Children's Emotional Intelligence: What Is Measured

Key takeaways

Across 47 studies, non-supportive responses to a child's negative emotion were associated with slightly more anxiety and low mood. Supportive responses showed a protective effect only about a third as large.
Elaboration — discussing an emotional event afterwards rather than reassuring during it — showed a protective association roughly twice that of supportive responses generally, though on six studies, far thinner evidence.
The categories are narrower than “be supportive” suggests: solving the problem counts as supportive; the parent's own visible distress counts as non-supportive.
The famous 11-percentile-point achievement gain comes from a meta-analysis of 213 school-based programmes — evidence for structured curricula delivered by trained staff, not for activities at the kitchen table.
The causal arrow is uncertain: an already anxious child pulls more exasperated responses, which could produce this pattern with no parental effect. What survives is its shape, not a verdict on any parent.

The advice arrives as a project. Build your child’s emotional intelligence: name the feelings, do the activities, put the chart on the fridge. It is framed as something you add, like a language or an instrument.

The research on parents and children’s emotions does not describe anything you add. It describes what you do in the thirty seconds after a child becomes upset — and its most useful result is that those thirty seconds are asymmetric.

Not doing the harmful thing is better evidenced than doing the helpful one

A 2025 meta-analysis pooled studies of how parents respond to their children’s negative emotions and what those responses track in the child. It separated supportive responses — comforting, helping with the problem, encouraging the child to express what they feel — from non-supportive ones: minimising it, punishing it, or becoming distressed yourself.

Both were associated with the child’s internalising problems, meaning anxiety and low mood. They were not associated equally.

0.18 vs 0.06

Non-supportive responses correlated with children’s internalising problems at r = 0.18 across 47 studies and 10,970 participants. Supportive responses correlated in the protective direction at only r = −0.06, across 50 studies and 10,698 participants. The association with harm is roughly three times the size of the association with help.

One specific supportive behaviour did better than the rest. Elaboration — actually discussing an emotional event afterwards, rather than reassuring in the moment — came in at r = −0.11, though on far thinner evidence: six studies and 867 participants.

Very few moderators reached significance, and there was no difference between mothers and fathers.

What this meta-analysis can and cannot say

These are correlations, and the causal arrow is genuinely uncertain in both directions. A child who is already anxious is harder to soothe, is upset more often, and pulls more exasperated responses out of a tired parent — which would produce this pattern with no parental effect at all.

What survives that objection is the shape: the dismissive end of the range is consistently and substantially more strongly associated with poor outcomes than the supportive end is with good ones. Whatever is driving it, the difference between “stop crying, it’s not a big deal” and a neutral response is the larger measured gap.

What the research is actually about

The field has a name — parental emotion socialisation — and a framework set out by Nancy Eisenberg and colleagues in 1998. It is deliberately narrow. It concerns how parents react to their children’s emotions, how they talk about emotion, and what emotion they express themselves, and it makes no claim to be teaching a general capacity.

Underneath it sits an older observation. John Gottman and colleagues, working with families in the 1990s, found that parents hold what they called a meta-emotion philosophy: a largely unexamined position on whether emotions are useful information or a problem to be ended. Parents rarely know they have one. Children encounter it every time they are upset.

That is the reframe worth taking from all of this. Your child is not receiving a curriculum. They are repeatedly observing what happens in your house when someone feels something inconvenient.

What counts as supportive is narrower than it sounds

“Be supportive” is not advice; it is a mood. The coding schemes this research uses are specific, and the specificity is the useful part.

On the supportive side sit three distinct things: comforting the child, helping them with the problem that caused the upset, and encouraging them to express what they are feeling. On the non-supportive side sit three others: minimising the reaction, punishing it, and the parent’s own distress at it.

Two of those deserve attention because they are easy to misread. Problem-focused support counts as supportive — solving the thing that went wrong is not, in this literature, a failure to attend to the emotion. And parental distress counts as non-supportive, which surprises people, because it does not feel dismissive from the inside. A parent visibly undone by their child’s unhappiness is nonetheless communicating that the feeling is more than the room can hold.

