← All domains, the protocol, and the register’s through-lines
What the Evidence Actually Says About Tutoring
Private tutoring in Australia is close to mainstream. A 2026 peer-reviewed study of Sydney public primary schools estimates around 17% of Australian students receive private tutoring at some point in their schooling, rising to 25% in Sydney and as high as 59% in schools with gifted or selective-entry classes [TUT-08]. A separate 2025 University of Sydney and University of Wollongong study puts it higher still, describing tutoring as mainstream for at least one in seven Australian students and estimating roughly 80,000 private tutors working nationally in 2024 [LRX-01]. The academic literature has a name for this: 'shadow education' -- fee-based supplementary instruction outside school hours that mimics and expands alongside the mainstream curriculum [LRX-02]. What is striking, given how common the practice is, is how thin the evidence trail becomes the moment you ask what it costs the country. The Australian Tutoring Association's CEO has put the industry's value at around $1.3 billion, but this figure carries no published survey or methodology behind it -- a media estimate from an interested industry body, not measured data [TUT-09]. The same 2025 study that logged 80,000 tutors also found the sector carries no regulation, no accreditation requirement, and no enforced child-safety screening standard [LRX-01]. So the honest starting point is this: a great many Australian families are buying a service whose total size nobody can verify, in a market nobody is required to police.
Set the market question aside and ask the narrower, more answerable one: does tutoring work? Here the evidence is genuinely reassuring, within limits. The most rigorous answer comes from a 2024 peer-reviewed meta-analysis in the American Educational Research Journal, pooling randomised controlled trials of PreK-12 tutoring programs and finding an overall effect of 0.288 standard deviations on learning outcomes [TUT-02]. An earlier, non-peer-reviewed working-paper version of the same team's analysis had reported a similar 0.37 SD [TUT-01], but the published, peer-reviewed figure is the one worth citing. The UK's Education Endowment Foundation, drawing on 123 qualifying studies, frames the same finding in classroom terms: one-to-one tuition delivers roughly five additional months' progress on average, with the security of that evidence rated 'moderate' rather than definitive [TUT-03]. Evidence for Learning, the Australian mirror of that toolkit, confirms the five-month figure but adds a local caveat: primary-school gains (six months) exceed secondary gains (four months), and literacy gains (six months) exceed maths gains (two months) -- and it states plainly that little Australian-specific research isolates the one-to-one element's effect from everything else around it [TUT-04]. The headline number holds up under scrutiny, but almost none of the underlying trials were run here.
The figure most people have actually heard of is not 0.29 SD or 'five months' -- it is Bloom's famous claim that one-to-one tutoring produces a two-standard-deviation advantage, pushing the average tutored student above 98% of a conventionally taught class [TUT-05]. It is worth being honest about what that claim is and is not. It comes from two 1984 doctoral studies at the University of Chicago, using narrow, unfamiliar-topic tests, and combining tutoring with intensive formative testing, correctives and retesting -- not tutoring in isolation [TUT-05]. A 2020 meta-analysis of 96 tutoring RCTs found that not a single study reached a two-sigma effect; typical effects ranged from about 0.27 SD on broad standardised tests up to 0.84 SD on narrow, tutoring-aligned tests, depending heavily on how the outcome was measured [TUT-06]. A separate systematic review traces the gap further, citing an estimate that modern one-to-one tutoring produces an average effect around d=0.79 -- large, but well under half of Bloom's original figure -- and attributes much of the remaining difference to Bloom's tutees being held to a stricter 90% mastery standard than the 80% used in the comparison, a difference in the standard being taught to, not proof tutoring itself is twice as powerful as the RCT literature suggests [TUT-07]. The honest version of the two-sigma story is that it describes something real but narrow and unreplicated at scale -- a landmark finding worth knowing, not a promise to make.
If the two-sigma figure overstates what one-to-one tutoring alone can do, recent trials show what actually moves the needle -- and dosage and structure look like bigger levers than 'having a tutor' as a blanket category. A 2025 RCT of a high-dosage, one-on-one early-literacy program embedded in the school day found kindergarten students receiving it were more than twice as likely to reach the target reading level by year's end (70% versus 32% for controls), with a 0.23 SD gain on oral reading fluency [TUT-10]; a year-two follow-up on the same cohort found a 35% of a standard deviation gain in reading fluency and a 16% reduction in being classified at-risk [TUT-11]. A comparison of tutoring formats found one-on-one virtual literacy tutoring was nearly twice as effective as two-on-one tutoring, with two-on-one tutors spending 63% more of their time managing behaviour and transitions rather than teaching [TUT-12] -- a direct account of why group size erodes the benefit. A 2024 RCT of high-dosage maths tutoring, run at a 4:1 ratio with 2,065 disadvantaged US ninth-graders, found a smaller but still significant 0.19 SD gain [TUT-13] -- evidence that even at a diluted ratio, structured, high-frequency tutoring moves outcomes, just less than true one-to-one does.
What ties the strongest results together is not raw hours but feedback quality. A large-scale replication of Hattie's Visible Learning research, spanning 435 studies and over 61,000 participants, found feedback carries a medium overall effect on learning (d=0.48) -- but high-information, elaborated feedback that explains an error and how to correct it (d=0.99) was roughly four times more effective than simple reinforcement or praise/punishment feedback (d=0.24) [TUT-14]. The same distinction -- feedback content, not feedback presence, as the active ingredient -- reappears in a domain-specific meta-analysis of corrective feedback in second-language learning, which found a durable medium effect that held up over time, more so for implicit than explicit correction [TUT-15]. This is not incidental to tutoring; it is close to the whole explanation for why one-to-one instruction outperforms a classroom: a meta-analysis across performance domains found the sheer amount of practice a person accumulates explains only a small share of variance in educational achievement specifically -- far less than in music or sport -- meaning practice quantity, or tutoring hours, is not by itself sufficient to explain learning gains [PRC-09]. Retrieval practice and self-testing are separately rated a high-utility learning technique across ages and contexts [PRC-01], consistent with why the better tutoring RCTs pair instruction with active recall and correction rather than passive re-explanation.
