All topics › Behavior discipline emotional regulation
What does the evidence say about smacking/physical punishment and child outcomes?
What does the evidence say about smacking and physical punishment — does it work, does it harm, and what does the law actually say in the UK?
Across decades of observational research, smacking is consistently associated with worse — not better — child outcomes: more aggression and antisocial behaviour, more mental health problems, and a poorer parent–child relationship [1][2]. No meta-analysis has found a credible long-term benefit. But every one of those studies is observational — no experiment can ethically randomise children to be smacked — so how much of the link is causal, and how much reflects the fact that harder-to-manage children get smacked more, is genuinely debated [3][4]. In the UK, smacking is now unlawful in Scotland and Wales, while in England and Northern Ireland parents still have a legal defence of "reasonable punishment" [8][9].
The largest meta-analysis, by Gershoff and Grogan-Kaylor, pooled 111 effect sizes covering 160,927 children and found smacking was significantly associated with 13 of the 17 outcomes examined — every significant association pointing the same way: more aggression, more antisocial behaviour, more mental health problems, lower cognitive performance, a poorer parent–child relationship, lower long-term compliance, and higher risk of physical injury or abuse [1]. The average effect size was small-to-moderate (d = 0.33), and the link held across the US and 13 other countries, across cohorts from the 1950s to the 2000s, and however spanking was measured [1]. An earlier landmark review of 88 studies found the same pattern [2].
That consistency matters: when a finding shows up in dozens of studies, in different places and decades, it is harder to dismiss as a quirk of one sample [4].
Here is the honest problem. Parents don't smack at random. Children who are temperamentally harder to manage — more impulsive, more defiant — both get smacked more and go on to have more behaviour problems anyway. Parental stress, poverty, and parents' own mental health push in the same direction. These confounders can explain part, perhaps a large part, of the association [3][4].
A meta-analysis that restricted itself to longitudinal studies and statistically controlled for children's earlier behaviour found much smaller effects: correlations of around .07–.10 for externalising and internalising problems — real, but small [3]. The author's conclusion was that psychologists should be more modest in their public claims. Critics respond that controlling for earlier behaviour can over-control: if smacking gradually worsens behaviour year by year, then stripping out the earlier behaviour strips out part of the effect itself [4]. Both sides agree on the direction; they disagree on the size.
The sceptics have one more card to play, and it's worth understanding. A study that re-ran the best longitudinal evidence for spanking alongside three alternatives — grounding, privilege removal, sending the child to their room — found the apparent adverse effect of every tactic, spanking included, shrank to non-significance once pre-existing behaviour was measured comprehensively [15]. Their argument: when every corrective action looks harmful in the data, the "harm" is probably an artefact of which children get disciplined most, not something specific to smacking. The counter from the other side is that quasi-experimental designs (comparing siblings, or tracking within-child change) still find worsening behaviour over time [11] — which shouldn't happen if the whole association were selection bias.
What nobody has found is a benefit. No meta-analysis shows smacking improving long-term behaviour, and even the one outcome parents care about most — getting the child to comply — looks unimpressive on close inspection: spanked children appeared more compliant in the moment, but once researchers accounted for the fact that those children were less compliant to begin with, the effect disappeared [1].
The only experiments in this literature are four small randomised trials — and they did not randomise ordinary parents to smack at home. They tested spanking used solely to enforce compliance with time-out in clinically defiant 2- to 6-year-olds already in treatment programmes [4][15]. That narrow exception aside, the field's claims about everyday smacking rest on observational data.
Since the two big meta-analyses, the field has moved — mostly in the same direction.
A 2021 narrative review in The Lancet summarised 69 prospective longitudinal studies — designs where behaviour is measured after the punishment, so the timeline is clear [11]. Seven themes emerged: physical punishment consistently predicts increases in behaviour problems over time; it is not associated with positive outcomes over time; it increases the risk of involvement with child protective services; children elicit punishment mainly through externalising behaviour; quasi-experimental studies (which compare, say, siblings, or track changes) also find worsening behaviour; the pattern holds across child and parent characteristics; and there is some dose–response evidence — more frequent or severe punishment predicts worse outcomes. The authors' conclusion: "physical punishment is harmful to children and policy remedies are warranted."
The newest meta-analysis (2024) pooled 35 studies and 144 effect sizes, covering 159,213 participants, and found a significant association between corporal punishment and a spectrum of violent behaviours (aggression, antisocial behaviour, bullying, violent crime) of r = 0.24 [12]. Punishment severity moderated the link: harsher punishment predicted more violent behaviour — a dose–response pattern that is harder to wave away.
Two caveats. First, a narrative review is not a meta-analysis — it doesn't pool numbers, so it can't tell you how big the effects are. Second, "prospective" doesn't solve the confounder problem on its own; the researchers who worry about residual confounding say the same about these studies [15]. The Lancet authors' conclusion is theirs, stated as the evidence suggesting it — you should know it's the authors' reading, not a law of nature.
One review (Larzelere & Kuhn) argues the research has been unfair to a very specific practice: brief, calm, controlled smacking used only as a backup for outright defiance in 2- to 6-year-olds, which in their reading outperformed several alternatives [5]. This is a genuine minority position. Most researchers reply that what parents actually do bears little resemblance to this laboratory-precise version, and that the meta-analytic effect sizes don't shrink when spanking is separated from harsher physical punishment [1]. Included here so you know the debate exists — not because the evidence is evenly balanced.
