kids in Delhi giving them instruction at their level, or really any instruction at all, (iv) for middle school kids in Maine, giving them some “hints” with well-aligned practice problems so that they don’t get stuck.
Common theme in all these winners was that homework was either marked (by a computer or human). I suspect that matters quite a bit.
-Highest RoI, subject-wise, seems to be in math, though it might just be relatively overstudied (all 5 of the studies included math or math-adjacent stuff), and there were high returns to Hindi as well in the Mindspark study, and no impact at all on math in Study 1.
-Highest RoI, distribution-wise, seems to be with lower-performing pupils, though caveat emptor: my guess is that they are also unusually responsive to feedback; simply assigning low-performing kids homework and not correcting it may backfire.
I wonder if non-teachers are aware that there is a growing anti-homework movement that's equity motivated. You can check out Grading for Equity by Joe Feldman if you want to get those arguments in one place.
Lots of teachers are encouraged to both drop the % that HW makes up of the total grade and also award half credit even when students don't do the work. I'm of two minds of the approach. It clearly reduces failing grades, but I'm not sure if that is helping people learn more. You could make the argument that homework is more about compliance, but then it's not always clear what a grade is supposed to represent. Is a grade the record of your ability to meet a specific standard, a measurement of your growth in learning, or an assessment of the effort you undertook? If we had a perfect way of measuring those things, which would we choose?
I‘m a (new-ish) teacher and at university we heard that John Hattie was the go-to guy for statistical analysis which aspects/methods/… work and which don’t have an effect on furthering education. His meta-meta analysis „Visible learning“ combines 960 meta studies and encompassing 260 million students to check out everything from class size and teaching method to TV hours at home and nutrition.
The median effect over all of them is d=0.23. So basically most things have some kind of positive effect on learning. The goal for teachers then should be to find those things that are better that most and concentrate on those. So he suggests taking 0.23 as „helps just a little“ (and below that as „hardly helps“. Effects over 0.4 are labelled „works well“.
Homework is ranked 88th, d=0.33. The meta-analyses were
Anyway, my take-away is that I mostly don’t bother with homework. The kids who understand the subject in class waste time doing things they already know. The kids who don’t know how to do the task don’t get any help so they will hardly get better. More time spent doing the thing will increase capabilities a bit but I have to be conscious of my pupil’s private life as well. They need time to learn the piano or play soccer or be kids or whatever. There’s so much learning possible if I don’t waste their time.
That said (as a primary school teacher), they will have to learn to read and write. There’s no way around that, if you don’t practice reading at home it will take you soo long before you get your 10.000 hours or whatever in school. So I‘d argue if there is ever a place for homework then it’s daily reading and writing in first and second grade.
Well, that's a sad end to my smugness right here! I am (was) pretty much the only teacher I know who read Hattie and tried to at least keep some of his results in my mind at times. Everyone else just does whatever they feel is "right" - I went with "science", and lost (time and maybe intuition).
I will poke around your suggested bestevidence.org but many misunderstandings lie this way. I'm German and my German Hattie edition had a lengthy preface detailing that you can't just translate the words as educational concepts in both countries differ quite a bit. E.g. how to teach "reading first grade" may be quite different so there's a lot of additional research involved.
One would have to start with a typology of homework. Some learning outcomes need skill drill. For others (too much) skill drill is harmful. Finding the right balance between skill drill and free exploration is crucial in so many areas of learning (whether sports, music, maths, etc). How to get this balance right is the hard question for a teacher.
And without putting this balance between skill drill and free exploration into focus there is, in my opinion, no hope to assess whether homework works.
Btw, some people seem to be able to learn a subject without skill drill. In my experience, this is only the case because they integrate skill drill into their free exploration. (I first noticed this when listening to a Dennis Bergkamp interview where he described how he learned to play football (soccer) as a kid.)
As someone who always hated homework, and spend countless train rides in the morning and countless breaks in between lessons to copy homework from others, I fully support your conclusion! Very grateful to those who always let me copy their finished work without ever complaining. Homework didn't count for our grades, but not doing homework would lead to different kind of punishment, including prominently to have to stay in school for some additional hour(s).
On a more theoretical note: Even besides the evident shortcomings of cited studies I think the question of 'does homework work' is grossly underspecified. As already mentioned in other comments, it's probably much more 'which type of homework' 'for whom' and 'under which conditions'. With 'which type of homework' not being algebra vs. French as a foreign language, but the specific type of exercices in each of those.
I've always sort of assumed that the value of homework came from two things: First, that school is over at 3pm or whatever, but the longer you spend thinking about something the more of it you retain (80,000 hours and whatnot); and second, that working on something closer to bedtime (than 3pm) makes it something that is more reinforced by whatever the mysterious reinforcement mechanism of sleep is.
I say "assumed". I have no more idea than anybody else. But it is enough to make me scowl when I hear about kids not doing homework any more, and not *just* because I've paid *my* dues.
The skill my kids seem to have learned by doing homework is how to work independently (compared to me who wasn't set homework for most of my school career). This has spilled over into them being able to do none school work with more focus and be much better at revising for exams etc. It's probably a useful life skill.
If you look at countries who have extraordinary amounts of homework, like South Korea, they seem to do astonishingly well from an academic achievement point of view. While this might indicate that homework is effective, it is a very high price to pay in terms of happiness and time spent learning on things mostly useless in real life.
Whenever I think about homework I think the one-size-fits-all approach to school is really the underlying problem. The super smart kids in the class that find the material super easy and boring are way more likely to do the homework. They want the good grade and homework is part of the grade. The kids that are struggling with the material and that could use extra help or tutoring are least likely to do the homework. The first group gets no benefit from the homework that they did and the second group that did not do the homework actually could have benefited from it.
There should be a skills based system where based on how you did on the test/quiz determines how much if any homework you get on the topic. If you just aced the test you don't need the homework. If you got a D there should definitely be some homework as you clearly do not know the material as well. Apparently, we don't do this because it would not be "fair".
Personally, I did all "homework" during the school day in other classes as I found school too easy and boring and the homework didn't even being to be challenging. This resulted in homework without the "home" portion.
in this case, it seems like we should default to homework works.
Science is conflicted because not enough good studies, but a priori homework should work (deliberate practice and whatnot) and common sense says homework should work. Maybe a good reason not to have a billion hours of homework, though.
I find the workload at a 40 hr job is less than the workload at an elite 11th-grade high school, which feels wrong.
Studies on learning suggest that repetition of material is valuable for learning.
As such, I feel that it is likely that any sort of repetition of material - be it in the form of reading a textbook then going thrugh the lecture and taking notes on the lecture, or actually doing homework problems based on the material covered in the lecture, or doing a project which relies on the skills you are supposedly learning, or watching videos about the subject matter - will reinforce learning.
Speaking as someone who enjoys writing fiction, repeatedly writing fiction (in the form of fanfiction) greatly improved my skill at writing fiction, and that I started out at a higher level than most people in the fanfiction community because I'd spent years roleplaying and reading books - and while neither of those things are precisely writing prose fiction, they rely on many of the same skills.
I think that there's often a "pool" of skills that surround various things. Writing prose fiction is not the same as reading prose fiction or roleplaying (be it freeform or RPGs), but there's a significant overlap in the skills here, and if you engage in gainful practice (i.e. pushing yourself to improve your skills by repeatedly doing these things) you improve your skills.
This is obvious with art as well; practicing art is how people get better at art. If you look at artists who practice for a long time, their skill levels tend to go up; in some cases, this can be especially dramatic if you catch people very early in their artistic careers. A good example of this would be Miss Mab; if you look at the very oldest comic she posted, back in high school ( http://www.missmab.com/Comics/Vol_001.php ) and then compare it to what she does today ( http://www.missmab.com/Comics/Vol_2084.php ) the difference is stark. Even if you look at a lesser time span within that, like five years between this ( http://www.missmab.com/Comics/Abel_03.php ) and this ( http://www.missmab.com/Comics/Ab_098.php ), you can see significant improvements.
However, at some point, unless you push yourself to do new things, these gains can become increasingly limited. For instance, this comic from 2013 ( http://www.missmab.com/Comics/Vol_1389.php ) and this comic from 2019 ( http://www.missmab.com/Comics/Vol_1896.php ) is a larger time gap, but the gains are clearly not as significant - she was already quite good at art in 2013, and while she did get better over time (particularly at backgrounds) the advances have not been as large.
As such, doing homework is valuable, but doing infinite amounts of homework isn't - there's declining returns, and you get larger returns by pushing yourself further along the road rather than endlessly repeating the skills you already have.