The categories also do not map onto warmth. It is entirely possible to be an affectionate parent whose reflex, when a child cries about something small, is to explain briskly why it is not worth crying about. That reflex is the measured one.

The school evidence is real, and it is not parenting advice

The most cited number in this whole area comes from somewhere else entirely, and it is routinely borrowed to sell things it does not support.

In 2011 a meta-analysis of 213 school-based social and emotional learning programmes, covering 270,034 students, found improvements in skills, attitudes, behaviour — and an 11-percentile-point gain in academic achievement against controls. A 2017 follow-up meta-analysis found effects persisting at follow-ups ranging from 56 to 195 weeks after the programmes ended.

Those are good results and they are worth knowing. They are also results for structured curricula, delivered by trained staff, over months. None of it is evidence that a set of feelings flashcards at the kitchen table produces an 11-percentile gain, and the frequent slide from one to the other is the single most common misuse of this literature.

What the school work does support is that the underlying skills are teachable and that teaching them at scale changes measurable outcomes. What it cannot tell you is what your Tuesday evening should look like.

Start by subtracting. The best-supported move is not adding a practice; it is removing the dismissive one. “You’re fine”, “it’s not a big deal”, “don’t be silly” and visible irritation are the responses on the strong side of the asymmetry.
Let the feeling exist before you solve the problem. Most non-supportive responses are not cruelty; they are efficiency. The upset is inconvenient and ending it is the fastest route. Ending it is the thing being measured.
Come back to it afterwards. Elaboration — discussing what happened once everyone is calm — was the single supportive behaviour with the strongest association, on the thinnest evidence. It costs a conversation on the way to school.
Watch your own reaction, not just your reply. Becoming distressed yourself counts as non-supportive in this research. A parent visibly overwhelmed by a child’s feeling teaches the same lesson as one who dismisses it.
Do not expect a large effect. These are modest correlations, not levers. The honest expectation is a small, cumulative shift in what your child learns about whether feelings are safe to have out loud.

Notice that four of those five are about restraint. That is not an accident of how they were written; it is what the effect sizes say.

Try this

For one week, keep a private count. Every time your child is upset, note only whether your first response was to end the feeling or to acknowledge it. No score, no judgement — just which one.

Most parents are surprised by the ratio, and by how strongly it tracks their own tiredness rather than anything about the child.

That is the actual finding in domestic form: the variable is not your knowledge of emotional intelligence. It is what you have left at six o’clock.

It does not stop mattering when they get older

The meta-analysis covered childhood and adolescence, and found very few significant moderators — including, notably, little that distinguished younger children from older ones.

That is worth saying because the advice tends to evaporate around age eleven, as though emotion coaching were a phase of parenting rather than a way of responding. The window in which a teenager volunteers a feeling is narrower and closes faster, which raises the cost of getting the thirty seconds wrong rather than lowering it.

The parts of this that are genuinely hard

The reverse-causality problem is not a technicality. Children differ enormously in how much emotional support they require, and the parent of a highly reactive child is running the same experiment on much harder settings. Reading a correlation of 0.18 as a verdict on parenting would be both statistically wrong and cruel.

Much of this literature also rests on parents reporting their own behaviour, which is exactly the kind of measurement that flatters. And “internalising problems” is a broad outcome measured many different ways across fifty studies.

The deepest difficulty is practical rather than statistical. Responding well to a child’s distress requires regulating your own first, and nothing in the research makes that easier at the end of a long day. The finding that non-supportive responses matter more is genuinely useful information, and it is also, for an exhausted parent, an unwelcome thing to be told. Both are true.

The bottom line

Emotional intelligence is not a thing you install in a child through activities. What has been measured is far narrower and more ordinary: what happens in your house when someone is upset, repeated a few thousand times.

And the measurement is lopsided in a way that should be a relief. You do not have to become a better emotional educator. The larger and better-documented difference is made by not shutting the feeling down — which is a lower bar than the advice industry has any interest in telling you about.

Sources (5)

Every factual claim in this article comes from one of these sources.