Skip to content

Mentoring and Coaching

Principle · All principles
Evidence · 67 claims (32 for, 28 mixed, 7 against) · 50 studies (12 causal, 11 review, 10 associational, 9 qualitative, 3 quant-synthesis, 3 design, 2 theoretical), q1–q4 · 6 of 50 report an effect size · 63 claims rest on one study

Conditional relationship

This page owns the relationship itself: a learner who is doing real work in a role, programme or practice that is new to them, paired over time with a more experienced person (a mentor teacher, an instructional coach, a success coach, a faculty or research mentor, a near-peer, or, increasingly, a software coach). The relationship is expected to help when the more experienced person sees the learner's actual work (a lesson taught, a draft, a lap flown, a semester's choices), responds to what that work shows with modelling, questions and specific feedback, meets often enough to follow the learner through a cycle of attempts, and hands judgement over to the learner as the learner succeeds. Under those conditions, mentoring and coaching are expected to improve observed performance in the practice (skill-performance, observed in the setting) and persistence in the role or programme (staying in teaching, re-enrolling in college), more than no such relationship or one that is infrequent, generic, mostly about logistics, or tied to the learner's evaluation. Effects on downstream outcomes (the coached teacher's students' test scores, a degree completed, earnings) are expected to be smaller and less certain than effects on the practice coached.

The relationship is conditional on four things: contact is anchored in the learner's own work, not in general advice; there is enough continuity for a cycle of attempt, feedback and reattempt; the person in the support role has the skill and time to observe and give specific feedback; and the role is separated from judging the learner, so that the learner can show what they cannot yet do.

What the wiki's evidence tests, and what it does not. The strongest test is of teachers coached on their classroom practice: a meta-analysis of 60 causal studies (Kraft et al. 2018, several claim pages) found coaching raised observed instruction by a pooled 0.49 SD and students' test scores by 0.18 SD (0.12 SD on state tests), with effects shrinking as programmes grew and no relationship with hours of coaching. For college students, a randomised evaluation of one success-coaching programme (InsideTrack, reviewed by the What Works Clearinghouse) found higher persistence (about five percentage points, effect size 0.13) and no discernible effect on degree completion, and a matched comparison of another (Success Boston) found higher persistence and no difference in grades. For new teachers, one large randomised study of comprehensive induction (intensive mentoring plus professional development) found no first-year improvement in practice, retention or achievement, while correlational surveys and policy reviews associate strong mentoring with plans to stay and lower attrition. For software coaches, one small study found large within-learner gains in a motor skill from an adaptive coach and weaker or no gains from simpler baselines, and one pilot found no advantage for a scaffolded study coach. The rest of the cited claims are descriptive: how much mentoring people receive, what it covers, what mentees and coaches value. No claim in this wiki isolates the components this page names (observation of real work, contingent feedback, fading, role separation) as causes; they are carried from neighbouring evidence and labelled where used.

How the converted siblings sit around this page: the Professional Development pattern owns the full coaching cycle for one teaching practice (diagnose, model, rehearse, enact with feedback, fade, check maintenance) and its doses, and is where the teacher-coaching meta-analysis is read in detail; this page does not repeat it. Ongoing study teams and peer coaching owns which ongoing school structure carries teachers' learning (study teams, peer coaching, a school-based coach) alongside occasional sessions. Mentor Support in Self-Paced Online Learning owns mentoring when learners study asynchronously: on-demand and proactive contact, progress signals and the split between automated and human outreach. Cognitive Apprenticeship owns the modelling, coaching, scaffolding and fading of a complex cognitive skill within instruction. Scaffolding and Fading owns how support within a task is raised and withdrawn. This page holds what is common to all of them: who the mentor or coach is, what the relationship is for, how often and about what they meet, how the role is kept separate from evaluation, and how the relationship ends.

Default design, while the relationship is untested

The earlier page's guidance (a clear developmental focus, trust and continuity, calibrated support, a path toward independence; risks of overdependence, uneven quality, role confusion and unequal access), kept as a concrete default and extended into an arc. Each step says what evidence, if any, stands behind it; numbers are untested proposals unless a claim is named.

