Discipline when the two of you haven't agreed

Bottom line

  • Settled: be warm, hold firm limits, praise what you want, use consequences that are connected and calm, and don't smack.
  • Not settled: time-out versus time-in. The research on time-out is better than its reputation, and nobody has tested time-in by itself. Pick one together and both use it.
  • At 18 months, change the room and redirect. He can't hold a rule yet.
  • At three, most kids still can't reliably stop themselves. Keep the rules few and the response the same from both of you.
Evidence: Strong Evidence: Contested Evidence: Weak

Last reviewed

This page is a parent's summary of published guidance and research. It is not a screening tool and not medical advice. If you are worried about your child, talk to your GP or child health nurse. In Queensland you can call 13 HEALTH (13 43 25 84) and ask for a child health nurse. Disclaimer.

My partner and I have not settled how we discipline our boys, who are three and 18 months.

So I read what the evidence says, with one aim: find out which parts are settled, so we only have to argue about the parts that aren't.

Start with what they can actually do

A lot of arguments about discipline are really arguments about what a child that age is capable of.

In an American survey of 2,200 parents, 36% thought a child under two could resist doing something forbidden, and 56% thought that came before three. Most children can't do it reliably until about three and a half to four (Zero to Three, read from a summary; grade: moderate).1

So our 18-month-old has almost no brakes. He can know the rule and still not stop himself. Our three-year-old has some, on a good day.

Other things that are normal, however they look at the time:

  • Tantrums. About 87% of children have them at 18 to 24 months and about 91% at 30 to 36 months. Three quarters last five minutes or less (Potegal and Davidson 2003, via a review; figures not verified against the original paper).2
  • Hitting and biting. Physical aggression shows up around the first birthday and peaks somewhere between two and three and a half, then falls from about four (Alink et al. 2006; figures not verified against the original paper).3
  • Not sharing. Children aged three to six will tell you sharing equally is right and then keep more for themselves. What they do doesn't match what they say until about seven or eight (Smith, Blake and Harris 2013; grade: moderate).4

None of that means you do nothing. It means a three-year-old who hits isn't broken, and an 18-month-old who goes back to the same forbidden thing isn't defying you. What the lists do and don't expect at these two ages is in why the milestone lists disagree.

What is settled

Warmth and firm limits, together. A meta-analysis of 1,435 studies found that warmth, clear control of behaviour and some room for the child to choose each go with fewer behaviour problems. Harsh control goes with more. So does being permissive. The effects are small and run both ways: difficult kids also draw harsher parenting (Pinquart 2017; grade: strong for the link).5

Praise, and consequences that are connected. A meta-analysis of 154 trials of parenting programmes looked at 26 separate techniques. Three were tied to bigger improvements: positive reinforcement, praise in particular, and natural or logical consequences (Leijten et al. 2019, from the abstract; grade: strong).6 A logical consequence is linked to what happened. The truck that got thrown is put away.

The same analysis found something a tired parent will be glad of. For children without serious behaviour problems, plain behaviour management worked as well as anything more elaborate.

Not smacking. A meta-analysis covering 160,927 children, limited to ordinary open-handed smacking, found 13 of 17 outcomes were significantly worse and none was better (Gershoff and Grogan-Kaylor 2016, from the authors' summary; grade: strong).7 Raising Children Network says plainly that it doesn't help children learn to behave and can lead parents to escalate.8

Most of us were smacked: 62.5% of Australians aged 16 and over, in the Australian Child Maltreatment Study (Haslam et al., from the abstract).9

What is not settled

Time-out. It is often said to damage attachment. A University of Sydney review concluded the opposite: time-out is effective, and inside a warm relationship it doesn't harm attachment. It was never meant to be isolation or a withdrawal of affection, and it only works when there is plenty of good time together the rest of the day (Dadds and Tully 2019, from a summary).10 The Australian Institute of Family Studies says there is no scientific evidence of harm when it is used as part of a broader parenting programme.11

Two honest caveats. Much of this research comes from people who build or teach programmes that include time-out. And I could find no evidence for the "one minute per year of age" rule. Grade: moderate, and contested in public.

Time-in. Staying with a child and helping him calm down is how every small child settles, and good time-out depends on it. But I found no trial comparing time-in alone with time-out (grade: weak as a method by itself).

Emotion coaching. This means noticing the feeling, naming it, accepting it, and then setting the limit anyway. Two Australian randomised trials, one of them with 162 fathers, found parents who learned it reported better behaviour in their children. The children were four and over, the research comes from one group, and the results were mostly reported by the parents themselves (Havighurst et al. 2010; Wilson, Havighurst and colleagues; grade: moderate).1213 It is an addition to limits. It doesn't replace them.

Gentle parenting. The first systematic study of it, 100 parents, found it centres on affection and staying calm. Limits are part of it but applied inconsistently. Over a third of the gentle parents described uncertainty or burnout (Pezalla and Davidson 2024).14 There are no outcome studies under that name. The parts that match warmth-plus-limits and emotion coaching are well supported. The parts that go further, such as never using a consequence, are not (grade: weak).

What that looks like at each age

This is a working approach reasoned from the evidence above, not a tested programme.

18 months 3 years
Main tool Change the room. Move what he can't have. Redirect and distract One clear instruction, up close, then wait
Words One short phrase, the same every time: "Gentle hands" Praise the exact thing: "You put your shoes on when I asked"
Hitting or biting Block it. "No biting. Biting hurts." See to the hurt child first. Move him away briefly Warn once, then follow through. A short, calm, boring break if you use time-out. Then reconnect
Tantrums Hold him or stay close Stay nearby, don't negotiate. Name the feeling afterwards
Don't expect Him to remember next time Sharing on demand. Use turns with a clear end

The questions we are answering together

The evidence leaves a short list of real choices.

  1. What are the three or four things that are never OK in our house?
  2. When the three-year-old hits: time-out, time-in, or one then the other? Where, and for how long?
  3. What is one real example of a connected consequence for our most common problem?
  4. What do we do in a tantrum, and what do we never do?
  5. Are we both ruling out smacking?
  6. Whoever starts dealing with it finishes. What is the signal for "take over"?

The answers go on one page that we can both see. A blank copy of that page will be in Tools.

Triple P Online, developed in Queensland, is free to Australian parents of children under 12.16 Its content is standard and well supported; how big its effect is across a population is disputed (Wilson et al. 2012; grade: moderate, effect size contested).15

Sources

  1. Zero to Three National Parent Survey (summary).
  2. Potegal and Davidson 2003, via review (not verified against the original).
  3. Alink et al. 2006 (not verified against the original).
  4. Smith, Blake and Harris 2013.
  5. Pinquart 2017.
  6. Leijten et al. 2019 (abstract).
  7. Gershoff and Grogan-Kaylor 2016, authors' summary.
  8. Raising Children Network, smacking.
  9. Haslam et al., Australian Child Maltreatment Study (abstract).
  10. Dadds and Tully 2019, summary.
  11. Australian Institute of Family Studies on time-out.
  12. Havighurst et al. 2010, Tuning in to Kids.
  13. Wilson, Havighurst and colleagues, Dads Tuning in to Kids (abstract).
  14. Pezalla and Davidson 2024.
  15. Wilson et al. 2012, Triple P meta-analysis.
  16. Triple P Queensland.

“Read” means the figure was read from the source itself. “Abstract” and “secondary” mean it was confirmed only from an abstract, press release or summary. “Not verified” means it could not be confirmed from the original. How we grade evidence.

What would change this advice

A trial comparing time-in alone with time-out; any outcome study of gentle parenting.