, ,


I have much stronger opinions about the best way to educate the children I am likely to have than I do about the best way to educate children in general. However, I understand that an educational reform proposal is an important part of being a prospective homeschooling parent who also blogs, and luckily there do seem to be some obvious pieces of low-hanging fruit. Picking these can justify the effort of homeschooling in and of itself.

For instance, high schools, and to a lesser extent middle schools, should run on Silicon Valley time. There is absolutely no reason to start classes before 10, much less at 7am (!!!), as the public high school near where I grew up did. Teenagers like to go to sleep at 11am or midnight, this is an extremely predictable fact about teenagers, and you do not get millions of people to change their preferences by yelling at them to be more virtuous and have more willpower. Chronic sleep deprivation causes depression, anxiety, behavioral problems, poor memory, and poor concentration (interestingly, these are all common complaints about teenagers). And I shudder to think of the consequences of causing chronic sleep deprivation in such a crucial time for brain development. Please, for the love of god, homeschooling parents, let your teenagers sleep in.

Similarly, many schools have cut recess and physical education to create more time for academics, in spite of the evidence that exercise improves children’s academic outcomes (as well as, obviously, their physical health). Again, this is another easy fix: homeschooling parents can and should ensure that their children have sufficient time for physical activity, including plenty of time for unstructured free play. On a related note, play-based kindergartens appear to outperform academically oriented kindergartens.

The homeschooling parent may also be able to adopt some evidence-based learning techniques which are not necessarily common in the classroom. The two techniques with the most evidence are practice testing and distributed practice (also called spaced repetition). People seem to learn better if they regularly have to recall the information they’re supposed to be learning, such as by using flashcards, doing practice problems, or having to write a short essay without referencing your notes. Distributed practice/spaced repetition is spreading out what you’re learning over time: for instance, instead of teaching about the theory of evolution all in one week, spread out the lessons over several weeks, and regularly return to the concepts to review them. Promising techniques with less evidence include interleaved practice (mixing up problems of different kinds, such as having addition and subtraction problems on the same worksheet), elaborative interrogation (trying to explain to yourself why facts are true), and self-explanation (explaining to yourself why you solved a problem in a particular way or how a fact relates to other facts you already know).

While there’s not much the non-homeschooling parent can do about their teenager’s chronic sleep deprivation, non-homeschooling parents can also pick these low-hanging fruit, although with somewhat more difficulty. Prioritize physical activity and unstructured play in choosing your child’s after-school activities, flee any kindergarten which involves a worksheet, and teach your child to test themselves and spread out their studying over time. Also, if you find yourself interested in activism, please consider campaigning for your child’s high school to start at a reasonable hour.


An interesting question is whether homeschooling tends to outperform non-homeschooling. Unfortunately, most of the data that purports to show that it does is selection bias hell, overrepresenting wealthy and college-educated homeschooling parents and underrepresenting educationally neglectful or just generally shitty homeschooling parents.

However, the Coalition for Responsible Home Education has done some very important– albeit preliminary– research with fewer selection bias issues. (Interested readers may fund less preliminary research here.) Poor homeschoolers tend to outperform poor publicly schooled children in reading and writing and slightly underperform them in math. Non-poor homeschoolers tend to slightly underperform in reading and writing and massively underperform in math.

We don’t have enough information to know for certain why homeschoolers and non-homeschoolers differ. However, my speculation is that poor homeschoolers tend to do better because poor homeschoolers are different from poor non-homeschoolers: for instance, they may be more likely to be culturally and educationally middle-class people who are poor because one parent quit their job to homeschool. The lower performance of children in math seems to me to be a result of the average American’s attitude towards math, namely, hatred, fear, and distrust. Many Americans can barely perform elementary-school-level math, such as simplifying fractions. No doubt this is due to American schools’ appalling math education, but one would not expect high-quality reading education from a parent who struggles reading Goosebumps, and one should not expect high-quality math education from a parent who does not know algebra.

For this reason, I would advise the homeschooling parent to put serious thought into how they’ll teach mathematics. In my case, I’m not particularly worried, because my local homeschooling coop is going to include an absurd number of physics majors, a former math tutor, and an Ivy League mathematics PhD. However, people who are less lucky should consider budgeting some money for math tutoring, perhaps from a local grad student, or a high-quality after-school math program.


