Adolescents take more risks than children or adults. This is consistent across cultures and observed in other species, which suggests it isn't primarily a failure of parenting or education.
Understanding why changes what conversations are likely to work.
The developmental account
A widely discussed model proposes that two systems develop on different timelines.
Systems associated with reward sensitivity and sensation-seeking increase in activity around puberty, driven by hormonal changes.
Systems associated with cognitive control — planning, impulse regulation, resistance to immediate reward — develop gradually and continue maturing into the twenties.
The gap between them, largest in mid-adolescence, is proposed to explain elevated risk-taking.
The model is debated and the timeline is not as clean as popular accounts suggest. The broad observation — heightened reward sensitivity alongside still-developing regulation — has reasonable support.
What this implies
Several things that contradict standard approaches.
Information alone doesn't work. Adolescents generally know the risks. Studies have found that adolescents can articulate risks accurately, sometimes overestimating them, and take the risk anyway.
Which means education campaigns based on informing about danger have limited effect, and evaluations of several such programmes have found exactly that.
Peers change the calculation substantially. Experimental work has found that adolescents take more risks when peers are present or believed to be watching, an effect not seen to the same degree in adults.
This is one of the more robust findings and it means that risk assessment made alone is a poor predictor of behaviour in a group.
Immediate social consequences outweigh distant physical ones. For an adolescent, the immediate cost of refusing something in front of friends is vivid and the distant cost of a health consequence is abstract. That's a rational weighting given how the reward systems are operating.
What works better
Focus on the immediate and social. Consequences that are proximate and socially relevant land better than long-term health outcomes. This has been used effectively in some campaigns.
Provide exit routes. Since the difficulty is frequently social rather than informational, giving an adolescent a way to decline without social cost is practical. Agreed arrangements — a code word in a message that means come and get me, no questions tonight — are used by many families and appear to work.
The no-questions-tonight part matters. If calling for help produces an immediate interrogation, they won't call.
Harm reduction alongside prevention. Given that some risk-taking will occur, information about reducing harm has value. This is contested by people who feel it condones behaviour, and the evidence generally favours providing it.
Autonomy-supportive conversation. Research on parenting styles consistently finds better outcomes from approaches combining warmth with clear expectations and explanation, compared with either authoritarian control or permissiveness.
Practically that means explaining reasoning rather than asserting rules, and negotiating where negotiation is possible.
Monitoring through disclosure rather than surveillance. Research distinguishes between parental knowledge obtained through a young person volunteering information and knowledge obtained through tracking. The former is associated with better outcomes; the latter is associated with reduced disclosure over time.
Which suggests that maintaining a relationship in which they tell you things is more protective than any monitoring technology.
What tends to backfire
Exaggeration. If a warning turns out to have been overstated, credibility on everything else is damaged. Adolescents check.
Sharing your own history selectively. Complicated — honesty can build trust, and framing that implies you took the same risks and were fine undermines the point.
Lecturing. A monologue produces disengagement. Shorter exchanges, more often, work better than a set-piece conversation.
Conversations only when something has gone wrong, which establishes that the topic is associated with trouble.
The timing
Conversations in the car, while walking, or during an activity work better than sitting facing each other. Side-by-side removes the intensity of eye contact and gives both parties something else to do.
They also work better before rather than after — discussing what to do if a situation arises, before it does, rather than reacting afterwards.
The relationship point
The consistent finding across this literature is that the quality of the relationship predicts outcomes more than any specific technique.
An adolescent who feels they can tell a parent something difficult without the relationship rupturing is considerably safer than one who can't, regardless of what rules exist.
Which means the most protective thing is frequently the least specific: maintaining connection through a period when they're actively pushing away, and being someone they'd call.
General information only. Concerns about a young person's wellbeing or safety should be discussed with a qualified professional.
Where they get information
Worth knowing, since it shapes what you are competing with. Adolescents obtain a substantial proportion of their information about risk from peers and from online sources rather than from adults.
Much of that information is inaccurate, and some of it is actively harmful. What it generally is not is judgemental, which is a large part of why it gets used.
The practical implication is that being a source they will actually consult requires being someone who answers questions factually rather than reacting. A single alarmed response to an early question tends to close the channel for years.
It also means checking what they believe rather than assuming. Asking what they have heard about something, without correcting immediately, reveals a great deal.