Raising smart kids, by the evidence

Heritability describes how much of the differences between people in a group line up with genetic differences. For measured intelligence it is substantial and it rises with age, from roughly 20% in infancy to 40% in adolescence and 60% in adulthood, partly because older children increasingly choose experiences that suit them. No trait is fully heritable, and a population figure is not a ceiling on your child.

Two findings set the scale for the rest of this guide. Schooling itself raises measured ability: studies built on changes in school-leaving laws find about 2 intelligence quotient (IQ) points per extra year. And early gains fade: preschool programs often produce real test gains that shrink in the first years after the program ends. In education research, one fifth of a standard deviation already counts as a large effect, so a product promising far more is promising more than the best-tested programs deliver.

The trap is reading a parenting correlation as a recipe. Parents pass on genes as well as a home, so a habit can travel with a child's outcome because of the genes they share. In one family study, most home-literacy habits stopped predicting children's reading once the parents' own reading skill was counted. Suppose Dana reads that children whose parents quiz them at dinner get better grades. Before adding quizzes, she should ask whether the study accounted for the parents' own ability.

Rule: look for a randomized trial before believing that doing one thing will change another, and expect modest effects.

Many products for children rest on findings that were small, brief or never there. The Mozart effect began as a slight, short-lived bump on a spatial task in adults after they heard one sonata. Later work traced it to small studies and measurement bias, and the claims of lasting gains for children came afterward, with products to match. In a controlled study, babies aged 12 to 18 months learned none of the words in an educational video, while babies taught the same words in person by a parent did learn them. Teaching to a child's learning style has little support: a 2024 meta-analysis found only 26% of outcomes showed the pattern the theory needs, yet in one US survey 93% of the public believed in it.

Brain-training games follow the same pattern. Training improves the trained task and close variants, which researchers call near transfer, but reviews find far transfer, to general thinking or school achievement, close to zero. The Federal Trade Commission (FTC) has acted against products that claimed otherwise: Lumosity's makers paid $2 million in 2016, the makers of a children's focus game settled in 2015, and the marketers of an infant reading program settled in 2014.

The trap is trusting the label. Suppose Leo buys a memory-game app that promises better grades in every subject. His daughter will get better at the app.

Rule: ask whether a randomized trial with a fair comparison group backs the claim, and expect practice to improve only what is practiced.

Language grows from back-and-forth exchanges. When your baby smiles or makes a sound and you answer in kind, you are doing what researchers call serve and return, and taking turns making sounds is a milestone most babies reach by about 6 months. In lab studies, babies who got prompt responses to their babbling changed how they babbled, and programs that coached parents to answer this way increased conversational turns and toddlers' vocabulary. When a baby points, answer in words; when a toddler says "Dog!", repeat the word inside a fuller sentence. Questions that start with who, what, where or why draw longer answers than yes-or-no questions do.

How much talk matters is debated. A 1995 study reported a 30-million-word gap by age 4 between better-off and poorer families, but a recount that included other caregivers and overheard speech found more talk than first reported, and researchers still argue about what to count. Two languages from birth do not confuse children or delay them: counted across both languages, bilingual children reach milestones like first words on a similar timetable.

Screens compete with talk. In an Australian study that recorded 220 homes, each extra minute of screen time at 36 months went with fewer adult words and fewer conversational turns.

The trap is talking at a child instead of with them. Suppose Priya narrates her whole day to her baby but rarely pauses. Leaving a gap after she speaks gives the baby a turn.

Rule: answer what your child just said or pointed at, and leave room for a reply.

Book sharing helps, modestly. A meta-analysis of 19 randomized trials of programs that taught parents to share picture books with children aged 1 to 6 found small vocabulary gains, larger when parents had more program contact, and some large parent programs have found no effect at all. Being read to explains only about 8 percent of the differences in children's language and reading skills, so a missed season of bedtime stories is not a verdict.

How you read matters. In dialogic reading the child does most of the talking: the adult prompts, evaluates the answer, expands it with a little more information and asks the child to say the fuller version, a routine called PEER. Prompts that ask the child to finish a line, recall the story, answer a wh-question or link the story to their own life keep them talking. Pointing at the words themselves, called print referencing, draws attention to written language.

When a child starts reading, the evidence favors teaching sounds. The National Reading Panel found that systematic phonics, teaching letter-sound links in a planned order, beat instruction without it, with the largest effects in kindergarten and first grade. It also found that guided oral reading with feedback improved fluency and comprehension.

The trap is teaching a child to guess. Suppose a first grader stuck on "horse" is told to look at the picture. Helping him sound out the letters first builds the skill he will need on the next page.

Rule: get your child talking about the book, and when they read, point them to the letters before the picture.