  1. Name what the relationship is for, in terms of the learner's work. One or two targets the mentee can see in their own work ("questions that get more than one student talking", "a weekly plan you keep", "a lap without a collision"), agreed in the first meeting and revisited each month. In one state survey, first-year teachers' mentoring was most often about noninstructional topics (The mentoring topics first-year teachers most frequently received guidance on were noninstructional rather than instructional [~W]); and in coached student teaching, coaching on specific areas of instruction went with better-prepared and more effective first-year teachers (Student teachers who perceived their mentors modeled effective instruction and provided coaching, feedback, and job assistance felt better prepared, and domain-specific coaching was related to higher first-year REACH ratings [+W]; one Chicago survey, associational). The targeting rule itself is untested.
  2. Anchor every contact in an observation of real work. The mentor watches the learner do the work (sits in on a lesson, reads the draft, reviews the semester plan, watches the run) or reviews a recording or a shared folder, before talking about it. Contact without observation becomes advice. Access to work in progress is what made coaching possible in one web-based course (Instructor access to students' individual Web folders enabled coaching on work-in-progress in the Web-based course [+W]); in one state, a fifth of first-year teachers had never been observed by their mentor (Time spent on specific mentoring activities varied widely, with some teachers never observed by mentors and others observed an hour or more a month [~W]). The rule is a proposal.
  3. Respond to what the work shows: model, ask, give specific feedback, set the next attempt. More help after a failure, less after a success; show how an expert would do the step that went wrong, ask the learner to explain their choice, and agree one change for the next attempt. Contingent support: Contingent scaffolding improves learning more than fixed or absent support. [+M] (one-to-one tutoring of children and a dynamic-assessment meta-analysis; carried from tutoring a task to mentoring a practice). An adaptive software coach that varied its help by the learner's skill produced large pre-post gains in an unassisted motor task, where simpler baseline coaches did not (L2C coaching significantly reduces learners' lap times and failure counts from pre- to post-coaching in FPV drone racing [+M]; Baseline coaches produced weaker or absent learning gains: RBF improved only failure count, MIA neither outcome [-W]; 33 adults, drone racing, one study).
  4. Meet on a regular rhythm, and protect it. A default of a short check-in weekly and an observation with a debrief every two to four weeks for the first term, then monthly. Hours alone are not the lever: across causal studies of teacher coaching, more hours did not go with larger effects (Coaching dosage shows precisely estimated null effects: more coaching hours are not associated with larger program effects [~M]; exploratory, between programmes). The rhythm is an untested proposal; what it protects is step 2.
  5. Keep the mentor out of judging the mentee. The mentor or coach is not the person who rates, grades or decides on the learner's continuation; if that cannot be avoided, say which conversations are evaluative and which are not. Teachers who saw coaching as voluntary and non-evaluative reported more improvement (Teachers who report coaching as a partnership report more improvement in classroom challenges and higher technology-use confidence; voluntary participation and non-evaluative support correlate with improvement in classroom challenges [+W]; one programme's survey correlations); teachers whose principal both coached and evaluated described hiding what they struggled with (Administrators' dual role as evaluator and coach can undermine the observation process's professional learning benefits [-W]; interviews in six schools).
  6. Choose and prepare mentors for the job, not only for their own performance. Select mentors who can observe and explain, give them preparation in giving feedback, and check what their meetings cover. In Chicago student teaching, mentors' formal qualifications were generally unrelated to their student teachers' outcomes while their own evaluation ratings were related (Mentor teachers' own evaluation ratings were positively related to their student teachers' first-year effectiveness as teachers-of-record [~W]); principals in another Chicago study often lacked the coaching skill for deep conversations about practice (Principal-teacher conferences were perceived as more reflective and objective than past evaluations, but many principals lacked the instructional coaching skills for deep discussions of practice [~W]). Preparation content and length are untested.
  7. Offer a say in the match. Let mentees state preferences (field, background, language, schedule) and change a match that is not working, without making matching on any one characteristic a rule. Apprentices of colour in one programme described mentors of colour as especially valuable and asked for more intentional matching (Mentors of color were especially valuable to apprentices of color, who wanted more intentional mentor matching [~W]; interviews in one programme); a review in sport reports same-gender coach-athlete pairs rating their relationship higher (Same-gender coach-athlete dyads perceive higher relationship quality (closeness, commitment, complementarity) than other-gender dyads [~W]). These are learners' perceptions of a relationship, not learning outcomes, and they describe what mentees value, not what any group needs.
  8. Reach the learners who would not ask. Assign a mentor or coach to every learner in the target group, with the mentor making the first contact, rather than relying on learners to seek one out. In the one randomised evaluation of student coaching recorded here, coaches were assigned by lottery and reached out to students, and persistence rose (InsideTrack© Coaching has potentially positive effects on college credit accumulation and persistence, based on one RCT meeting WWC standards [+M]). The earlier page's warning that informal mentoring favours learners who already know how to seek support is a proposal the wiki holds no test of.
  9. Hand over judgement, then end or change the relationship. Over the arc, the mentee moves from watching the mentor model, to doing with the mentor's prompts, to doing alone and explaining their reasoning, to setting their own next target; the mentor asks more and tells less. End or change the relationship when the mentee can meet the agreed targets unaided and plan their own next step. Fading support promotes the transfer of responsibility from instructor to learner. [+M] (a scaffolding review and worked-example experiments; carried from fading within a task to fading a relationship).
  10. Plan the arc across sessions. A default, all untested proposals:
    • First two to four weeks (setting up). One meeting to agree targets and ways of working; one observation of the learner's work and a debrief; weekly 15–30 minute check-ins. The mentor models more than asks.
    • Rest of the first term (cycles). An observation and 30–45 minute debrief every two to four weeks, each ending with one change to try; the learner brings evidence of the change to the next meeting. The mentor asks the learner to analyse their own work before giving feedback.
    • Later terms (handover). Monthly meetings led by the learner, who brings their own target, evidence and plan; the mentor comments rather than directs. The relationship ends, or becomes occasional, when the learner meets the targets unaided for two cycles in a row.