I think this is where people get unahppy about homework. The more skill you already have, the worse the gains will become by doing the same thing over and over.
I suspect this is why homework appers increasingly more valuable as students get older - in elementary school, a lot of the higher end students already baically know everything or most everything elementary school teaches them. As such, elementary school homework is a waste of time for them.
However, lower end students are not nearly as good at this stuff, and as such, will see significantly larger benefits because they don't already know this stuff.
A kid who enters elementary school reading the kinds of books you read in 6th grade or even high school is going to get almost nothing from elementary school English classes.
I also suspect this is why math homework seems so valuable by comparison - few parents teach their children much math independent of school, and particularly not higher levels of it, and people generally don't naturally do much repetition of math at that age, so math homework will seem very valuable. And once you hit algebra, I think that goes way beyond what 90%+ of parents will ever spend time drilling kids on independently.
Conversely, a lot of kids enjoy reading books, so a lot of other subjects are more likely to be covered. I enjoyed reading Zoobooks and various science books, so I had a huge level of knowledge matter on the subject. I even read some textbooks for fun because I was a horrible little monster like that.
This meant that I had a ridiculous advantage over everyone else in science class, even through high school and early college. I knew this stuff back and front and could spend very little time on work and get better grades because I had already put in the work through years and years of independent "study", so the homework was entirely pointless for me.
But if I tried the same thing with college calculus, I was doomed.
So, yes. I think that homework is valuable because it is forced repetition of learning ,but I think that "pre-existing knowledge" is a greatly underappreciated confounding variable.
I think you will see larger benefits of people doing homework in subject matters that they don't independent study than those that they do, and I think you will see larger benefits of homework the more advanced the subject matter is, as that means that fewer and fewer students will have natural background in it, and that you will see larger benefits on lower end students for lower-level materials than you will see for higher-level students on lower-end materials.
Alfie Kohn is a longtime anti-test, anti-grades fanatic. Similar to the FairTest idiots.
There's homework, and there's homework. When I was in school most of it was busy work. In college though, most of my problem sets were actually a good way to practice things.
I wonder if parents could just say to their kid that they didn't have to do any elementary school homework. Would a kid who otherwise performs well get away with not doing any homework? It would probably depend on the school.
Let's just continue with something of unknown utility. Why not?
Schools went overboard on homework for common core. It sucks. Most of it is poorly designed. It seems more designed to indoctrinate students to be complaint than to learn.
As a parent dealing with schools for almost 20 years, learning doesn't seem to matter anymore.
Thanks for this post! Short, crisp, interesting summary about something I've meant to research but never got around to.
the graph does look like returns diminish quickly though - you've already gotten nearly all of the benefit at 50% the amount.
The purpose of homework isn’t learning. The purpose of homework is to sort those who will conscientiously put the new cover sheets on the TPS reports from those who won’t. It’s to filter for conscientious and agreeable personality traits as well as work ethic.
Kohn is not "[taking] the studies seriously anyway." He argues that the studies are, as you wrote, inconclusive at best and that one should recognize the demonstrable harms of homework and limit its use (seeing as it's not particularly compelling that it does much). It is easy to find compelling articles on the harms, including physical health, and it extends to even competitive environments where it actually negatively impacts learning (but helps with the educational rat race). e.g. https://www.tandfonline.com/doi/abs/10.1080/00220973.2012.745469
I would hazard a guess that homework vs proficiency would produce the exact same results in elementary school. If you throw more time at a skill the better you get, that seems pretty straight forward.
Practicing piano is another good example they should study. If you don’t practice every day, your progress would be limited. And the age works as well because the most amount of kids in piano lessons is likely elementary school. If they don’t have the drive to practice (such as myself) eventually they will quit.
The same goes for sports. Starting early and spending more time practicing will produce oversized results.
However if you graph enjoyment vs homework, I’m sure the graph would be inverted but that’s the whole point of parenting, isn’t it? To force your kids to learn skills they otherwise wouldn’t on their own?
One of my friends was essentially neglected as a child and spent most of the 80s and 90s watching TV. He’s in his 50s now and has no skills except for being a substitute teacher. I love him to death but I would never let him teach my child anything. He’s basically the poster child for forcing your kids to do homework.
Another potential confounder is that incompetent teachers almost certainly will either assign far too much or far too little homework. My own homework requirements were much more reasonable than those of my friend's kids, for example.
Speaking personally, I found it useful in some subjects, not useful in others. It depends on what you are given and how much of it there is to do. I think it made a lot more sense in the context of classes where they said "OK, we have this much work to do, and it's due tomorrow." If you finished in class, you finished in class - nothing wrong with that.
I remember a lot of homework through junior high that consisted of "colour this map; ensure the entire Pacific is blue crayon"
From a quick glance at the Nawaz and Welbourne study, it seems that they didn't look at performance in classes other than the algebra class.
I take the default model to be that with no required homework, students will nevertheless spend some time at home trying to keep up with the material in all of their classes. When a large amount of homework is required for one class, they will do more work for that class, and do better in that class, but will do less work for other classes (there's only so much time in a day), and do worse in those classes, for zero (or negative) net benefit. If you don't measure performance in other classes, you have no evidence against this default model.
Even if the total time spent at home on school work goes up as a result of requiring homework, and overall performance in all classes improves, it would not, of course, follow that this was a net gain. Kids have other important uses of their time, which may well include work on projects unrelated to the classes they are taking, but which would be considered "academic" work if it happened to have been assigned in class (eg, reading novels).
It’s so sad that we have almost no good data on this, despite the massive amount of time spent on it. My prior is that if it’s a skill (math, writing, analyzing arguments, probably reading), then everything else being equal time spent practicing it makes you better. The biggest confound is that a lot of homework is busywork, at least in elementary school.
It seems like edtech should help us get more data points like the algebra study more easily. Hope there’s more data out there that we just don’t know about yet.
Wait, this is really all the literature??? We have education departments at every university and this seems like a really easy thing to do randomized studies on.
I've had a teacher friend tell me that homework in elementary school is useless. But as someone who spends an hour a night helping a kindergartener and first grader with homework, I'll be annoyed if it doesn't help.
I don't doubt that the material is better ingrained in their minds having done homework. However, whether that has lasting importance is another question. They'll eventually learn reading, writing, and arithmetic with or without homework.
If I were doing this I might is a hot control or comparison with tutoring. Most tutoring and time spent with a tutor is essentially 'doing homework' and/or working through sets of problems. Especially if one were to look at tutoring centres vs scores...this could be a good approximation of how much 'practice makes perfect'. Though you'd be hard pressed to not be swamped with bias fo who goes to tutoring. But with a bit of hand waving statistics one could compare similar SES kids who get tutoring vs those who don't to see if practices helps.
Another way, which I'm surprised to see is missing, is to look at homework completion rates and scores. Few teachers bother to do this consistently, but there are some who collect and mark all homework as part of a homework score. This happens a lot more in Maths or Science classes than in something like english or history but can be found all over if homework is a formal part of their summative grade.
I found it was a good opener and I'd have a 'do now' problem at the start of class and/or I'd have them do a quick peer marking to swap their homework and mark each other based on answers I'd put on the board (with some QC from me). Then I'd collect that to have them do the marking for me and get the valuable experience of engaging in marking. This sort of actual homework submission is a decent proxy for 'practice' and better than time spent. If a kid spends 5 minutes per problem or 1 minute per problem because they are smarter/already know it etc. who cares? The main thing is the number of practice problems they completed, in my view.
And they would be a good method to compare if completing x% of practice sets with x% accuracy helps or not using their marked homework submissions. So yea I'd think to look at homework completion rates of a given band of high/middle/low achievers and their homework completion rates to show effort and the scores they get in that homework to show mastery.
e.g. If two students gets a 60/100 on their prelim test at the start of the school year, then one does 80% of their homework with 90% accuracy and the other does only 20% of their homework with 90% accuracy....what will their final exam results be? Will that extra practice by the 80% homework effort kid result in a higher score than the other kid who clearly didn't practice as much?
Time spent is always going to be fraught with incredible inaccuracies akin to people self reporting their dietary/caloric intakes...only worse since it relies on children to report timespans.
Maybe no longer, but in my days, homework was sometimes used as punishment or mis-behavior.
Math teacher here. Homework is unquestionably good for math proficiency for many students if it is the right kind and there is the right amount of it, but the problem is that some students might need 10 problems while others might need 30 so there is a danger of assigning too much for some and too little for others. My proposal is to offer a large number of homework problems, not count homework performance for the students’ grades, but guarantee that X% of the problems on the test will literally be from the homework, where the ideal value of X may vary depending on the class and subject.
