The phrase is used loosely, the underlying behaviour is real, and the evidence is better than the headlines suggest. This page sets out what the term means clinically, when the mechanism was built, how many people meet the criteria, and what the measured effects are, with the study behind every number.
No diagnostic manual contains it. The DSM-5 has no entry for smartphone addiction, and the ICD-11 recognises gaming disorder but not phone use in general. What exists instead is a research literature built on self-report scales, the most widely used being the Smartphone Addiction Scale, Short Version, a ten item instrument published by Kwon and colleagues in 2013 and since translated into dozens of languages.
Those scales do not measure hours. They measure the features that make any behaviour compulsive: using the phone longer than intended, failing at attempts to cut down, checking it first thing and last thing, feeling restless when it is not to hand, and continuing in situations where the use is clearly costing something. A person on the phone for four hours of deliberate work scores low. A person who checks it ninety times without deciding to once scores high.
This distinction matters for the whole argument. Screen time is a measure of quantity and it is a poor predictor of harm. Loss of control is a measure of quality, and it is the part that predicts the outcomes further down this page.
The behaviour is recent enough to date precisely. Its ingredients were added one at a time, each for a defensible commercial reason, and none of them was designed as an addiction mechanism. The result was assembled in roughly a decade.
The underlying schedule is much older. Variable ratio reinforcement, in which a reward arrives after an unpredictable number of actions, was described by B. F. Skinner in the 1950s and is the most persistent reinforcement schedule known in behavioural psychology. Slot machines run on it. So does a feed that sometimes contains something worth seeing.
| Year | What was added | Effect on the behaviour |
|---|---|---|
| 2006 | Infinite scroll, patented by a Microsoft search team in February and widely credited in public debate to Aza Raskin, who has since said he regrets it | Removes the page boundary, and with it the natural moment to stop |
| 2006 | The algorithmic news feed | Content is ordered by predicted engagement rather than by time |
| 2007 | The iPhone | The feed is now in a pocket, available in every gap in the day |
| 2008 | App stores and third party apps | Competition for attention becomes a market with measurable winners |
| 2008 | Pull to refresh, built by Loren Brichter for Tweetie | A deliberate physical gesture that produces an unpredictable reward, the slot machine handle |
| 2009 | Push notifications and the like button | The device initiates contact, and social approval becomes a countable quantity |
| 2016 | Fully algorithmic short video feeds | No follow graph required, the schedule tunes itself per person within minutes |
| 2018 | Screen Time and Digital Wellbeing ship | The platforms concede the problem exists, and make the counter-measure optional |
Read down that column and the pattern is plain. Each step either removed a stopping point or made the reward less predictable. Nothing in the list required the user to be weak.
The best single estimate for young people comes from a 2019 systematic review and meta-analysis in BMC Psychiatry. Sohn and colleagues pooled 41 studies covering 41,871 children and young adults and found a median prevalence of problematic smartphone use of 23.3 per cent, with an interquartile range of 14.0 to 31.2 per cent. Roughly one in four.
The spread between countries is larger than the spread between people. Olson and colleagues pooled 83 samples from 24 countries, 33,831 participants surveyed between 2014 and 2020, and found that country level factors accounted for 74 per cent of the variation in scores. China and Saudi Arabia scored highest, followed by Malaysia, Brazil, South Korea, Iran, Canada and Turkey. Germany and France scored lowest. Scores also rose year on year across most countries over the period studied.
For context on the denominator, DataReportal put the number of unique mobile phone users at 5.83 billion in April 2026, with 6.12 billion people online and 5.79 billion active on social platforms. A quarter of the young users of a technology at that scale is not a niche clinical population.
The same meta-analysis reported the associations below. These are odds ratios from cross-sectional data, so they establish that problematic use travels with these outcomes, not that it causes them. The confidence intervals are shown because three of the four are wide.
Attention is affected even when the phone is not used. Ward and colleagues, working with nearly 800 participants, found that the mere presence of a person's own smartphone reduced available cognitive capacity on tests requiring full concentration. Phones in another room produced the best performance, phones face down on the desk the worst. Whether the phone was switched on or off made no measurable difference.
The interruption cost compounds it. In observational studies of office work, Mark and colleagues found that after an interruption people took roughly 23 minutes to return to the original task. That figure was measured for workplace interruptions in general, not for phones specifically, but the phone is now the most frequent source.
The volume is documented. Common Sense Media instrumented the phones of about 200 11 to 17 year olds for a week in 2023. Half of them received 237 or more notifications a day, some received thousands, and 23 per cent of those alerts arrived during school hours. Median pickups exceeded 100 a day.
Reported use has kept climbing. The global average sits near 4 hours 37 minutes a day on the phone, with the United States near 5 hours 16 minutes, against 4 hours 45 minutes a day of internet use across all devices and 2 hours 39 minutes on social and video apps.
Any honest page on this subject has to include the counter-evidence, and it is substantial. Orben and Przybylski analysed three large datasets covering more than 17,000 adolescents in the United Kingdom, the United States and Ireland, and found that technology use explained at most 0.4 per cent of the variation in adolescent wellbeing. In the same analysis, wearing glasses had a more negative association with wellbeing than screen time did.
Their target, though, is the quantity measure. Total hours is a crude variable that lumps a video call with a grandparent together with three hours of passive scrolling, and it predicts almost nothing. The prevalence and outcome studies further up this page use compulsion scales rather than hours, which is why they find larger effects. Both results can be true at once: time on a screen is a weak predictor, loss of control over that time is a strong one.
The practical implication is not that the problem is imaginary. It is that a person who wants to change something should stop measuring hours and start measuring pickups they did not decide to make.
Put the findings in order and a single mechanism accounts for most of them. The pickup is cued by a gap rather than by a decision: the queue, the ad break, the pause between two tasks, the ten minutes in bed. The reward is unpredictable, which is the schedule that sustains behaviour longest. The stopping point has been engineered out. The device also initiates contact hundreds of times a day, so the gap does not even have to be found.
Every part of that runs below deliberation, which is why advice pitched at deliberation keeps failing. Resolutions, screen time dashboards and app deletion all assume a moment of choice that the behaviour does not contain. Deleting one app reliably moves the reflex to the nearest substitute, because the reflex was never about the app.
What is left is the reach itself. An intervention has to sit there, it has to impose a cost at the moment the hand moves, and the cost has to be one the person cannot waive for free, or it stops mattering within days.
Most blockers take the opposite route and try to make the block impossible to remove, which is why they end up deleted. If you are weighing up the paid apps in this category, the Opal Alternative page sets the two approaches side by side, with prices.