When a condition is not met: if the mentor cannot observe real work, treat the relationship as support and advice and do not expect change in the practice; if the mentor must also evaluate, keep evaluative and developmental meetings apart and tell the learner which is which; if there is no time for regular contact, put the time into fewer observed cycles rather than more unobserved meetings.

Fitting the design to a situation

The three facts that most change the decision:

  • What the learner is doing and what the relationship should change. Coaching an experienced teacher's practice, inducting a new teacher, keeping a first-generation student enrolled and helping a trainee master a motor skill are different jobs with different outcomes and different evidence. If unstated, ask: whose work will the mentor see, and what would be different in it after a term?
  • Who the mentor is and what else they are to the learner. A trained coach, a colleague, the learner's evaluator, a near-peer and a software coach bring different skill, time and power. If unstated, ask: who is available, how much time do they have each month, and do they also rate or grade this learner?
  • Scale. One mentor for a few learners can follow each learner's work; a programme across many sites cannot assume the same quality. If unstated, ask: how many learners per mentor, how many sites, and who prepares and supports the mentors?
If the brief says… Then change… Basis
Learners are novice in the role: first-year teachers, apprentices, new graduate students Weekly contact at first; put instruction or the core practice, not logistics, at the centre of each meeting; observe early Strong mentoring is linked to novice elementary teachers being 25 percentage points more likely to plan to remain in the same school, with a similar but non-significant pattern for high school teachers [+W] (survey, plans to stay); Comprehensive teacher induction generated no significant improvements in novice teachers' classroom practices in the first year [-M] (randomised; do not promise first-year gains in practice)
Learners are experienced practitioners changing one practice (in-service teachers, principals) Run coaching cycles on one observable practice and pair them with group sessions and materials; use the Professional Development pattern Teacher coaching programs improve teachers' instructional practice by a pooled 0.49 SD across 43 causal studies [+S]; Pairing coaching with group trainings is associated with larger effects on instruction and achievement; instructional materials with larger, and video libraries with smaller, effects on instruction [~M] (between-programme association)
Learners are adult college students at risk of leaving Assign a coach to each student with proactive outreach on enrolment, planning and obstacles; judge the programme on persistence first, and do not expect grades or completion to follow automatically InsideTrack© Coaching has potentially positive effects on college credit accumulation and persistence, based on one RCT meeting WWC standards [+M]; InsideTrack© Coaching shows no discernible effects on college degree completion (attainment) [~M]; In the Linkow et al. (2019) scale-up QED, coaching showed no statistically significant effect on cumulative college GPA (effect size 0.01, p = 0.82) [-M]
Learners are child or adolescent school students A named adult or near-peer who meets the learner regularly about their own schoolwork, with the teacher's knowledge; keep sessions short and frequent untested proposal; the wiki's mentoring evidence is about adults (teachers, college students, trainees)
The mentor or coach also evaluates the learner (a principal, a supervisor, a course instructor) Separate developmental and evaluative meetings, say which is which, and keep notes from developmental meetings out of ratings; where possible give the coaching role to someone else Administrators' dual role as evaluator and coach can undermine the observation process's professional learning benefits [-W]; Teachers who report coaching as a partnership report more improvement in classroom challenges and higher technology-use confidence; voluntary participation and non-evaluative support correlate with improvement in classroom challenges [+W]