Hey, I know Cooper! He was my dad's thesis advisor back in the day. Very cool to see him cited in the wild
Seems obvious that practicing anything makes you better at it. That homework may be so poorly designed that doing it does not lead to improvement is possible but seems a different issue -- e.g. quality of teachers, of educational materials.
Personal anecdata: I did very well in almost every class that assigned homework (I have always been diligent about doing what was assigned). I did quite poorly in almost every class where homework was not assigned. This pattern repeated from middle school, through high school, undergrad and later grad school.
I know others whose results were completely uncorrelated with the amount of homework assigned and/or performed.
My guess is that the distribution is so wide, individual differences are drowned out by the inter-person variations. Or worse, the distribution might be multi-modal.
Homework is good if/when it gives children a chance to practice concepts they have not mastered. For a study to show homework to be helpful, it would need to take into account when students master material. I have four very different kids, and can not make grand sweeping statements about homework in general, but it has always been useful for each kid to practice skills until mastered.
I also recently reviewed homework research and summed up what I found here: https://michaelpershan.com/homework.html
I agree that the literature is a mess, and its messiness is frequently mentioned by researchers. I think I ended up feeling happier about the non-experimental studies than you did -- check out those Trautwein papers, I think they're starting to figure out where to look for this stuff.
Here was my conclusion: "For the youngest kids, homework matters a lot to parents but probably has less of an impact on achievement. Parents are often helpful, but sometimes it can turn into a negative experience at home. To handle that, teachers can assign higher quality assignments that won’t trigger negative interactions. Sounds tough, but OK.
For adolescents, classes that assign homework frequently are associated with more learning. But not classes that ask for more time on homework. And not classes that assign more homework that students think is more challenging. But yes classes where students think the homework is interesting or well-selected. And students reporting they put more effort into homework is associated with learning.
It sounds like teachers need to thread the needle for teens. They need to frequently assign homework that is purposeful – it can’t be busy work – but it shouldn’t be too hard. And there’s a possibility that homework isn’t really as much about practice as it is about taking the class seriously, feeling in control, taking responsibility for their own learning."
Basically: homework will be a net gain if kids actually do it. But when you're dealing with adolescents, realistically there are a lot of ways for them to get away with not actually doing it. (A lot.) Completion rates in most places aren't terrific. So the first question is, how do you get kids to do it and do it seriously? And the second is, what are the benefits of that sort of practice? I think research (along with common sense) supports the idea that there are on average benefits to kids when they do it.
For early elementary reading and math, Slavin & Madden at Johns Hopkins developed “Success for All” and.. they say there is research… maybe it’s hiding. Also it is more than a homework program, it’s class work interlocking with homework in a specific way and classrooms divided by level rather than grade. So it is not directly comparable to “homework only” but it might be something to look at.
Also, the Kumon system.
Even if we were to accept that "homework," as some kind of undifferentiated treatment variable, has some net positive benefit, treating homework as an undifferentiated variable seems pretty useless from a pedagogical perspective. What kind of homework? For what kids? Going to what schools? Under whar kinds of conditions? As compared to what other type of intervention variable?
Between the ages of about 8-12 I had assigned homework that amounted to about an hour a week. I found it too easy and boring, just like most other schoolwork. But at that age I was messy, disorganised, terrible at time management, and good at coming up with excuses to get out of things I did not want to do. Because these things were challenges for me, it was good for me to have to a) remember to take it home, b) keep the pages clean and unwrinkled, c) initiate and complete a task I found boring and pointless, d) do this on a deadline, and e) remember to return it. My parents and teacher let me choose my own strategies to manage this, and then asked me to reflect on whether they worked - and this actually helped considerably with my organisational and task management skills, moreso than imposition of these strategies would have done. I also find it interesting that I can pinpoint things that are still issues for me (task initiation, leaving things to the last minute, losing items) back as far as early primary school.
From my memories of school, the quality and usefulness of homework varied wildly, even within the same subject ("math") and even with the same professor. My naive guess would be that repetition is differently-useful for different lessons in a way that resists sweeping generalizations, to say nothing of different students learning differently.
That said, most of my homework before the university level was crap.
On the plateau model: mathematics tends to be taught as a series of plateaus (plateaux?). You are introduced to a concept like "multiplication" or "sine function" or "Lebesgue integration" and then you work through a zillion examples until it's been thoroughly beaten into your head; you reach the plateau where you finally understand what a sine function is. Then you can move on to the next bit of mathematics, which assumes you understand the previous bit.
Unfortunately the number of examples you're obliged to work through is some constant, insufficient for the dumb kids and excessive for the smart kids. So the smart kids are bored doing the same thing over and over again while the dumb kids never quite reach the plateau and get left behind.
Some other subjects are unlike this. You don't need to understand the history of Ancient Greece before you can understand the history of Ancient Rome. Furthermore, while you can reach the "sine function" plateau pretty easily, no high school student will reach the "Ancient Greek History" plateau. So you can spend an unlimited number of hours studying ancient history, if you like, understanding it more and more deeply, although possibly not in any way that can show up on a high school exam.
My primary experience of homework is that it contributes to my performance in classes if it's one of two categories:
1. homework that strongly resembles material on the test
2. homework that covers information that won't come up in class
Type 1 homework is a waste of time for every purpose except improving my GPA.
Type 2 homework is actively useful (obviously, because I wouldn't have learned that information otherwise). This has been true for my entire life - I remember independent learning assignments as early as first or second grade, and I've always found them more useful than lecture-based instruction. My capacity for self-teaching has improved since I was six, and I can go longer without support/feedback now, but the basic principle is still the same. This is probably less useful if you have unmedicated ADHD or otehrwise aren't much of an independent learner.
(Unfortunately, type 2 homework tends to correlate with useless busywork during class time. The most notable example I've run into was high school AP bio, which was a "flipped class". 100% of my time outside of class - mostly reading the textbook - was useful; 90% of my time in class was spent doing repetitive worksheets or listening to the teacher repeat the textbook in smaller words. Then again, I probably only did 30% of the assigned homework. It was a weird year.)
I guess there's a third category, which is homework with the goal of improving fluency/speed at some task. Teachers love to give this kind of homework, but they're often remarkably bad at designing it. IMO, this sort of homework should include lots of very short problems, instant feedback, and enough surprises that making an excel spreadsheet/python program that solves the problem for you is more time and effort than doing it by hand. Also, my experience this literally only works for symbol manipulation problems (eg the kind of math that engineers do/basic proofs/sudoku). I have experienced many painful homework assignments that attempt to turn interesting real-world problems into fluency drills - this is both painful and probably actively detrimental to learning.
Is the batch of no homework students pulling down the average that much in the final study cited? I'll gladly take the side of "some math homework is useful, and more is technically better on average, but less still represents better than average progress" if it isn't. Less is definitely more of my time back.
Challenge: imagining arbitrary resources were available, but no more cooperation from teachers, students and parents than you'd normally expect, how would you design a study to effectively look at this? How would you control for teaching styles/techniques/skill? Circumstances in student homes? Quality of homework curriculum? Alignment of homework type with subject matter (Algebra isn't Geography isn't writing)? How do you isolate variables we can't agree how to measure?
We struggle to assess teacher quality in general, or genuine variation in student ability. In that context, figuring out how effective homework is seems like determining how good a chisel is without knowing the skill of the sculptor or the quality of the marble.
The post already mentions all of these considerations, but let me come out with an anecdote.
I was a somewhat gifted student, finishing comprehensive school (9th grade) with the highest possible grades (outside of P.E. and art classes, anyhow) even though the last time I did any work whatsoever beyond showing up was during the first grade - before I figured out I didn't need to do any. I continued on this route in high school and while I maintained top grades in some subjects, my grades in mathematics, physics and chemistry tanked (more specifically, calculus was my breaking point and upon being introduced to the topic, I received a failed grade), and I dropped out of school. I've since continued studying in university and I'm receiving good grades, but this time I'm doing the bare minimum and subjectively, I would expect to fail most courses if I didn't do that much.
What I gather from my experience is the following: Homework is useful, but only up to a point dependent on the individual and the topic. For me, at least as far as the granularity of school grading system goes, there was no use up until high school, but at high school level failure to do any work was sufficient to drop the math grades from excellent to failed. Presumably, other students might find homework to be the difference between success and failure at an earlier or later stage than I did. But perhaps more crucially, it has the additional effect of building up work ethic: to this day I find myself unable to cope with a workload of doing more than half of the number of courses expected of students, and I can only assume this is because I never did any work prior to university. This alone seems like a good justification to have mandatory homework, although it would be better still if homework wasn't seen by some students as arbitrary punishments handed out in child prisons, but gifted students could skip grades to the point where doing homework is actually required of them.