Effect sizes are abstract, so here's the honest translation. A d of 0.33 means smacked children scored, on average, about a third of a standard deviation higher on measures of aggression and related problems — a small-to-moderate association, roughly comparable in size to many accepted findings in child psychology [1]. It does not mean most smacked children develop serious problems; most don't. It means the distribution shifts: at population level, more smacking predicts more problems, even though any individual child's outcome depends on far more than this one factor. Ferguson's controlled estimates (around .07–.10) would imply a much smaller shift [3]. The range between those two readings is the genuine uncertainty in this literature — not whether the direction is bad, but how bad, and for whom [4].
The UN Committee on the Rights of the Child — the body that interprets the Convention on the Rights of the Child — says eliminating all corporal punishment, "however light", is an "immediate and unqualified obligation" of states [7]. Sweden became the first country to ban it, in 1979, alongside a national public-education campaign. In the decades since, Swedish surveys show steep declines in both approval of and reported use of physical punishment [6] — though how much of that the ban itself caused, versus broader cultural change, is hard to separate (the ban came bundled with the campaign). Internationally, 65 states now fully prohibit physical punishment of children (per the RCPCH's April 2024 count), with Scotland and Wales among them [10].
The main US paediatric body has also taken a position: the American Academy of Pediatrics' 2018 policy statement advises adults to avoid all physical punishment, saying aversive strategies — including spanking and shaming — are minimally effective in the short term and not effective in the long term [13].
There isn't much UK-specific data, but there is some. Research using the Millennium Cohort Study — which followed around 19,000 UK children born in 2000–02 — found that corporal punishment of any frequency was most common for 3-year-olds (about 70% of boys and 64% of girls) and decreased as children got older, with boys experiencing more than girls and little variation across household income [14]. So if it happens in your house, it is — statistically — utterly ordinary. That tells you nothing about whether it's a good idea; it tells you you're not an outlier.
The UK is split. Smacking is unlawful in Scotland (since November 2020, under the Children (Equal Protection from Assault) (Scotland) Act 2019) and in Wales (since 21 March 2022, under the Children (Abolition of Defence of Reasonable Punishment) (Wales) Act 2020) [8]. In England and Northern Ireland, parents can still rely on a legal defence of "reasonable punishment" if charged with common assault or battery of a child — but that defence is only available against those lesser charges. It cannot be used where the force causes actual bodily harm [9] — so anything that leaves an injury moves into a different legal category. The Royal College of Paediatrics and Child Health published a 2024 report calling for the defence to be removed in England and Northern Ireland; its child protection lead, Professor Andrew Rowland, called the current laws "unjust and dangerously vague" because they leave a grey area over which forms of physical punishment are lawful [10].
This topic is usually framed as being about the child, but the research points back at the adults too — with unusual honesty about the limits of the data.
Smacking happens most when parents are at their lowest. The studies reviewed here measure child outcomes, not parents — but Gershoff's (2002) process–context model of corporal punishment treats parental strain and family circumstances as central to when physical punishment happens [2]. That doesn't make it anyone's "fault" — it reframes smacking as a sign that the parent's resources are depleted, not that the child is uniquely awful. No trial has measured what stopping (or starting) smacking does to a parent's mental health, sleep, or relationship — that evidence simply doesn't exist yet. But the pattern is consistent enough to say: if you find yourself smacking, it's worth asking what's going on for you, not just for your child.
There is a real escalation risk. Physical punishment is one of the few parenting behaviours where the research documents a gradient from punishment to injury: most smacks cause no physical injury, but smacked children are at higher risk of injury and of crossing the line into abuse [1][2]. A smack delivered in anger is harder to keep measured — which is why this is the strongest practical argument in the whole literature, and it doesn't depend on resolving the causality debate.
Guilt is not evidence, in either direction. Many parents who were smacked themselves feel uneasy about it and also defensive about their own parents. The evidence can't settle that for you — but it can tell you that effective alternatives exist (see the time-outs and praise topics), so "I don't know what else to do" doesn't have to be the end of the story.
There are no trial event rates here (no per-1,000 figures exist), so the table reports effect sizes with the honest caveat that they come from observational studies.
| Outcome | Finding | Evidence rating |
|---|---|---|
| Aggression / antisocial behaviour | Mean d = 0.33 overall — small-to-moderate, all detrimental [1] | B |
| Same outcomes, longitudinal studies with baseline behaviour controlled | Partial r ≈ .07–.10 — small [3] | B |
| Mental health problems (internalising) | Consistent detrimental association, small-to-moderate [1][2][3] | B |
| Cognitive performance | Slightly lower scores associated with physical punishment [1][2] | C |
| Parent–child relationship quality | Poorer relationship quality associated with physical punishment [1][2] | C |
| Long-term compliance (the intended benefit) | Associated with lower compliance, not higher [1] | B |
| Immediate compliance | Apparent benefit disappears once baseline differences are accounted for [1] | C |
| Physical injury / escalation to abuse | Higher risk; one of the most practically important findings [1][2] | B |
| Violent behaviour spectrum (aggression, antisocial behaviour, bullying, violent crime) | r = 0.24 across 35 studies; harsher punishment → stronger link [12] | B |
| Involvement with child protective services | Increased risk of involvement [11] | C |
Note: converting these effect sizes into "X in 1,000 children" would require baseline rates the studies don't provide — anyone who gives you such a number is guessing.