Scarce mentor time: an hour a month or less Spend it on one observed cycle (watch, debrief, one change) rather than several unobserved conversations; let peers or a shared folder carry the rest Coaching dosage shows precisely estimated null effects: more coaching hours are not associated with larger program effects [~M]; Time spent on specific mentoring activities varied widely, with some teachers never observed by mentors and others observed an hour or more a month [~W]
Remote or virtual coaching Use video of the learner's work and video calls for the debrief; keep the same cycle Coaching effects do not differ significantly across school levels, and in-person versus virtual delivery shows no significant difference [~M] (underpowered; equivalence not shown)
online-self-paced, no live teacher Use proactive outreach triggered by progress signals and on-demand contact untested here; modelled in Mentor Support in Self-Paced Online Learning
A software or AI coach instead of, or beside, a person Use a coach that adapts its help to the learner's observed performance and withdraws it; test unaided performance after coaching, not only satisfaction L2C coaching significantly reduces learners' lap times and failure counts from pre- to post-coaching in FPV drone racing [+M]; A scaffolded AI Study Coach produced no assignment-performance or concept-inventory advantage over unrestricted AI use in a master's programming course pilot [-M] (pilot, no inferential tests)
A programme across many sites or a whole system Expect smaller effects than in pilots; budget for selecting, preparing and supporting coaches, and monitor what meetings actually cover Coaching program effects shrink substantially at scale: effectiveness trials of larger programs show only a fraction of efficacy-trial effects [~M]; In five scaled early learning coaching systems, coach-to-site ratios averaged about 1 coach per 22 sites, dosage varied widely, and participants perceived benefits in educator satisfaction, practice, and program quality [~W]
Mentor quality is uneven or untrained mentors are all there is Give mentors a short preparation in observing and giving feedback, a meeting template, and someone who reviews their meetings Mentorship was important to apprentices, but mentor quality and time with mentors varied across programs [~W]; Principal-teacher conferences were perceived as more reflective and objective than past evaluations, but many principals lacked the instructional coaching skills for deep discussions of practice [~W]; the preparation itself is untested
Mentees ask to be matched by shared background (race, gender, language, first-generation status) Ask every mentee for matching preferences and honour them where possible; let a mismatch be changed without cost; do not assume what any group wants Mentors of color were especially valuable to apprentices of color, who wanted more intentional mentor matching [~W] (mentees' accounts in one programme, not outcomes)
The outcome the brief cares about is students' achievement, not the coached adult's practice Keep the coaching on the practice most closely tied to that achievement, add materials, and expect a much smaller effect on achievement than on practice Teacher coaching raises student achievement on standardized tests by a pooled 0.18 SD across 31 causal studies [+M]; Coaching effects on instruction and achievement are positively correlated, but large instructional improvements are needed for modest achievement gains [~M]
The brief names its own mentoring routine, categories or rubric (its own induction standards, observation form or coaching cycle) Keep that routine and its categories; check only that each contact is anchored in observed work, the mentor is not also the evaluator, and the relationship hands over judgement untested proposal
Observations, recordings or meeting notes are sensitive (a teacher's video, a student's academic or personal difficulties) Agree who sees recordings and notes before the first meeting; keep them out of evaluation and records unless the learner agrees; delete recordings after the debrief untested proposal
One meeting only Treat it as advice, not mentoring: review one piece of the learner's work and agree one change; do not claim mentoring effects untested proposal