The way the problem is posed (like so many of these sorts of reviews) is problematic.
What matters is not just how well homework does (or does not work) cross some aggregate of all children, but what works optimally, ideally for each child and, when that is not practical, for whatever subgroups are easily isolatable.
I would say, for example, that my experience (unlike Scott's) is that I learned a lot from homework, more so than I would have from whatever the alternative on offer is (either "just listen in class" or "read the chapters of the textbook"). Some of that may be my idiosyncratic personality and "learning style", some may be my school, but *I* would be very reluctant to throw out homework until I'm presented with what looks (to me) like a better alternative.
Certainly some (not all, not most, but some) of that homework was a waste of time insofar as I learned things I didn't much care about at the time and have forgotten. But that's a different claim from the efficacy of homework; and I'm not sure how we resolve it given that kids (even at college age!) often are really not sure what they will ultimately find interesting and helpful (let alone secondary social goals like insisting on the teaching of the "second" language in a bilingual society).
It's interesting that everyone is uncertain how well academic homework works, including myself, but I don't think anybody doubts that if one afternoon you showed a group of ten-year-olds how to throw a baseball for 40 minutes, and that night one of them went home and practiced throwing a baseball for two hours, the next day that one kid would be the most likely to have improved at throwing a baseball.
Homework *sucks*, and I personally find the prospect of inflicting it on my future children abhorrent (they have one youth, and *that's* what they're going to spend it doing?), but on a purely competitive level, those who put in the hours at any given activity are in general those you can count on to rise to the top performance-wise.
Sometimes the point of repetition in the form of homework is not to be able to answer the question at all, but to be able to answer the question quickly and easily without having to think about it. It's more like training than education; the point is to develop unconscious competence so that you can just do the task automatically. I've been a great math student, but I tend to do mental arithmetic relatively slowly; I always came up with tricks to compensate for a lack of immediate recall, so I never got to the point where I could multiply any two single-digit numbers instantly. For example, if you ask me to multiply 8 by 7, I might not remember immediately that the answer is 56, so I'll remember that 7*7 is 49 and add 7 to that to get 56. In second grade I discovered that I could calculate 13 - 7 by calculating 10 - (7 - 3) instead of by knowing that the answer was 6. When school math did become more complicated than arithmetic, being clever with numbers and manipulations was a big help, but at the level in which memorization is the fastest algorithm I didn't quite keep up.
Haha I love this post. "There's 1 fact, and that's the way it is. Goodbye now."
I think it also depends on what you are targeting. Homework is good for practicing a particular action so it becomes easier, and memorizing things. I'm less convinced that it is useful for understanding concepts.
50 years ago, the ability to memorize facts and be good at executing certain repetitive action was useful in some cases (math, writing) but I would argue it is far less useful now. I think in modern society, computers/internet/automation have removed a lot of the value in memorization and mastery over repetitive tasks. The way I'm able to be a high functioning member of society is specifically by *not* memorizing things (wastes time and mental resources) or being good at repetitive tasks and instead I'm good at looking stuff up on the internet and knowing how to ask WolframAlpha good questions.
If we want children to be good at taking closed book tests, which is completely non-analogous to the modern real world, then sure, homework may help. This feels like Goodhart's Law in action though as test taking just doesn't represent reality at all.
Did they include hours watching TV while doing homework? Homework took me a long time because I did it while watching my favorite after-school TV shows.
I'm going to conjecture that homework in elementrary school has some value if it's set to "overlearn" - what non-educationalists call "practice until you can do it in your sleep" or something - basic skills such as reading, writing and arithmetic (particularly number bonds, times tables etc.).
When you're trying to do 3 x 14 or something like that, not having to "context switch" to work out 3 x 4 really helps, and this is consistent with cognitive load theory in psychology, which as far as I know does have good studies behind it. How do you get 3 x 4 without having to stop and think? Lots and lots of practice. Homework is one way to find extra time outside classroom hours to do that practice.
A school with no homework but a longer school day would probably do exactly the same, there's no reason why the practice has to be done at home. But it does have to be done.
From my own school experience, homework was a mixture of practicing things we'd already done in class, reading new things in preparation for next class, and busywork. The latter of course doesn't have much learning value.
Back in the dear departed days of psychedelic experiment, there was a phrase to describe the most important predictors of the experience of a proposed 'trip': Set, and setting. Set, the person's emotions and attitudes to such a possible experience. Setting, the surroundings - kind of surroundings, comfort, trust in anyone there to support you. 'Bad trips' (unpleasant or disturbing ones) could be largely avoided by personal confidence, and pleasant surroundings with trusted friends.
I suggest this is a useful way of thinking of investigating the experience ('outcome') of homework. It becomes evident that in particular 'setting' is crucial - a noisy, chaotic small apartment is not conducive to calm 'homeworking' - no matter how long spent on it. I'll bet one factor not even thought of by the investigators was hunger. I'll bet not one of them had ever feared for security of food, or indeed been hungry beyond a half hour beyond usual mealtime.
In other words, cliched experimental designs give cliched results - useless for educational technique guidance.
If I understand the study, "the students with more homework" did *not* do better in an absolute sense. They improved their baseline scores more.
The public policy issues with homework are almost never studied. Instead, people diddle with this nonsense as if it matters, and oh, lord, the elementary school parents talking about it! Jesus, people. Homework at that level is irrelevant, because nothing in elementary school matters, gradewise.
But the public policy issues with homework dwarf the piddling possible improvement it may or may not reveal. Some of you have mentioned this. At the high school level, the only relevant issue with homework is that it substitutes compliance for ability. It's very common for kids with high abilities to get lower grades than kids with lower abilities (and also similar abilities, which is just as idiotic).
Grades in America are a fraud, and we are now in a world where these fraudulent grades are the only legal tender for college. It's revolting. Meanwhile, people spend time and energy wondering whether a kid who counts on his fingers will get one more problem right out of fifty if he spends 30 minutes a night laboriously practicing.
Crazy. Michael, I thought your summary was fine and very clear.
Parents of elementary school children: homework is irrelevant and you shouldn't waste a single second worrying about it. If your kid doesn't understand classwork, find a shrink to see what LD he or she has.
Parents of high school kids: do whatever you can, including bribery, to get your kid to do homework. Do not make it a moral issue or a learning issue. Make it clear that it's bullshit, asinine hoops they need to jump through to keep their grades up.
If, on the other hand, your kid copies homework just to get it done, judge them severely and wonder where the hell you went wrong. The first kid is far superior.
On the third hand, if you have a kid who just does the homework even though he or she knows it's bullshit and genuinely enjoys ticking off tasks, well, they're weird but it makes life easier for you.
I don't give homework. I turn anything important into classwork. I write it about it a lot, but here's The Rules that Matter:
https://educationrealist.wordpress.com/2018/10/26/homework-the-rules-that-matter
(don't know if the formatting will work)
I. Teachers assign homework or they don't assign homework.
A. If they assign homework,
then they grade homework or they don't grade homework.
1. If they grade homework,
homework is essential to understanding or it's not.
a) If homework is directly essential to understanding,
then the homework grade is the same as the assessment grade.
b) If homework is not essential to understanding,
then the homework grade is corrupting the assessment grade.
(1) If the homework grade is corrupting the assessment grade, then the school's administration will prevent this or accept this.
(a) If the school's administration prevents grade corruption then the
public will scoff.
(b) If the schools administration accepts grade corruptions, the district will
prevent this or accept it.
(i) If the district prevents grade corruption, see I.A.1-b-(1)-(a) (ii) If the district accepts grade corruption, this is a typical school. 2. If they don't grade homework, then students won't do homework.
B. If they don't assign homework, the public will scoff, and policy makers will scowl, but that's why god created tenure.
Well, if you want my personal thoughts:
I've always hated homework and frequently didn't do it when it was assigned. The particular details of when I would or wouldn't do it are complex.
In high school I ended up doing all the assigned math homework anyway, and in college I did much less than that. And it seems pretty clear to me that this directly impacted how much material I ended up retaining; the quality of my learning is much lower for the college material.
So I would say that yes, practice matters in math. On the other hand, I would estimate the amount of homework that's necessary to fully learn and retain "9th grade algebra" at zero. It's not at all complex and doing work further along in the curriculum will provide more than enough drill in the basics of algebra.[1]
Moving back to the topic of the post more generally, I would bring up the result that student performance in math generally tracks "exposure to math in school" pretty closely, whereas performance in other subjects (and most notably reading) does not. I find the argument persuasive that this is mostly due to the fact that students rarely do more math than they are assigned in school, whereas many students do a lot more reading than they are assigned in school.
