Screen Time: What the Evidence Actually Supports (and What It Doesn't)
One screen-time finding replicates reliably. Most of the rest is contested. Here is the difference, and what it means for the rule you set on a Tuesday night.
The screen-time finding with the strongest evidence is that screen use in the hour before bed delays sleep onset. Effects on attention and mood are more contested, and displacement — what the screen replaced — explains more than the raw daily total does.
Screen time is discussed almost entirely in moral terms and almost never in arithmetic ones. It is worth doing the arithmetic first, because it changes the question.
Four hours a day, from age three to eighteen, is 21,900 hours. That is about 912 full days. It is roughly 26% of a childhood's waking life.
Whether that matters is a real question with a contested answer. How much it adds up to is not a question at all — it is multiplication, and almost nobody does it.
The finding that replicates
If you take one thing from the screen research, take this: screen use in the hour before bed delays sleep onset.
This result appears across study designs, age groups and countries. It is stronger for a phone held close to the face than for a television across a room. It survives adjustment for most of the obvious confounders. And it has a clear mechanism — light exposure plus cognitive arousal at exactly the moment the body is trying to wind down.
It is also, conveniently, the finding with the clearest fix. You do not have to reduce total screen time to act on it. You have to move one hour.
The findings that are contested
Attention. Measured sustained attention has declined on some tasks over the period screens became ubiquitous. Whether screens caused it is much less clear. The studies that separate short-form video from screen use generally tend to find larger associations for short-form — which is at least consistent with a mechanism — but the designs are mostly cross-sectional, and children who already struggle to sustain attention may simply prefer shorter content.
Mood and anxiety. The literature here is genuinely divided, and the division is not between good and bad researchers. It is between studies that measure total time (which find small effects) and studies that measure specific behaviours like social comparison and appearance-focused content (which find larger ones). "Screen time" is too coarse a variable to settle it.
Language development. Reasonably strong for the under-threes, where passive video appears to displace the back-and-forth conversation that drives vocabulary. Much weaker after that.
Why "screen time" is a bad variable
A video call with a grandparent, a maths app, a Roblox session with three friends, and two hours of autoplay short-form video are all "screen time". They have almost nothing in common.
The research that replicates best separates three things:
- Content — what is on the screen
- Context — alone or with someone, chosen or served
- Timing — when in the day
A single number that ignores all three is going to behave inconsistently across studies, and it does.
The displacement question
The most useful reframing available to a parent is this: not how many hours, but instead of what.
Two hours that replace television is a different proposition from two hours that replace sleep, outdoor play or reading. The screen is often not the variable that matters; it is the thing that got pushed out.
This is why the same daily total produces such different outcomes across households, and why a rule that just reduces the number often achieves less than one that protects a specific thing — bedtime, mealtimes, one hour outdoors.
What the guidance actually says
| Age | Common guidance | What it counts |
|---|---|---|
| Under 18 months | Avoid, except video calls | All screen media |
| 18–24 months | High-quality content, watched together | Apps and video |
| 2–5 years | Up to 1 hour a day | Recreational media |
| 6–12 years | Consistent limits, commonly ~2 hours recreational | Games, video, social |
| 13–18 years | Consistent limits, protect sleep and activity | Recreational media |
| All ages | No screens in the hour before bed | Everything |
Two things worth noticing. The guidance excludes schoolwork and video calls, which most people forget when they read their weekly report. And after age five it stops giving a number and starts saying "consistent" — because the evidence supports consistency better than it supports any particular threshold.
What actually works in practice
Ranked roughly by evidence strength and ease of implementation:
- Devices out of the bedroom overnight. Strongest evidence, easiest to enforce, single biggest sleep effect.
- A screen-free hour before bed. Directly targets the best-supported finding.
- Written, agreed rules rather than verbal ones. Inconsistent rules are worse than lenient ones, because the child learns the rule depends on your mood.
- Adults following the same rules. In most household surveys the adults use screens more than the children, and children notice.
- Co-viewing under five. Changes outcomes more reliably than reducing total time does.
- Capping short-form specifically rather than screens generally.
Notice what is not on that list: a specific number of hours. It is not there because the evidence does not support one, and pretending otherwise makes the rest of the advice easier to dismiss.
What to expect if you change something
Families who sustain a reduction tend to report a similar sequence: two or three days of genuine resistance, a week of boredom, then sleep changes around week two and attention or mood changes around weeks three to four.
The reason most reductions fail is that they are abandoned on day three, which is the peak of the resistance and the point at which it looks like it is not working.
The honest summary
Sleep is the strong finding. Displacement is the useful frame. Consistency beats any specific number. And a large share of what gets blamed on screens may be the sleep the screens cost.
Everything beyond that is a live research question, and anyone who tells you otherwise is either selling something or has not read past the headline.