What the data actually shows
The most influential synthesis comes from Elizabeth Gershoff and Andrew Grogan-Kaylor. Their 2016 meta-analysis pooled decades of studies covering tens of thousands of children and found that spanking — defined narrowly as open-handed hitting meant to correct, not abuse — was associated with thirteen of thirteen measured outcomes pointing in an unfavourable direction: more aggression, more antisocial behavior, more mental-health difficulties, and a weaker parent-child relationship, with no association at all with better long-term compliance.
Crucially, the effect sizes for spanking looked similar in direction to those for physical abuse, just smaller — undercutting the common intuition that ordinary spanking is a categorically different, harmless thing. An earlier 2002 review by Gershoff had reached compatible conclusions, and later longitudinal work that follows the same children over time tends to find that more spanking predicts more behavior problems later, even after accounting for earlier behavior.
The honest limitation is causation. Because you cannot ethically randomise children to be hit, almost all of this is observational, and some of the link surely runs the other way — difficult behavior invites more spanking. Researchers use statistical controls and longitudinal designs to probe this, and the harm-direction association largely survives, but no single study settles the causal question. What the body of evidence does establish reasonably well is the absence of benefit: there is no credible signal that spanking improves long-term behavior.
Why this feels different from how it actually is
Spanking can feel effective because it often works in the moment — a child startled or hurt usually stops the behavior right then. That immediate compliance is powerful feedback for a parent, and it is exactly why the practice persists. But short-term stopping and long-term behavior are different things, and the research is about the second, where the picture reverses.
It also feels different because of personal experience. Many people who were spanked grew up fine and reasonably conclude it did no harm. That is a real observation, but it runs into survivorship and averaging: 'I turned out okay' tells you about one outcome under one set of circumstances, not about the average effect across millions of children, which is what the studies measure.
And there is the weight of tradition and the fear of permissiveness. Spanking can feel like the responsible, firm choice and its absence like letting a child 'get away with it.' But the research consistently separates firmness from physical punishment: clear limits and consistency are linked to better outcomes; the hitting part specifically is what carries no benefit and a risk of harm.
The research consistently separates firmness from physical punishment: clear limits and consistency are linked to better outcomes; the hitting part specifically is what carries no benefit and a risk of harm.
What the research says to do about it
The better-supported alternatives are not permissiveness but non-physical structure. Approaches built on clear, consistent, age-appropriate limits — explaining expectations, following through calmly, using time-outs or loss of privileges, and noticing and reinforcing good behavior — have stronger evidence for shaping behavior over time than physical punishment does, without the associated risks.
Warmth combined with structure is the broader pattern the developmental literature keeps returning to. Children tend to do somewhat better when discipline happens inside a reliably affectionate, responsive relationship, and when ruptures are repaired afterward rather than left. The relationship itself does much of the work that punishment is imagined to do.
Reducing the parent's own stress and reactivity is one of the better indirect levers, because a large share of spanking happens in moments of anger and overwhelm rather than as a considered strategy. Parenting-support programs that teach calmer, consistent responses are associated with less reliance on physical punishment and fewer behavior problems.
What the research says does not help
The belief that there is a 'right amount' of spanking that helps if you stay calm and don't overdo it is not supported. The reviews find the unfavourable associations across ordinary, non-abusive spanking, not only at extreme levels, so 'just a controlled swat' is not a category the evidence treats as safe or effective.
Swinging to the opposite extreme — no limits, no consequences, hoping reasoning alone will do it — is not what the research recommends either, and is not what 'don't spank' means. The evidence favours structure and consistency delivered without hitting, not the removal of structure.
Relying on 'I was spanked and I'm fine' as a guide does not help, because individual anecdotes — in either direction — cannot tell you about average effects, and the average is what a decision rule should rest on. The data does not say everyone who was spanked was harmed; it says there is no average benefit and a measurable average risk.
There is no credible signal that spanking improves long-term behavior.
What this looks like in real life
Why it feels effective in the moment
A child does something dangerous, gets a swat, and stops right away. That immediate compliance is powerful feedback, and it is exactly why the practice persists. But short-term stopping and long-term behavior are different things — and the research is about the second, where the picture reverses. The thing that works in the moment is not the thing that shapes behavior over time.
One life versus an average
Many people who were spanked grew up fine and reasonably conclude it did no harm. That is a real observation, but a single life cannot reveal an average effect. The studies measure the average across large populations, where spanking tracks worse outcomes on balance and no better behavior. Both things can be true at once: plenty of individuals are fine, and there is still no average benefit.
Real numbers in context
The headline figure from the Gershoff and Grogan-Kaylor 2016 meta-analysis is striking precisely because it is so one-directional: of the thirteen child outcomes they examined, spanking was significantly associated with worse results on essentially all of them and with better results on none. The individual effect sizes were modest, not dramatic — this is a consistent, small-to-moderate signal, not a claim that spanking devastates children.
It is also worth holding the correlational caveat in view: these are associations drawn largely from observational data, so the precise causal magnitude is genuinely uncertain and researchers disagree about how much of the link is causal. What is far less contested, across reviews and across major pediatric guidance, is the other half of the finding — that there is no reliable evidence spanking produces better long-term behavior than non-physical discipline.