[1] Another thing that seems obvious to me is that different people will require different amounts of practice to learn and to retain the same material.
I think that homework is important, because to me it's practice; you wouldn't say to a kid "did you learn that music piece/recipe/sport?" and have them say "Oh yeah, we did it in class" and then "well have you played the piece/cooked the recipe/played the sport?" "No need, we did it in class, I remember it!"
We see the value of practicing/trying out practical skills, I don't think mental ones are much different. For primary school, leave homework until they're older (7+) and only a little, because at that age it's (a) getting them into the habit of having a routine and doing it and (b) physical practice like writing so they get the skills.
I don't know how much homework secondary students get in America (or indeed what it's like in modern Ireland); in my time coming up to the Leaving Certificate (final year national exam for ages 17-18 school leavers) we were recommended by our teachers to do an hour per subject study every night. You could be doing 6-8 subjects, so I don't know how many people did spend 6-8 hours every night studying. But you do need *some* time studying and not just relying on "we did it in class and I remember it".
I agree that "amount of time spent on homework" does *not* correlate with "results on tests" except very crudely; good students may spend more time on homework because they are studying, and good students will probably do better on tests. But you have, as pointed out, smart students who skip or spend little time on homework, who will do well on tests anyway, and poor students who spend a lot of time on homework because it's hard for them and will do not so well.
https://www.lesswrong.com/posts/vNBxmcHpnozjrJnJP/no-one-knows-what-science-doesn-t-know (sorry, I couldn't resist)
(Just for clarification, I am German so that is the "your native language"-subject.)
I spent less time on homework in German (the subject, I mean) than in math but the reason was not that the German homework was less, it was actually more. I just didn't do it because it felt like it was too much and very not fun to me while math was always fun (either it was easy or it was like solving riddles designed to be easily solvable). I think another reason was that you get a lot of "I finished part of it" rewards in your head, since Math homework was usually a bunch of different tasks while German homework was usually one big essay kind of thing. I did English and French homework more regularly too and that was usually a bunch of little tasks and not one big essay.
Those students who did more algebra homework probably had less time to study other subjects. A good study would adjust all homework.
I’m a professor, and I research (among other things) learning and the development of expertise. I think that much of the (very interesting) discussion seems to ignore or downplay comments about the quality of studies in this area. The lack of good studies on HW specifically should motivate us to rely more on theory and on related literature (and improve studies).
We have lots of studies that are more theory driven, and we do have first principles. When we ask a question like “is homework effective?” What is the assumption about when students will practice the concepts that they have been exposed to in class, if not via HW? Just theoretically, if we had a complete dataset on the exact amount of time that every student spent doing HW, is anyone hypothesizing that in elementary mathematics (say), there is no relationship between the amount of time students spend practicing a task such as adding fractions with unlike denominators and (a) their accuracy on a test of that material, and (b) their facility / speed given a consistent level of accuracy (i.e., expertise) with which they would be able to execute concepts such as these? (Controlling for other relevant factors such as ability.) In other words, is the conclusion of the “anti-HW”/“HW is useless” crowd that, given a perfect dataset with full controls (or random assignment and perfect measurement in terms of time and attention), deliberate practice on tasks like learning times tables, adding fractions, working with percentages, word problems, etc. would have *zero relationship* with expertise as measured by tests?!?
If that were true, it would be very surprising, to say the least. In laboratory experiments where people are required to learn things, practice (and attention - “deliberate practice”) and time on task are strong predictors of learning. I used mathematics as an example, but if the concept were “knowledge of history”, or “knowledge of geography”, it does not seem possible that this relationship would be zero. Learning to write well involves writing, then seeing feedback on that writing (and preferably, a chance to fix whatever errors were made, without punitive grading), etc. All of these practice opportunities cannot be fit into a class day without sacrificing instructional time.
What many commenters seem to be getting at is that much elementary school HW is a waste of time in comparison to the hypothetical HW that I have postulated (one person mentioned coloring assignments, but there are so many “silly” things, many of which are really “parent assignments”). I do agree with other commenters that some of those comments are too strong, because it implies that academic skills are the only skills that elementary school is cultivating, which is untrue.
As a final point, the academic literature in Education does not tend to be the strongest in terms of statistical or analytical rigor. As just one example, the analyses in most of the HW literature fail to account for attenuation of the focal variable due to the (gigantic) measurement error with which HW time is measured (see https://en.wikipedia.org/wiki/Regression_dilution if this is unfamiliar to you). This is one of many similar issues that plague this kind of work, and which may or may not be affecting our conclusions.
Scott, your post is about about the need for meaningful, systematic data and the woeful, pervasive lack of that data. Therefore, I'm going offer an anecdote from my own personal experience, which of course will totally, definitely resolve the issue once and for all (for reference, I was born in the early 1970s and had the standard range of k-12 public schooling, along with college and grad school):
1. In elementary school, most homework seems to have been busywork to do so the teacher could say they assigned homework.
2. Exception to #1: Beginning around 5th or 6th grade (maybe as early as 4th grade?), we were assigned short stories to read and book reports, etc. There wasn't enough time in the class day to read in class, so we had to read those at home.
3. Another exception to #1: Well, it's only a partial exception. Even when it was busywork, it taught me a measure of self-responsibility which ultimately was, for me, empowering. I learned to schedule and to plan. I probably overlearned those skills and became a little obsessive about it. Or a lot obsessive. It also made it harder for me in college and, especially grad school, when I had to learn to put things aside for a while and focus on other things because there was too much to do all in one night.
4. In middle school and high school, homework was sometimes busywork, but just as often necessary to learn what I was assigned. With some subjects (history, French, English lit), I really looked forward to it. With other subjects (math, physics, chemistry), I struggled through the homework and didn't learn anything.....until I had a trig and calculus teacher, in my junior year in high school who had a different approach. She assigned the problems first and had us teach ourselves how to do them. The next day, we went over the problems and what we didn't understand from the textbook. I, for one, loved this approach, because it taught me to take control over my own learning and not depend on the teacher's lectures. (Of course, that approach didn't work for everyone.)
5. In college and grad school, homework was most of what I had to do when it came to learning. The lectures helped. I wouldn't've been able to get by by teaching myself and reading. But (ugh!) graduate seminars (where we all talked as if we knew what we were talking about, having pretended to read the first paragraph of the introductions of any given book) did very little and it was the homework. And, as I said above, the "responsibility" and work ethic I learned from my early years of homework made it difficult to triage.
Before I read this post, I pre-registered my thoughts about homework. Updated thoughts are at the bottom.
How important is homework?
I am sure tat if you look at large aggregate data, it will show low effectivness for homework with respect to learning an future earnings.
So where should you look?
First a definition.
Homework is type of practice of the concepts, processes, and skills one learned through instruction. It’s self-guided exploration in the zone of proximal development.
Second, some comparisons.
Singapore, Finland, South Korea all have homework; they are the three of the best education systems in the world, maybe they know something you don’t.
Third, the best interventions we know of in schooling are ability grouping and direct instruction. Both of these interventions would change the nature of homework. Turning inconsequential “practice” flailing about completing worksheets with deliberate practice that is up to the level of the student.
Fourth, I run two schools which only give homework two days a week and have three days a week of in-person instruction. We consistently achieve above average on standardized tests and one consistently rank in the top 20 of schools. So, I both am prone to belief that practice is important, and that the quantity of time spent in school is gratuitous.
Given these facts, I have the following hypotheses:
1. Homework will be as effective as one’s practice regimen.
2. Homework will be effective if it employs the right quantity of a student’s understanding.
3. Homework is necessary, because learning happens in the space between one’s ears. If one is not doing something with the brain, it is not learning.
Maybe these are uncontroversial and nearly tautologically true?
---
Update:
Scott brings up a question I haven't thought much about. How does homework's importance change with time? At the same time time, he admits that's practicing algebra problems probably makes one better at algebra. To the question about hand, I don't think the answer changes much. When I watch the curriculum closely, I notice that fourth grade homework is especially focused on memorization and dozens of 'microskills' like placement of punctuation, building up 3 digit multiplication, and the "states and capitals" world context stuff.
While we don't know the value of homework in the RCT Sense of 'know', I will still put my confidence levels in the value of homework as high as my confidence levels in the value of unguided deliberate practice.
About halfway through my engineering degree in college, I realized that if I wanted to get good grades, I either needed to attend class or do the homework. This realization served me well till I graduated.
