Reading the phone-and-sleep news story
A radio station's website runs the poll question: 'Do you agree that phones are damaging teenagers' sleep?' 1,240 people respond and 78% say yes. The station reports that '78% of New Zealand teenagers say phones damage their sleep'. The poll link was shared on social media by a parenting group critical of screen use. In the same bulletin, a study of 2,000 14-year-olds is reported: those who used screens for four or more hours a day were twice as likely to report poor sleep a year later as those who used screens for under an hour (16% versus 8%). The headline reads 'Phones double teen sleep risk'. The study measured screen use and sleep by self-report questionnaire at the start and again at the end, and did not randomly assign screen time. It also found that heavy screen users were more likely to report anxiety symptoms and to have a phone in the bedroom.
1. Evaluate the claim that 78% of New Zealand teenagers say phones damage their sleep.
Read the worked answer
The 78% figure describes the people who chose to respond, and nothing wider.
- Self-selection. Respondents decide for themselves whether to take part. People with strong views, or those mobilised by the parenting group that shared the link, are heavily over-represented. The sample is not random and there is no sampling frame of teenagers.
- The population is wrong. The respondents are website visitors, likely including many parents - not a sample of New Zealand teenagers at all.
- No margin of error is meaningful. A '±3%' figure computed from n = 1,240 would assume a random sample. With self-selection, non-sampling error dominates and cannot be estimated; the true figure could be far from 78% in either direction.
- Question wording. 'Do you agree that phones are damaging...' is leading: it presumes damage and offers no neutral or disagree-with-the-premise option, which pushes agreement upward. It also asks about a belief, not about the teenagers' own sleep.
Verdict: the only defensible statement is '78% of the 1,240 people who responded to an online poll said yes'. Generalising to New Zealand teenagers is not supported. A credible estimate needs a probability sample of teenagers, a neutral question, and a reported response rate.
2. Interpret the 16% versus 8% finding, distinguishing relative and absolute risk, and comment on the headline.
Read the worked answer
Relative risk: 16% ÷ 8% = 2.0. Heavy screen users were about twice as likely to report poor sleep as light users. This is what 'double' refers to.
Absolute risk: 16% − 8% = 8 percentage points. In concrete terms, about 8 extra poor sleepers per 100 heavy users compared with light users (16 in 100 versus 8 in 100). That is the figure that matters for how many young people are affected.
The headline. 'Double the risk' is arithmetically correct but rhetorically loaded: doubling a small risk is a small absolute change, whereas doubling a large one is serious. Here the absolute increase is substantial, but the headline gives the reader no way to judge that. It also converts an association into a suggested cause ('phones ... risk'), which the design cannot support, and it drops the comparison group ('compared with under an hour') entirely.
A fairer headline would be: 'Teens using screens 4+ hours a day were more likely to report poor sleep a year later (16% vs 8%)'. Both numbers should appear, plus the fact that the study is observational.
3. Assess whether the study shows that heavy screen use causes poor sleep, and say what evidence would strengthen the claim.
Read the worked answer
No - this is an observational study. Screen use was not assigned, so the groups differ in many other ways:
- Confounding. Heavy users were more likely to report anxiety and to have a phone in the bedroom. Anxiety disrupts sleep; a phone in the bedroom provides notifications and light at night. Either could produce the whole difference without screen time itself being the cause.
- Reverse causation. Poor sleepers may lie awake and use their phones more; the direction could run the other way, or both ways.
- Shared measurement method. Both screen use and sleep came from self-report questionnaires, so recall error and a tendency to give consistent answers inflate the apparent link.
- Dose and content unspecified. 'Four or more hours' bundles passive scrolling, gaming, homework and video calls.
What would strengthen the claim: randomised assignment to a screen-time reduction or a phone-curfew condition, with objective sleep measurement (e.g. actigraphy); adjustment for anxiety, bedtime routine and bedroom phone access in the analysis; pre-registered outcomes; replication in different ages and settings.
Scope: even a strong result would apply to 14-year-olds in this study, over one year - not to all teenagers, and not to long-term outcomes.