Eleven of the sixteen rows cite a claim. The teacher-coaching rows carry a meta-analysis of causal studies; the student-coaching rows one randomised and one matched evaluation; the rest carry surveys, interviews or one study each, and are labelled as such.

Observation, state and explanation

What can be observed: who meets whom, how often and for how long; whether the mentor saw the learner's work before the meeting; what each meeting covered (instruction, logistics, emotional support, career); the change agreed and whether it appears in the learner's next piece of work; the learner's unaided performance on the target; and whether the learner stays in the role or programme. "Supported", "dependent", "trusting" and "ready to work alone" are inferences.

Response Plausible explanations Observation that could change the interpretation
The mentee reports the mentoring helped, but their observed practice does not change The meetings were about logistics or reassurance, not the practice; the mentor never saw the work; the target is too broad; the change needs longer than a term Read the meeting notes for topics; check whether an observation preceded each debrief; look for the agreed change in the next lesson or draft.
The mentee's work improves while the mentor is present and falls back without them The mentor is still doing part of the work (prompting, deciding); the support was never withdrawn; the improvement was in compliance, not in judgement Observe one cycle with no prompts; ask the mentee to plan the next change themselves and see whether it is sound.
The mentee stops coming or says little in meetings The mentor also evaluates them; the meeting time clashes with something they value; the match is poor; the mentee does not see what the meetings are for Ask privately (or through someone else) whether meetings feel evaluative; offer a different mentor or time; restate the targets in the mentee's own terms.
A programme shows good retention but no change in practice or achievement The relationship works as support and belonging rather than as instruction; the practice measure is not aligned with what was coached; selection (those who stayed would have stayed anyway) Compare those assigned with those not assigned, if assignment was by lottery or rule; check whether the practice measure covers what meetings worked on.
Mentoring helps some learners and not others Mentor quality differs; contact time differs (some never observed); learners differ in what they wanted from it Tabulate outcomes by mentor and by contact actually received; ask learners what they hoped for at the start.

Evidence and qualifications

Teachers coached on their practice: one meta-analysis of causal studies, plus later trials. Kraft, Blazar and Hogan (2018) pooled 60 experimental and quasi-experimental studies of teacher coaching (pre-kindergarten to grade 12, mostly by a dedicated coach). All its findings below are one source; only the pooled instruction estimate is marked S, within the cap its recorded evidence allows (a q4 synthesis, restated in a second report), and the rest are marked M at most.