Regardless if having homework for elementary schoolers is conducive of their learning, it does seem useful to have it at all if doing homework is useful for high schoolers, right? I had barely any homework in elementary/middle school and remember having a culture shock early in high school when I realized I couldn't finish all my work in ten minutes anymore.
I know it's only anecdotal evidence, but when studying partial differential equations in college, it was not enough for me to attend class and do the assignments; i had to spend significant extra effort doing all the problems at the back of each chapter to really learn the material well enough to pass the course. I was motivated to do that and ended up with a B in a class I was initially failing. So at least in this one case, 'homework' made a big difference.
For how important education is, I find it amazing that we don't seem to have more research on this.
I am a fan of this talk https://www.youtube.com/watch?v=zDZFcDGpL4U
Public school teacher here. I’ve observed the current trend to be moving away from assigning homework. In my master’s program for teaching, we were told homework was not equitable because not all students have the privilege of free time or the proper environment at home. Honestly, I’m not sure what to make of that. Many of my high school students are not motivated by failing grades. Like others have said, if they don’t want to do it, they won’t. Granted, some subjects do require more practice and time, so I don’t see a way out of some homework. Like others, I developed my own intrinsic motivation to learn in university. In education research, the studies are quite pitiful. Over time, I’ve just settled on the belief that you can’t force anyone to care about anything- especially youth existing in the current state of public education.
I wonder if the main benefit of homework for elementary school students isn't directly academic, so much as developing good habits.
Which would allow them to actually benefit more from homework later.
But good luck disentangling all the variables.
I doubt that there are trustworthy data sets that accurately track student homework loads over that span of time.
Mostly in response to some other users:
1 - Even if younger students don't benefit as much from HW, you'd be hard-pressed to start instilling the habit starting in their teens. That's a losing battle for parents, it's like asking your kid to do chores starting at age 14. Extensive amounts are probably redundant, but I'd foresee a benefit to their assignment in the long-run.
2 - Some have shared anecdotes that they performed well in school despite ignoring HW wherever they could. This doesn't account for study time ahead of tests and exams, which, for classes like math, amounts to the exact same thing as HW. I've seldom encountered people make the claim that they excelled in school despite doing no HW *and* not studying. Those that have likely possess superior focus and memory skills.
Understanding the concepts is one thing, but retention (let alone of the details) can be fickle, and benefits enormously from repetition. It's an important foundation for understanding and applying more advanced concepts in layers. I think it's a mistake to blow up the importance of long-term retention such that performance in tests is then thought inconsequential. The skill instilled from study or hw is probably more important than retaining some abstract concepts in hs-level STEM. If most of us don't use calculus in the long-run, difference in learning outcomes (if outcome is measured well after the fact) in that domain is negligible; we barely remember how to put it to practice. Use it or lose it. Basically everyone keeps writing and speaking their native language beyond school, that doesn't usually erode. If I pick up my guitar now after having it collect dust for several years, I can recall scales and a few songs, but what I can muster out of it is tantamount to doing basic arithmetic. Goodbye Paganini and Eruption solo. What are the concepts doing for me now, in isolation? If you only remember coles-notes versions of everything you're still basically impotent in those respective disciplines.
(a propos writing: many novelists will tell you that writing is a craft, and there's no magic behind it. If you hack away at it for hours a day, you improve)
I haven't looked at all of the comments, so someone may have already gotten here and I'm late to the party. But it seems to me that one solution to both the homework problem and the diminishing-connection-to-our-IRL-communities problem is to make homework an in-class team task (a type of flipped classroom) that has to be completed by everyone in the class or credit is not given to anyone. The students would have to be given support from their teacher in how to troubleshoot issues like a kid not being able to do it, so they would learn a lot about different capacities and the ability to support others on a team IRL. Yes, it requires a lot of teacher training, but it's what they are missing the rest of the day (on their phones) and enforcing it in the classroom may be the only way we heal that.
Fascinating question.
Homework consists of certain activities. Are those activities better performed (better learning) with:
A) No teacher available?
B) With a teacher available?
I can't think of any advantage to A) other than not needing to pay the teacher.
Why wo
If the goal is to filter students, not to make them "smarter" or "more educated," then more homework helps select for independent learning
This appears to assume the the objective of schooling is student learning.
Tangentially, regardless of the inconclusive evidence (or trusting of the mediocre evidence?) my teacher training program emphasized that it *doesn't* help and urged us to give homework sparingly.
Even the algebra study is based off a very specific counterfactual of homework as it's generally done in algebra classes, where the teacher lectures the whole class and students only get practice doing it on their own. It would seem much more logical to completely invert that, where students watch a lecture on their own time with the ability to fast forward through or rewatch parts as they deem necessary then go through exercises during class time where they can get help from the teacher and classmates when they get confused.
Art of Problem Solving (https://artofproblemsolving.com/) changed my feelings about homework -- they run these amazing math (and two programming) classes where you are in class for only 1.5 hours a week, but mostly you learn through doing the excellent problem sets, which both build skills in the material and give practice in doing challenging problems, and are fun problems to boot. If all homework were like this, I'd be in favor of it. And it also clarified to me that in the best case, homework can act as a practical way to build skills, just as people have brought up music or sports above.
Though as it is, I'm glad my kids' school doesn't give homework so I can have them do AoPS and Beast Academy (for lower grades, run by the same people) instead.
This is actually something I've thought about quite a bit while I was teaching introductory chemistry and physics at a private high school (7 years before transitioning to software).
First, I have yet to see a *non* garbage educational study (coming from a hard science background). Small sample sizes, buckets of confounding variables, ad hoc "controls" or no controls at all, studies that either don't replicate or (most commonly) don't scale at all beyond the small focused lab schools, etc.
"Time spent" is a horrible measure of anything. First, because it's horribly unreliable. It was a running joke that students, teachers, and parents would all estimate completely different amounts of time for the same set of homework. Part of that was that students would include things like "I have the book open, but am snapchatting with my buddies". For tests, I generally used a 5-6x modifier--if I could do it in 10 minutes, I'd expect it would take the median student 50-60 minutes.
Beyond that, I actually ran an experiment (because I was lucky enough to have an administration that gave me carte blanche). For the last few years, I didn't assign "home"work. Instead, I did a few things:
1. Practice time and practice sets in class, with me walking around helping people. If you didn't finish, you'd have to do so at home. But they were scaled to take up the class time.
2. Those practice sets were *irregularly* graded. But rarely. Instead, I did a combination of two assessment methods between major summative assessments, both fairly low stakes (~5% of the total grade, with the median effect being ~1% either way).
2a. Pop quizzes on the previous-day's work. 3-4 questions, 10 minutes. Basically taken from the problem set the previous day with numbers changed. Graded right there in front of everyone, usually by a peer as we'd walk through the problems in detail.
2b. "Online Practice Sets", which involved a tool I wrote to generate a pocket static website that presented questions (multiple choice and fill in the blank, mostly) and then auto-graded them (including some allowed variance for numeric questions). They could do them as many times as they wanted (including later on to prepare for the tests), and the system would only report the highest score. The problems and answers were randomized (in order at least). If there was a pop quiz, it would often come from those problems.
I changed because I realized (both from surveying students and my own observations) that traditional homework ended up in one of two buckets--
1. Common enough to be useful...but then the grading load was too much so they didn't actually get feedback and it ended up being graded on completion. This was utter garbage for everyone--no one took them seriously, including myself.
2. Rare enough to have time to give meaningful feedback and be scored properly. These, in turn, ended up being basically just another summative assessment. And weren't frequent enough to actually provide any reinforcement.
So switching to a basically "no homework, but lots of practice embedded throughout" system was win-win. Can I say that it had substantial effects on performance? No, because we also were revamping the curriculum fairly constantly (one of the reasons I left was that we were just always chasing the new hot thing instead of actually measuring whether the changes were doing anything useful). But it was way more honest and the kids liked it more and were more engaged. Or had fewer people to blame when they weren't. And allowed me to target more practice for some things than others and let them redo their practice. Plus, the panic on their faces when they'd have a pop quiz (even when it was announced) was beautiful to my jaded, evil soul.
Bear with me here...
Say I wanted to improve some practical skill, e.g. batting in baseball (a relatively intricate motion required coordination, full disclosure I don't play baseball). I expect it would be hard to ask a question like if I swing faster will I bat better in a single experiment/study.
1) A sudden shock to the learned motion most likely will make things worse before they get better (at least, this is my personal experience).
2) To accomplish this goal I will need to adjust other elements of my swing, it is hard to isolate one element at a time...