Guided play sits between free play and direct teaching: the adult sets up a playful activity with a goal in mind, and the child keeps the lead. A 2022 meta-analysis found it beat direct instruction on early math and shape knowledge and beat free play on spatial vocabulary, but almost every study was at high risk of bias. Showing a child how a new toy works can narrow their exploring; in a well-known experiment, children taught one feature found fewer of the others. Pretend play is common and valuable as play, but evidence that it causes developmental gains is weak.

Curiosity deserves protecting. Young children can ask more than 100 questions an hour, and preschoolers are less satisfied when a why gets a non-answer than when it gets a real explanation. When you do not know, saying so and finding out together shows how people learn.

Math starts in conversation. In home observations, some parents used more than 200 number words with their toddlers and others fewer than 10, and early number talk predicts later math. Counting is not the same as knowing how many, so after your child counts, ask how many there are. A simple board numbered 1 to 10 in a straight line improved young children's number skills more than a circular one, and spatial skills such as picturing a turned shape improve with practice, for girls and boys alike.

The trap is turning play into a lesson. Suppose Ben's daughter starts building a circus from blocks and he redirects it into a zoo. Asking which animal goes where would keep her in charge and still bring in spatial words.

Rule: follow your child's lead in play, and weave in a question, a number or a spatial word.

Executive function is three skills: working memory holds information in mind, inhibitory control stops an automatic response, and cognitive flexibility shifts between rules. They predict achievement beyond intelligence and family income, which makes them worth attention. The marshmallow test made waiting famous, but a larger 2018 replication found a weaker link with later achievement that largely disappeared once family background was accounted for. One short wait at age 4 is no forecast.

What parents do is linked to these skills. A meta-analysis of 41 studies of children up to age 8 found warm, responsive parenting correlated about .25 with executive function, and controlling, intrusive parenting about −.22. Correlations do not prove cause, and calmer homes partly reflect genes that parents and children share. Programs give a mixed picture: one kindergarten trial of the Tools of the Mind curriculum found gains, most preschool trials found none, and a classroom trial of short executive-function games found no gains on direct tests.

The trap is taking over. Scaffolding means giving just enough help for the child to take the next step, a question or a hint, and then handing the task back. Suppose Mia's son is stuck on a puzzle. Placing the hard pieces finishes the puzzle and skips the practice; asking which piece has a straight edge keeps him working.

Rule: give the smallest hint that lets your child take the next step alone, and practice the real routine you care about.

Person praise credits a fixed trait, as in "you're so smart." Process praise credits what the child did, such as effort, a strategy or checking the work. Long-term home studies link parents' praise of toddlers' effort with more persistence and growth-mindset beliefs years later, though these studies are correlational. Inflated praise, words like incredible for an ordinary result, carries its own risk: about a quarter of the praise adults give is inflated, and in one experiment children judged a classmate who got lavish praise for an average score as less smart.

A growth mindset is the belief that ability can be developed. In a national US experiment with 12,490 ninth graders, an online lesson under an hour long raised lower achievers' grades by 0.10 grade points and did nothing for higher achievers. Meta-analyses disagree about whether such programs help beyond targeted groups, so expect small effects at most.

Rewards need care too. Across 128 experiments, promised tangible rewards tended to undermine intrinsic motivation, the drive to do something for its own sake, while verbal praise tended to increase it. Researchers still debate when the effect appears.

The trap is paying for what a child already loves. Suppose Ava reads every night for fun and her parent starts paying a dollar a book. Reading risks becoming a job.

Rule: praise the action your child can repeat, at the size of the achievement, and save promised prizes for tasks they have no interest in.

The Centers for Disease Control and Prevention (CDC) recommends 10 to 13 hours of sleep a day for ages 3 to 5, naps included, 9 to 12 for ages 6 to 12 and 8 to 10 for teens. Too little sleep is linked to attention and behavior problems that can hurt schoolwork, and in CDC's 2025 data 75% of students were not getting enough. Federal guidelines ask for at least 60 minutes of moderate-to-vigorous activity a day for ages 6 to 17. Activity is good for health, but the evidence that it raises grades is inconsistent.

Lead is the clearest environmental harm to learning. No safe blood lead level has been identified, and in a pooled study of 1,333 children the steepest losses came at the lowest exposures. CDC's reference value, 3.5 micrograms per deciliter, flags the highest 2.5% of young children; it is not a safe level. Most children with lead in their blood show no symptoms, so only a blood test can tell. Ask the child's doctor, especially if your home was built before 1978.

Some popular claims are smaller than they sound. Iron deficiency anemia in young children can lead to learning difficulties, but extra iron for children who are not deficient has not been shown to help thinking. Much of the link between breastfeeding and test scores reflects family background.

The trap is treating slipping grades as a motivation problem before checking sleep. Suppose Noah's grades drop in fifth grade; the first question is when he actually falls asleep.