I currently favor unschooling as a method of homeschooling, but this is pretty much about traits of my child, rather than traits of children in general. It is pretty much inevitable– given genetics– that any children I have will have their own particular, passionate interests which they are extremely enthusiastic about learning about, and that they will respond to attempts to get them to learn about other topics with something between dutifulness and rebellion. This seems to me to imply that unschooling, which involves following the child’s interests, is an ideal choice: the children will be much happier and I won’t have to spend a bunch of time coercing them into doing well on tests on subjects they are uninterested in.

No doubt this will lead the child to have a remarkably unbalanced education: they may understand everything there is to know about sailing, or Broadway musicals, or ancient Greece, while remaining unclear on things like how molecules work or the Civil War. However, conventionally educated people are also generally unclear on these things: for instance, 19% of college students know what the Manhattan Project is, 16% know that the Raven in the poem of the same name says “Nevermore,” and 14% know that Mendel is the man who first studied genetic inheritance in plants. (In the interests of not presenting an unfairly biased list, I will add that college students are generally extremely accurate about the definitions of the words “zebra,” “hibernation,” “hockey puck,” “fossil,” and “ruby.” So science education at least is not a total failure.) It is a commonplace observation that most people go through twelve years of mathematical and scientific education, and graduate with no ability to do anything beyond arithmetic and only the vaguest understanding of Newton’s laws or the theory of evolution. If my children are ignorant about Edgar Allen Poe, at least they will have a firm understanding of ancient Greece, which is more than can be said for the general public.

In addition, one does not have unlimited time to educate children: you can either give them a broad overview of many topics or a deep understanding of a few topics. You can have a world history class which gives two days to Sumeria and one day to the Vietnam War, or you can have a Sumerian history class that doesn’t talk about anything else. It’s not obvious to me that the former is obviously better than the latter, and the difficulty of coercing children whose brains work the way mine does leads one inevitably to the latter.

Perhaps the most important part of unschooling is not what it does but what it doesn’t do: that is, unschooling does not crush the love of learning out of children. Peter Gray writes in Free to Learn that adults who attended Sudbury schools as children are often behind in academic knowledge, but they catch up quite quickly once they go to college. And they are routinely baffled by other college students: these college students are paying tens of thousands of dollars to learn from experts in the field, and yet their primary interest is doing the minimum they can to get an A. It was simply incomprehensible to them.

It seems to me that mandatory schooling is likely to reduce the love of learning, because you will regularly have to learn about things you don’t care about and aren’t interested in, and if you are interested in a subject you cannot explore it in as much depth as you would prefer. Extrinsic rewards tend to decrease intrinsic motivation: the more you’re working to get an A in the class, the less you’re working because you actually care about the subject. I’ve personally experienced this– there’s nothing that kills my motivation to write five-thousand word essays about feminism than getting a grade on it.

However, I do think there is a certain amount of wisdom in the three R’s, reading, writing, and arithmetic, one that overcomes my objections to coercion. Reading, writing, and arithmetic (plus statistics) are unique among subjects in that they make it easier to learn everything else. While there are other subjects that make everything else easier to learn, they generally only require a few days’ worth of explicit education (the scientific method) or are related to so many different interests that there’s not much reason to explicitly teach it as its own thing (the ability to smell bullshit).

You could also add “a foreign language” to the list of things that make it easier to learn other things, and I certainly would if my children were not native English speakers. However, teaching a foreign language such that the child actually learns it is an enormous pain unless you happen to already have lots of friends who are fluent speakers, and there doesn’t seem to me to be much point in replicating the standard American four-years-of-high-school-Spanish-and-can’t-ask-where-the-bathroom-is. However, if you happen to know lots of people who fluently speak Spanish, Chinese, or a similarly useful language, it seems well-advised to ask them to babysit regularly and refuse to speak any language other than the one you want your child to learn.

Of reading, writing, arithmetic, and a useful foreign language, reading is the most important: while writing is primarily useful if you want to communicate something, and math is useful for the natural and social sciences but generally unnecessary for the humanities, literally everything you want to know– from cooking to woodworking to economics to the history of the Indus valley civilization– is easier to learn if you are capable of reading a reference text.

The good news is that all pretty much all unschooled children learn to read eventually, on average at the age of 8, which is only one year behind most children. (Note, however, the caveats above about self-reported data: this evidence is likely to massively overstate how early unschoolers are reading.) The bad news is that nearly a fifth of unschooling children learn to read after the age of 10, which means that they have at best three to four years of fundamentally impaired ability to learn, and perhaps almost a decade. Imagine what they could have accomplished in those three to four years with appropriate reading education!