To assess whether batting harder is better I would instead consider the mechanism (yes, hitting things harder make them go further) and some empiricism (yes, athlete that bat further hit harder).
This strikes me as similar to the case for homework and teaching in general, the reason that we believe that homework improves performance is that there is a clear mechanisms and successful teachers seem to have settled on a certain balance of homework. The type of study Scott seems to be looking for seems to be inherently fraught, like asking a someone to suddenly change their swing and concluding something from the experiment. Perhaps (speculating here) this is why there are not so many studies of this flavor.
Has anyone asked whether the student is motivated to learn? I suspect that homework might be of value to a student wanted to learn and didn't get the concept or have the vocabulary mastered yet.
When I was teaching myself foreign languages, I assigned myself brutal amounts of homework and forced myself to do it because I wanted to learn. I don't think I could have otherwise developed whatever capabilities I did in fact develop.
When I was an undermotivated student, which was most of the time in any sort of coursework, homework was just a hoop I had to jump through, and cats hate jumping through hoops.
Thanks Scott, great post. Here are a few other RCTs that might add some additional nuance on the question:
1. "Variations of homework amount assigned in elementary school can impact academic achievement"
Link:
https://www.tandfonline.com/doi/pdf/10.1080/00220973.2020.1861422
a. Abstract:
"The utility of homework and its impact on academic achievement has been an ongoing debate for more than 100 years. To date, there is no rigorous experimental research to show whether, and the extent to which, variations in the amount of homework assigned can impact the academic performance of elementary school students. In this study, 440 second grade students were randomly distributed in 3 groups within the classes they attended. Each group received different amounts of homework in writing and math for 20 days. The results showed a significant immediate effect of homework quantity on writing competency (but not on math competency). The writing homework effects were sustained 4 months later, but only for the group that had allocated a moderate amount of homework to writing skills practice. This study shows that the additional opportunities for practice offered by homework can have differential short-term and medium-term effects on the academic achievement of elementary school students and supports the assignment of a moderate amount of writing homework at this age"
b. Commentary
I can understand not including this one in your review.
There's not much to like about the design of the study, or the figures within the writeup (both are overly complicated, particularly the figures).
Still I did their findings (and underpinning theory) plausible for 2nd grade kids, and so for all completists, note this RCT and its takeaways.
2. "The Role of Homework in Student Learning Outcomes: Evidence from a Field Experiment"
Link:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1592889
a. Abstract:
"In this article, the authors describe a field experiment in the classroom where principles of micro- economics students are randomly assigned into homework-required and not-required groups. The authors find that homework plays an important role in student learning, especially so for students who initially perform poorly in the course. Students in the homework-required group have higher retention rates, higher test scores (5 to 6 percent), more good grades (Bs), and lower failure rates. The authors also study the relationship between endogenous homework submission and test performance using instrumental variable estimation and find that homework submission has a large positive effect on test performance."
b. Commentary
The gains appear real, and while modest also meaningful, across outcomes.
It was also striking how much homework students in the “no-homework” group opted out of; making it a requirement changes behavior to a much larger extent than I would have expected in this setting (college).
3. "Online Mathematics Homework Increases Student Achievement" (Michael Pershan mentioned this in an earlier comment.)
Link:
https://journals.sagepub.com/doi/pdf/10.1177/2332858416673968
a. Abstract:
"In a randomized field trial with 2,850 seventh-grade mathematics students, we evaluated whether an educational technology intervention increased mathematics learning. Assigning homework is common yet sometimes controversial. Building on prior research on formative assessment and adaptive teaching, we predicted that combining an online homework tool with teacher training could increase learning. The online tool ASSISTments (a) provides timely feedback and hints to students as they do homework and (b) gives teachers timely, organized information about students’ work. To test this prediction, we analyzed data from 43 schools that participated in a random assignment experiment in Maine, a state that provides every seventh-grade student with a laptop to take home. Results showed that the intervention significantly increased student scores on an end-of-the-year standardized mathematics assessment as compared with a control group that continued with existing homework."
b. Commentary
I can also understand not including this one in your review, as it’s NOT a "to homework, or not to homework, that is the question" study in the strictest sense. But an interesting comment on what *kind* of homework works.
One notable finding is that it helped low-achieving kids more (same as study 2). Effects were ~0.29 for kids below the median, ~0.12 for kids above it.
In contrast to study 1, and I know this is a matter of taste in some respects, I really liked the study. No deception or tricky maneuvering, as far as I can tell.
4. "Disrupting Education? Experimental Evidence on Technology-Aided Instruction in India"
Link:
https://econweb.ucsd.edu/~kamurali/papers/Working%20Papers/Disrupting%20Education%20(Current%20WP).pdf
a. Abstract:
"We study the impact of a personalized technology-aided after-school instruction program in middle-school grades in urban India using a lottery that provided winners free access to the program. Lottery winners scored 0.37σ higher in math and 0.23σ higher in Hindi over just a 4.5-month period. IV estimates suggest that attending the program for 90 days would increase math and Hindi test scores by 0.6σ and 0.39σ respectively. We find similar absolute test score gains for all students, but much greater relative gains for academically-weaker students. Our results suggest that well-designed technology-aided instruction programs can sharply improve productivity in delivering education."
b. Commentary
Whether this is "homework" or "an after-school program" is a matter of debate. I think 85% former and 15% latter in my interpretation. The software just happened to be really good, targeted homework.
I also think the broader context of education in low and middle income countries might be useful here. Schools, by and large, are staggeringly unproductive in generating learning relative to those in high-income countries. (More: https://blogs.worldbank.org/education/global-education-crisis-even-more-severe-previously-estimated)
Case in point in this study: kids in the bottom tercile in the comparison group literally learned nothing at all between baseline and endline:
"Students in the treatment group see positive value-added in all terciles whereas we cannot reject the null of no academic progress for students in the bottom tercile in the control group.."
So homework, in this case, might be one (of many potential?) piece of low-hanging fruit that could get kids to learn more than they otherwise would in a low-productivity setting.
*
Summary
-These four studies, plus the algebra one Scott cites, tend to support that homework *can* work, under some conditions.
-The key variables, by study, appear to be (i) for the younger kids in Romania, giving them novel contexts to practice already-learned skills, (ii) for college kids at East Carolina State, making them do the homework (and make it relevant to success on exams), (iii) for the kids in Delhi giving them instruction at their level, or really any instruction at all, (iv) for middle school kids in Maine, giving them some “hints” with well-aligned practice problems so that they don’t get stuck.
Common theme in all these winners was that homework was either marked (by a computer or human). I suspect that matters quite a bit.
-Highest RoI, subject-wise, seems to be in math, though it might just be relatively overstudied (all 5 of the studies included math or math-adjacent stuff), and there were high returns to Hindi as well in the Mindspark study, and no impact at all on math in Study 1.
-Highest RoI, distribution-wise, seems to be with lower-performing pupils, though caveat emptor: my guess is that they are also unusually responsive to feedback; simply assigning low-performing kids homework and not correcting it may backfire.
I wonder if non-teachers are aware that there is a growing anti-homework movement that's equity motivated. You can check out Grading for Equity by Joe Feldman if you want to get those arguments in one place.
Lots of teachers are encouraged to both drop the % that HW makes up of the total grade and also award half credit even when students don't do the work. I'm of two minds of the approach. It clearly reduces failing grades, but I'm not sure if that is helping people learn more. You could make the argument that homework is more about compliance, but then it's not always clear what a grade is supposed to represent. Is a grade the record of your ability to meet a specific standard, a measurement of your growth in learning, or an assessment of the effort you undertook? If we had a perfect way of measuring those things, which would we choose?
I‘m a (new-ish) teacher and at university we heard that John Hattie was the go-to guy for statistical analysis which aspects/methods/… work and which don’t have an effect on furthering education. His meta-meta analysis „Visible learning“ combines 960 meta studies and encompassing 260 million students to check out everything from class size and teaching method to TV hours at home and nutrition.
The median effect over all of them is d=0.23. So basically most things have some kind of positive effect on learning. The goal for teachers then should be to find those things that are better that most and concentrate on those. So he suggests taking 0.23 as „helps just a little“ (and below that as „hardly helps“. Effects over 0.4 are labelled „works well“.