New teachers: one randomised study against correlational surveys. Comprehensive teacher induction generated no significant improvements in novice teachers' classroom practices in the first year [-M]: Mathematica's randomised study of two comprehensive induction programmes (intensive mentoring with professional development) found no first-year improvement in practice, retention or achievement; the claim records two reports of the same study, so it counts as one. Against it, Strong mentoring is linked to novice elementary teachers being 25 percentage points more likely to plan to remain in the same school, with a similar but non-significant pattern for high school teachers [+W] and Combining intensive mentoring with other supports is associated with the largest retention-plan gaps: 48 percentage points for elementary and 55 for high school novices [~W] are one Chicago survey of plans, reported in a press release; teachers who got strong mentoring may differ in other ways. The review and policy statements (Mentoring strongly influences new teachers' decisions to stay in or leave teaching (review attribution) [+W], New teachers who receive little mentoring are twice as likely to leave the classroom as well-mentored beginners [+W], High-quality induction and mentoring programs are reported to reduce new-teacher attrition, and early-career teachers without coaching are reported twice as likely to leave [+W], Well-prepared and well-mentored teachers stay in teaching at more than twice the rate of unprepared entrants, and teacher residency programs typically retain more than 80% over five years [+W]) report earlier correlational research second-hand. In Missouri, No statistically significant mentoring–retention relationships emerged in districts with any characteristics examined, and subgroup differences exceeding 10 percentage points showed no clear pattern [~W] (small subgroups). Together: the wiki holds a correlation between mentoring and staying, and one randomised test that did not find a first-year effect. The randomised study's later years are not recorded here.

College students: one randomised and one matched evaluation of success coaching. InsideTrack© Coaching has potentially positive effects on college credit accumulation and persistence, based on one RCT meeting WWC standards [+M] (retention 66.4% against 61.4%, effect size 0.13, 3,527 students in seven lotteries) and InsideTrack© Coaching shows no discernible effects on college degree completion (attainment) [~M]; In the Linkow et al. (2019) scale-up QED, coaching produced a statistically significant positive effect on second-year college enrollment [+M] (effect size 0.17) and In the Linkow et al. (2019) scale-up QED, coaching showed no statistically significant effect on cumulative college GPA (effect size 0.01, p = 0.82) [-M]. Both programmes are proactive, assigned coaching; neither tests a peer or faculty mentor. Career coaching in one workforce programme (Career coaching participants showed increased employment rates and earnings after participating in coaching activities [+W]) is a before-and-after outcome study with no comparison group.

Software coaches: two small studies. L2C coaching significantly reduces learners' lap times and failure counts from pre- to post-coaching in FPV drone racing [+M] (dz = −1.08 and −2.13, within the coached group), Baseline coaches produced weaker or absent learning gains: RBF improved only failure count, MIA neither outcome [-W] and Learners rated the L2C coach higher than both baselines on safety, performance, agency, and satisfaction [~W]: 33 adults learning a motor skill; a preprint. A scaffolded AI Study Coach produced no assignment-performance or concept-inventory advantage over unrestricted AI use in a master's programming course pilot [-M]: 33 students, no inferential tests. In a pilot with a biology simulation, groups coached by a teacher or a researcher did no better than groups using the software alone (No significant differences in learning measures across teacher-coached, researcher-coached, and BioWorld-alone conditions [-W]; six groups), although both coaches used a modelling-then-fading approach (Teacher and researcher coaches both used a cognitive apprenticeship approach but differed in scaffolding style [~W]).

Mechanisms carried from neighbouring evidence. Contingent scaffolding improves learning more than fixed or absent support. [+M] and Fading support promotes the transfer of responsibility from instructor to learner. [+M] test support within tutored tasks, not mentoring relationships; the earlier page marked fading [+S], lowered here because it is carried to a different object. A training-design argument, not tested by any study recorded here, holds that practising whole complex tasks improves transfer of complex skills [~W] is an argument, not a finding, and is marked down from [~M] for that reason.

What mentoring covers and how much people get: descriptive. Time spent on specific mentoring activities varied widely, with some teachers never observed by mentors and others observed an hour or more a month [~W], The mentoring topics first-year teachers most frequently received guidance on were noninstructional rather than instructional [~W] and Mentoring experiences did not differ statistically significantly across districts with different poverty, achievement, or rurality levels [~W] are one Missouri survey of first-year teachers. Mentorship was important to apprentices, but mentor quality and time with mentors varied across programs [~W] and Mentors of color were especially valuable to apprentices of color, who wanted more intentional mentor matching [~W] are interviews in one teacher-apprenticeship programme. Colleagues, not coaches, were teachers' main source of technology support, and over a third of post-survey respondents reported no support [-W] and Teachers described an ideal technology coach as prepared, patient, observant, and using current research-based methods [+W] are one territory's survey and interviews.