Some unschooling advocates point to hunter-gatherer societies, in which children often learn solely through play. However, the games hunter-gatherer children play have undergone literal millennia of cultural evolution to make sure they teach the skills hunter-gatherer children need to learn. This is not true for modern children; our games are not generally optimized for teaching reading, writing, and arithmetic. For that reason, I suggest using Montessori methods in the early grades, which are play-based.


Both my family and my husband’s family have a disproportionate number of weird, awkward nerds. Having no social skills is deeply unpleasant: it feels like you are surrounded by incomprehensible monsters who hate you for no reason, perhaps because they are evil, perhaps because you are an inherently terrible person. For this reason, it makes sense to prioritize teaching social skills.

However, explicitly teaching social skills often goes poorly. For one thing, social skills classes are often based more on what parents and teachers want to be true than on what is actually true: for instance, students will be told to tell bullies firmly that they don’t like that, but will not be told to punch the bully or, failing that, befriend someone tough who can punch the bully for you, even though the latter are far more effective techniques. Sometimes students are even told that sitting quietly in class and doing your homework will make you friends. For this reason, I suggest avoiding social skills classes.

I am currently moderately socially competent. When I think about how I became moderately socially competent, two things stand out. One is that I began communicating online, which stripped away the incomprehensible tone and body language and allowed me to quietly observe interactions for months before I participated myself. For this reason, I plan to encourage my children to engage in online interaction.

The second is that online I could talk to people whose neurotypes (for lack of a better word) were similar to mine. (This is not about diagnoses: there are many autistic people I can’t relate to with anything other than polite incomprehension; weird awkward nerds can have a wide variety of different diagnoses and often do not qualify for any diagnosis at all, except perhaps recurring depression.) I don’t think people often think about how important the typical mind fallacy is in developing cognitive empathy, particularly when you are first learning. The first step to cognitive empathy is going “I would be sad if I lost my doll, she lost her doll, so she is probably sad.” Only once you have a firm grasp of that can you move onto “I would be sad if I lost something important to me, I don’t care about tea sets, she cares about her tea set, she lost her tea set, so she is probably sad.”

Interacting with people who are very different from you is interaction on hard mode. We normally place children with social-skills impairments in environments where reasoning based on their own minds is utterly useless. If you like listening to other people’s infodumps, you might infodump about your own interests and then be puzzled about why no one likes you. If you misunderstand subtext, you might politely decide to be blunt about whether you like another child’s haircut. If you don’t care about hygiene, you might be confused about why the other children make fun of you for wearing the same pants three days in a row. Weird awkward nerds are probably different from other children in other ways: for instance, they tend to have different interests, which makes it hard to bond over shared passions or hobbies. The situation is even worse for autistic children, who not only have all these difficulties but also have a characteristic affect (stimming, lack of eye contact) which is unpleasant for most neurotypicals. So you put children who are already bad at social interaction in a situation where they have to do very complicated social reasoning and they don’t share many interests with the children they’re talking to and the other children are already biased towards disliking them. This is a recipe for disaster.

It seems to me a better way would be to put weird awkward nerdy children in an environment of solely weird awkward nerdy children. As young children, they can learn empathy, confidence, and how to make friends around people like them. Once they’re a bit older, they can interact with ordinary children and children who are differently socially impaired, and learn how to expand their empathy to people who are more different from them: since they already have friends, their failures won’t make them feel like they are inherently unlikeable and alien, and perhaps they can compare notes with their friends and together learn to understand more normal people. Autistic children can learn to fake eye contact and to stim subtly in adolescence, after several years of being allowed to stim freely, and when they aren’t trying to learn all the other rules at the same time. (I do believe that being able to pass is, sadly, a useful skill.)

I guess this is partially an argument against inclusion, even though inclusion has been a big disability rights push for decades. I want to defend myself against this a little bit. A disabled-children-only classroom seems very silly: most disabled children don’t have any sort of social impairment, so this argument doesn’t apply to them. Not all socially impaired children should be put in the same classroom: there are lots of different ways children can be socially impaired, and a socially impaired child may have even more difficulties understanding a differently socially impaired child than they do understanding a child with ordinary social skills. And certainly there is no reason to separate disabled and nondisabled weird awkward nerds.

Like I said, I am not really capable of suggesting strategies for educational reform: I shudder at the idea of turning “weird awkward nerd” into a set of criteria that decides which classroom you get to go into. And I have zero evidence (other than my own personal experience) that suggests this is actually a better way to raise socially impaired children. That said, personally I plan on setting up my homeschooling so that my children mostly interact with similarly weird and awkward children in the elementary-school years, and to me this is the major advantage of being able to homeschool.