Rule: protect sleep, prevent lead exposure, and let the child's doctor decide on any supplement.

Since 2022, each CDC milestone sits at an age by which at least 75% of children are expected to show it, so even one missed milestone is a reason to talk with the doctor. The checklists are communication tools, not screening tests; pediatric guidance calls for standardized developmental screening at 9, 18 and 30 months.

Early intervention is the publicly funded system of services, such as speech therapy, for children under 3 with delays. You do not need a referral: call your state's program and ask for an evaluation, and federal rules then give the program 45 days to evaluate and meet with you. From age 3, ask your local public school.

Preschool evidence needs care. Intensive programs such as Perry Preschool and the Abecedarian Project found lasting gains into adulthood, but they cost far more than public preschool and were never fully replicated. Head Start's measured benefits mostly faded by third grade, and children who attended Tennessee's state pre-K did worse by sixth grade, for reasons researchers dispute. In child care, a large federal study of more than 1,000 children found sensitive, encouraging caregiving among the strongest predictors of development, and family life mattered more than child care features.

The trap is judging a program by its worksheets. Suppose Rosa tours two centers. The one where caregivers keep talking with and answering the children is the better sign.

Rule: act on a missed milestone now, and judge child care by how the adults interact with the children.

A 2023 review of 20 studies found that supportive homework help, the kind that leaves the child in charge, went with slightly higher math scores, while intrusive help went with lower ones. In one study of fourth graders, controlling help and weaker grades fed each other over a few months. Homework time itself is barely linked to achievement before high school.

Activities chosen for spillover rarely deliver it. In well-controlled studies, music lessons showed no clear effect on general thinking skills or grades, though the debate continues, and children taught chess for 25 hours did no better at math than children who played checkers. Tutoring is different: literacy tutoring averaged an effect of 0.35, largest in the earliest grades and when trained staff delivered it.

Age within a class matters too. Across 32 studies, the youngest children in a school year were more often diagnosed with attention deficit hyperactivity disorder (ADHD), and in France the youngest were 64% more likely to start speech therapy. Advanced students who are accelerated tend to gain academically without social harm on average, while gifted programs that rely on referrals miss many capable children.

The trap is fixing the work for the child. Suppose Sam checks and corrects every answer before it goes back to school. The page improves, but the habit he is building is the one linked to lower scores.

Rule: support how your child works, and leave the work itself in their hands.

In adolescence, academic socialization, linking schoolwork to a teen's goals and future, has the strongest link with engagement among forms of parental involvement, while intrusive homework help is most strongly linked to lower math achievement in high school. Autonomy support, explaining rules and inviting initiative, is the opposite of psychological control, which presses through guilt. Feeling competent is the strongest predictor of self-driven motivation.

Sleep changes for biological reasons. Puberty shifts the body clock later and teens need 8 to 10 hours, yet in 2011–12 the average US public middle or high school started at 8:03 a.m. In a small randomized trial, starting an hour later gave young teens about 26 more minutes of sleep. Parent-set bedtimes and keeping devices out of the bedroom are both linked to more sleep.

Social media is a contested risk. The 2023 Surgeon General's advisory said there is not yet enough evidence that it is safe for youth. Large studies find small average links with well-being and possible sensitive ages, while critics argue that the averages hide larger links for heavy users, especially girls.

The trap is treating a late bedtime as defiance. Suppose a 15-year-old cannot fall asleep before 11:30. Stricter lights-out orders will not move the body clock; a phone-free bedroom and a steady wake time help more.

Rule: talk about goals, step back from supervising homework, and protect sleep before arguing about effort.

The National Institute of Child Health and Human Development says learning disabilities are not related to intelligence: the brain processes information differently, and earlier help raises the chance of success in school. Dyslexia usually shows up as trouble linking letters to sounds, spelling and recognizing words, and a bright, talkative child can still have it. A speech or language delay deserves a doctor's visit, often with a hearing test, and a passed newborn hearing screen does not rule out later loss.

ADHD means attention or hyperactivity problems that are frequent, appear across settings and disrupt daily life. No single test diagnoses it, and poor sleep or anxiety can look similar, so diagnosis starts with the child's healthcare provider. For children under 6, parent training in behavior management is the recommended first treatment, and medication decisions belong with the provider.

At school, either you or the school can request an evaluation under the Individuals with Disabilities Education Act, and once you consent it must be completed within 60 days unless your state sets its own timeframe. An Individualized Education Program (IEP) provides special education; a Section 504 Plan adjusts the environment so the child can learn alongside others. A child can also be both gifted and learning disabled, with each hiding the other.

The trap is waiting to see. Suppose a teacher suggests giving a struggling second-grade reader another year before testing. Waiting costs time that early help would use.

Rule: put an evaluation request in writing, note the date you consent, and follow up when the deadline nears.