The learners, settings and outcomes differ across these (teachers' observed practice, students' test scores, college persistence, drone lap times), so do not rank them by effect size. Effects on observed practice and on persistence are recorded; effects on the learner's own sense of belonging or identity, which the earlier page named as goals, are recorded only as mentees' accounts.

The evidence by question

Further evidence, not yet read against this model

Claims the earlier page cited that this model does not use, each with a note on what it is. They are kept so every earlier citation stays live.

Objective and learner-valued goal

The designer's objective is usually a change in the learner's practice (observed teaching, a skill performed unaided) or continued participation (staying in teaching, re-enrolling), at a stated horizon. Elicit separately what the mentee values: to survive the first year, to be seen as competent, to get a job or a promotion, to belong among colleagues, to have someone to ask. Agreement and divergence are both observations. A mentee who values being seen as competent may hide difficulties from a mentor who also evaluates them; one who values having someone to ask may want reassurance when the design offers observation and feedback.

What would revise this model?

The expectation should weaken if randomised comparisons of mentoring or coaching anchored in observed work show no advantage over unobserved advice or no relationship; if the teacher-induction null effect holds in later years and in other programmes; if student-coaching persistence gains disappear at scale as teacher-coaching effects shrink; or if software coaches match human ones on unaided performance, in which case the relationship is a delivery choice rather than the active ingredient. If role separation, matching or mentor preparation turn out to make no difference when tested, the default's steps 5–7 are costs without benefits. Do not protect the model by calling every null result "not real mentoring" after the fact.

Observed practice, the coached person's students' achievement, persistence, completion, earnings and the mentee's own valued goal are separate claims. The present evidence establishes no contact frequency, session length, mentor-to-mentee ratio or matching rule.

Claims

Examples

Key Sources

  • Kraft, M. A., Blazar, D., & Hogan, D. (2018). The effect of teacher coaching on instruction and achievement: A meta-analysis of the causal evidence. Review of Educational Research, 88(4), 547-588. doi:10.3102/0034654318759268
  • Ladyshewsky, R. K. (2006). Peer coaching: A constructivist methodology for enhancing critical thinking in postgraduate business education. Higher Education Research & Development, 25(1), 67-84. https://doi.org/10.1080/13600800500453196
  • Topping, K. J. (1996). The effectiveness of peer tutoring in further and higher education: A typology and review of the literature. Higher Education, 32(3), 321-345. https://doi.org/10.1007/BF00138870
  • Colver, M., & Fry, T. (2016). Evidence to support peer tutoring programs at the undergraduate level. Journal of College Reading and Learning, 46(1), 16-41. https://doi.org/10.1080/10790195.2015.1075446
Page metadata
Field Value
type principle
id mentoring-and-coaching
aliases mentoringcoaching
title Mentoring and Coaching
description A learner doing real work in a role or programme new to them, paired over time with a more experienced person who observes that work, responds with modelling and specific feedback, meets regularly, stays out of judging them and hands judgement over, is expected to improve observed practice and persistence; the strongest test in the wiki is a meta-analysis of causal teacher-coaching studies (practice 0.49 SD, students' test scores 0.18 SD, effects shrinking at scale), one randomised student-coaching evaluation found higher persistence but not completion, and one randomised induction study found no first-year gains.
status review
canonical True
generated by claude/unspecified on 2026-10-19
trust unverified — no one has confirmed this page's content
sources 4 — kraft-2018, ladyshewsky-2006, topping-1996, colver-2016

Frontmatter is the page's machine-readable record — see the Schema & Guide.