Homework is ranked 88th, d=0.33. The meta-analyses were
- Pascal, Weinstein & Walberg, 1984 (15 studies, d=0.36)
- Cooper, 1989 (20 studies, d=0.21)
- DeBaz, 1994 (77 studies, d=0.39)
- Cooper, 1994 (17 studies, d=0.21)
- Cooper, Robinson & Patall, 2006 (32 studies, d=0.48)
Anyway, my take-away is that I mostly don’t bother with homework. The kids who understand the subject in class waste time doing things they already know. The kids who don’t know how to do the task don’t get any help so they will hardly get better. More time spent doing the thing will increase capabilities a bit but I have to be conscious of my pupil’s private life as well. They need time to learn the piano or play soccer or be kids or whatever. There’s so much learning possible if I don’t waste their time.
That said (as a primary school teacher), they will have to learn to read and write. There’s no way around that, if you don’t practice reading at home it will take you soo long before you get your 10.000 hours or whatever in school. So I‘d argue if there is ever a place for homework then it’s daily reading and writing in first and second grade.
Well, that's a sad end to my smugness right here! I am (was) pretty much the only teacher I know who read Hattie and tried to at least keep some of his results in my mind at times. Everyone else just does whatever they feel is "right" - I went with "science", and lost (time and maybe intuition).
I will poke around your suggested bestevidence.org but many misunderstandings lie this way. I'm German and my German Hattie edition had a lengthy preface detailing that you can't just translate the words as educational concepts in both countries differ quite a bit. E.g. how to teach "reading first grade" may be quite different so there's a lot of additional research involved.
Anyway, thanks for setting me straight, Michael.
One would have to start with a typology of homework. Some learning outcomes need skill drill. For others (too much) skill drill is harmful. Finding the right balance between skill drill and free exploration is crucial in so many areas of learning (whether sports, music, maths, etc). How to get this balance right is the hard question for a teacher.
And without putting this balance between skill drill and free exploration into focus there is, in my opinion, no hope to assess whether homework works.
Btw, some people seem to be able to learn a subject without skill drill. In my experience, this is only the case because they integrate skill drill into their free exploration. (I first noticed this when listening to a Dennis Bergkamp interview where he described how he learned to play football (soccer) as a kid.)
Btw^2, if you want to enjoy the sublime skills of Bergkamp see eg https://www.youtube.com/watch?v=mvREhmTptSc
As someone who always hated homework, and spend countless train rides in the morning and countless breaks in between lessons to copy homework from others, I fully support your conclusion! Very grateful to those who always let me copy their finished work without ever complaining. Homework didn't count for our grades, but not doing homework would lead to different kind of punishment, including prominently to have to stay in school for some additional hour(s).
On a more theoretical note: Even besides the evident shortcomings of cited studies I think the question of 'does homework work' is grossly underspecified. As already mentioned in other comments, it's probably much more 'which type of homework' 'for whom' and 'under which conditions'. With 'which type of homework' not being algebra vs. French as a foreign language, but the specific type of exercices in each of those.
I've always sort of assumed that the value of homework came from two things: First, that school is over at 3pm or whatever, but the longer you spend thinking about something the more of it you retain (80,000 hours and whatnot); and second, that working on something closer to bedtime (than 3pm) makes it something that is more reinforced by whatever the mysterious reinforcement mechanism of sleep is.
I say "assumed". I have no more idea than anybody else. But it is enough to make me scowl when I hear about kids not doing homework any more, and not *just* because I've paid *my* dues.
The skill my kids seem to have learned by doing homework is how to work independently (compared to me who wasn't set homework for most of my school career). This has spilled over into them being able to do none school work with more focus and be much better at revising for exams etc. It's probably a useful life skill.
If you look at countries who have extraordinary amounts of homework, like South Korea, they seem to do astonishingly well from an academic achievement point of view. While this might indicate that homework is effective, it is a very high price to pay in terms of happiness and time spent learning on things mostly useless in real life.
Whenever I think about homework I think the one-size-fits-all approach to school is really the underlying problem. The super smart kids in the class that find the material super easy and boring are way more likely to do the homework. They want the good grade and homework is part of the grade. The kids that are struggling with the material and that could use extra help or tutoring are least likely to do the homework. The first group gets no benefit from the homework that they did and the second group that did not do the homework actually could have benefited from it.
There should be a skills based system where based on how you did on the test/quiz determines how much if any homework you get on the topic. If you just aced the test you don't need the homework. If you got a D there should definitely be some homework as you clearly do not know the material as well. Apparently, we don't do this because it would not be "fair".
Personally, I did all "homework" during the school day in other classes as I found school too easy and boring and the homework didn't even being to be challenging. This resulted in homework without the "home" portion.
in this case, it seems like we should default to homework works.
Science is conflicted because not enough good studies, but a priori homework should work (deliberate practice and whatnot) and common sense says homework should work. Maybe a good reason not to have a billion hours of homework, though.
I find the workload at a 40 hr job is less than the workload at an elite 11th-grade high school, which feels wrong.
Studies on learning suggest that repetition of material is valuable for learning.
As such, I feel that it is likely that any sort of repetition of material - be it in the form of reading a textbook then going thrugh the lecture and taking notes on the lecture, or actually doing homework problems based on the material covered in the lecture, or doing a project which relies on the skills you are supposedly learning, or watching videos about the subject matter - will reinforce learning.
Speaking as someone who enjoys writing fiction, repeatedly writing fiction (in the form of fanfiction) greatly improved my skill at writing fiction, and that I started out at a higher level than most people in the fanfiction community because I'd spent years roleplaying and reading books - and while neither of those things are precisely writing prose fiction, they rely on many of the same skills.
I think that there's often a "pool" of skills that surround various things. Writing prose fiction is not the same as reading prose fiction or roleplaying (be it freeform or RPGs), but there's a significant overlap in the skills here, and if you engage in gainful practice (i.e. pushing yourself to improve your skills by repeatedly doing these things) you improve your skills.
This is obvious with art as well; practicing art is how people get better at art. If you look at artists who practice for a long time, their skill levels tend to go up; in some cases, this can be especially dramatic if you catch people very early in their artistic careers. A good example of this would be Miss Mab; if you look at the very oldest comic she posted, back in high school ( http://www.missmab.com/Comics/Vol_001.php ) and then compare it to what she does today ( http://www.missmab.com/Comics/Vol_2084.php ) the difference is stark. Even if you look at a lesser time span within that, like five years between this ( http://www.missmab.com/Comics/Abel_03.php ) and this ( http://www.missmab.com/Comics/Ab_098.php ), you can see significant improvements.
However, at some point, unless you push yourself to do new things, these gains can become increasingly limited. For instance, this comic from 2013 ( http://www.missmab.com/Comics/Vol_1389.php ) and this comic from 2019 ( http://www.missmab.com/Comics/Vol_1896.php ) is a larger time gap, but the gains are clearly not as significant - she was already quite good at art in 2013, and while she did get better over time (particularly at backgrounds) the advances have not been as large.
As such, doing homework is valuable, but doing infinite amounts of homework isn't - there's declining returns, and you get larger returns by pushing yourself further along the road rather than endlessly repeating the skills you already have.
I think this is where people get unahppy about homework. The more skill you already have, the worse the gains will become by doing the same thing over and over.
I suspect this is why homework appers increasingly more valuable as students get older - in elementary school, a lot of the higher end students already baically know everything or most everything elementary school teaches them. As such, elementary school homework is a waste of time for them.
However, lower end students are not nearly as good at this stuff, and as such, will see significantly larger benefits because they don't already know this stuff.
A kid who enters elementary school reading the kinds of books you read in 6th grade or even high school is going to get almost nothing from elementary school English classes.
I also suspect this is why math homework seems so valuable by comparison - few parents teach their children much math independent of school, and particularly not higher levels of it, and people generally don't naturally do much repetition of math at that age, so math homework will seem very valuable. And once you hit algebra, I think that goes way beyond what 90%+ of parents will ever spend time drilling kids on independently.
Conversely, a lot of kids enjoy reading books, so a lot of other subjects are more likely to be covered. I enjoyed reading Zoobooks and various science books, so I had a huge level of knowledge matter on the subject. I even read some textbooks for fun because I was a horrible little monster like that.
This meant that I had a ridiculous advantage over everyone else in science class, even through high school and early college. I knew this stuff back and front and could spend very little time on work and get better grades because I had already put in the work through years and years of independent "study", so the homework was entirely pointless for me.
But if I tried the same thing with college calculus, I was doomed.
So, yes. I think that homework is valuable because it is forced repetition of learning ,but I think that "pre-existing knowledge" is a greatly underappreciated confounding variable.
I think you will see larger benefits of people doing homework in subject matters that they don't independent study than those that they do, and I think you will see larger benefits of homework the more advanced the subject matter is, as that means that fewer and fewer students will have natural background in it, and that you will see larger benefits on lower end students for lower-level materials than you will see for higher-level students on lower-end materials.