Kept Company
Kept Company

Does mentoring work

Mentoring works, and the size of the benefit is small: around 0.2 of a standard deviation on average across youth mentoring programmes, in meta-analysis by DuBois and colleagues in 2011.

Who this is for

A coordinator deciding whether to start or continue a mentoring scheme, and a volunteer weighing up a year of Tuesday evenings against what that year is likely to achieve.

What it protects

The year. Knowing the effect is small and comes from duration stops a scheme spending its goodwill on matches that fold by spring.

What goes wrong without it

A scheme is designed around matches made rather than matches surviving, reports a good first year, and never notices that its short matches are the ones doing damage.

Drawing: a kettle and two mugs on a worktop.

That is a real effect and not a large one. The average mentored young person ends up slightly better off on the measured outcomes than the average unmentored one, which is not the same as a mentor changing a trajectory. The finding that should shape a scheme is the second one: a well run programme beats a badly run one by more than mentoring beats no mentoring. What a scheme does about matching, training, supervision and match length matters more than whether it exists.

The size of the effect, and what 0.2 does not mean

The size of the effect is a shift in an average, not a change in a person. An effect of 0.2 of a standard deviation shows up reliably when hundreds of young people are compared with hundreds of others, and is invisible in any single match. Three conversions of that figure are wrong and all three appear in sector literature: it is not a 20% improvement, it is not a proportion of young people helped, and it does not convert into a reoffending rate, an attainment grade or a saving to a public service. No number of that last kind appears on this site, because none of the published ones survives scrutiny. Modest is not worthless: a small effect obtained cheaply and at scale is a reasonable thing for a council or a school to buy, and the honest way to bid for it is to say so.

Six conditions under which mentoring produces measurable benefit

Mentoring produces measurable benefit under conditions the research points at consistently, and every one of them is a design decision rather than a matter of luck.

  1. Ask for twelve months, and mean it. Twelve months is the minimum useful match length in youth mentoring, and it is the ask for a reason.
  2. Screen volunteers with references, an interview and the right check for the nation: a DBS check in England and Wales, PVG scheme membership in Scotland, an AccessNI check in Northern Ireland.
  3. Train before the first meeting, not after the first problem.
  4. Match on something real, such as interests, availability and what the young person says they want, rather than on who is next on the list.
  5. Supervise monthly, so a mentor who is drifting is noticed before the young person notices.
  6. Fund the coordination, which is 60% to 80% of the cost of a scheme and the part that produces the effect.
The ask
About one hour a week52 weeks

The far more common conditions under which it produces almost none

The conditions under which mentoring produces almost no measurable benefit are more common than the conditions under which it works, and they are recognisable in advance. A scheme built on a short term grant that can only promise six or eight weeks of contact is the clearest case: a mentoring scheme that cannot ask for a year should not be recruiting children into it. A scheme that matches on availability alone, trains with a single evening briefing, holds far more than 30 to 50 active matches per coordinator, or measures success by matches made rather than matches surviving, will produce numbers that look busy and outcomes that sit at zero. There is also a case where the effect is not merely absent but negative: matches that begin and then end before three months are associated with outcomes worse than no match at all, reported by Grossman and Rhodes in 2002. A young person referred because adults do not stay has been shown one more adult who did not stay.

If this happens

A mentor tells the coordinator, in week five, that they are not sure they can keep this up

The coordinator treats this as the most useful thing anyone will say all month, and does not talk them into staying. They agree either a realistic pattern the mentor can hold for the rest of the year, or a planned ending in which the young person is told what is happening and why, by the mentor, in person. A planned ending at week six is a manageable disappointment. A mentor who fades out over three months is the failure the research warns about.

Mentoring, coaching and training are three different methods

Mentoring, coaching and training differ in what they work on, how long they take and how they are judged, and calling one by another’s name is how schemes end up measured against the wrong outcome.

MethodWorks onTypical lengthJudged by
MentoringThe relationship first, goals second12 months or moreWhether the match survived and what changed over it
CoachingDefined performance goalsA short structured seriesWhether the goal was met
TrainingA defined body of knowledge or skillFixed and shortWhether the learner can now do it
Peer mentoringThe relationship, between people of similar standingA term or a yearWhether both parties gained

Mentoring and training are not the same method, and a workplace programme described as mentoring but structured as training will be assessed as training and will look weak. Befriending is different again, because it is not working towards a goal at all.

What the mentor gets out of it

Mentors report confidence, communication skill, a clearer view of a community they did not know, and for younger volunteers, something real to say at an interview. The evidence for those gains is weaker than it looks, because it is collected almost entirely from mentors still in post to be asked. Volunteers who felt unsupported or left early are largely missing from it. A scheme may honestly say that most people who complete a year are glad they did. It should not tell them mentoring will change their life either.

The mentor

  • Commits to about an hour a week for twelve months, and says no at the start if that is not realistic.
  • Tells the coordinator early when the commitment is slipping, rather than reducing contact quietly.
  • Ends the match in a planned way, face to face, if it has to end at all.

The young person

  • Is told at the outset how long the match is expected to last and what happens when it finishes.
  • Is not asked to fill in a measure they have not had explained to them.
  • Is told the truth about why a mentor has left, in words appropriate to their age.
And the scheme

The scheme decides whether the effect exists at all, because the effect lives in matching, training, supervision and length. It carries three obligations neither party can carry alone: not to open a match it cannot resource for a year, to hold each coordinator to roughly 30 to 50 active matches so supervision is real, and to report match survival alongside match numbers.

Where to go next