Put together, the evidence supports a case for tutoring that is real but bounded. The pooled RCT effect -- around 0.29 SD, or roughly five months' additional progress -- is a defensible, moderate-confidence claim, not a guarantee, and almost none of the underlying trials were conducted in Australia [TUT-02] [TUT-04]. Bloom's two-sigma figure is the wrong number to promise a parent; it was a narrow finding from a small, high-intensity, mastery-driven study design, not a general property of hiring a tutor [TUT-05] [TUT-06]. What the dosage and format trials show is that structure beats hours: true one-to-one beats small-group by a wide margin because group tutoring loses time to behaviour management rather than instruction [TUT-12], and the sessions that work hardest are built around high-information corrective feedback rather than simple encouragement [TUT-14]. For a family choosing tutoring in a market that is large, growing, and still largely unregulated [LRX-01], the evidence points less towards 'more sessions' and more towards a specific question worth asking of any tutor or program: is the feedback explaining the error and the fix, or just marking the page. That is the mechanism the RCTs keep finding, and it is the one lever a family can actually assess before they buy.
A 2025 peer-reviewed University of Sydney / University of Wollongong study (Zunica, Reid O'Connor & Cartwright, published in the Australian Educational Researcher) finds private tutoring is now mainstream for at least one in seven Australian students, with an estimated 80,000 private tutors working in Australia in 2024, but the sector lacks regulation, accreditation, or child-safety screening standards.
revealed that tutoring is now mainstream for at least one in seven Australian students, but lacks effective regulations to ensure safety, quality, or fairness. "Tutoring has grown to become a billion-dollar industry in Australia, but it lacks accreditation, enforced codes of conduct, or real protection for students," said Dr Zunica... Tutoring has also grown steadily in Australia in response to increased competition among students, with as many as 80,000 private tutors working in 2024, many without formal teaching credentials.
University of Sydney news item reporting on Zunica, B., Reid O'Connor, B., & Cartwright, K. · Unregulated and unfair: How private tutoring is falling short for families · The University of Sydney (news); underlying study in Australian Educational Researcher (Springer) · 2025
Peer-reviewed / causalVerified 2026-08-24#LRX-01
Mark Bray's foundational 'shadow education' definition: private supplementary tutoring in academic subjects, provided for a fee outside standard school hours, which 'mimics the mainstream' -- as the mainstream curriculum changes or expands, so does the shadow.
Although the term shadow education is used widely, it is not always employed with consistent meaning. It is therefore necessary at the outset to identify the parameters of the present study. It is concerned with tutoring in academic subjects that is provided for a fee and that takes place outside standard school hours... The study uses the metaphor of the shadow because much private supplementary tutoring mimics the mainstream... Thus, as the curriculum of the mainstream changes, so does the curriculum in the shadow. And as the mainstream expands, so does the shadow.
Mark Bray and Chad Lykins · Shadow Education: Private Supplementary Tutoring and Its Implications for Policy Makers in Asia · Asian Development Bank / Comparative Education Research Centre (CERC Monograph Series No. 9) · 2012
Peer-reviewed / causalVerified 2026-08-24#LRX-02
A meta-analysis of tutoring RCTs (2020 NBER working paper version) found a pooled effect size of 0.37 SD on learning outcomes, with effects concentrated in in-school and teacher/paraprofessional-led programs.
with an overall pooled effect size estimate of 0.37 SD
Andre Nickow, Philip Oreopoulos, Vincent Quan · The Impressive Effects of Tutoring on PreK-12 Learning: A Systematic Review and Meta-Analysis of the Experimental Evidence (NBER Working Paper 27476) · National Bureau of Economic Research · 2020
Peer-reviewed / causalVerified 2026-08-23#TUT-01
The peer-reviewed, published version of the Nickow/Oreopoulos/Quan tutoring meta-analysis reports a pooled effect size of 0.288 SD across PreK-12 randomized tutoring experiments, with larger effects for teacher/paraprofessional tutors, earlier grades, at-least-3x/week sessions, and in-school delivery.
tutoring programs yield consistently substantial positive impacts on learning, with an overall pooled ES of 0.288 SD (SE = 0.029, p < 0.001)
Andre Nickow, Philip Oreopoulos, Vincent Quan · The Promise of Tutoring for PreK–12 Learning: A Systematic Review and Meta-Analysis of the Experimental Evidence · American Educational Research Journal (SAGE), Vol. 61, Issue 1, pp. 74-107 · 2024
Peer-reviewed / causalVerified 2026-08-23#TUT-02
Evidence for Learning (the Australian mirror/localisation of the EEF Toolkit) confirms one-to-one tuition delivers ~5 months' additional progress on average, but notes there is little Australian-specific research isolating the one-to-one element's effectiveness, with primary-school gains (+6 months) exceeding secondary (+4 months), and literacy gains (+6 months) exceeding maths (+2 months).
Despite the popularity and success of one to one programs, there is little research that specifically examines the effectiveness of the one to one element of these programs in Australia.
Evidence for Learning · One to one tuition — Teaching & Learning Toolkit (Australian edition) · Evidence for Learning (Australia) · 2024
Peer-reviewed / causalVerified 2026-08-23#TUT-04
A 2026 Australian peer-reviewed study estimates that approximately 17% of Australian students receive private tutoring at some point during their schooling nationally, rising to 25% in Sydney, and ranging from 23% to 59% across six surveyed Sydney primary schools (highest in schools with gifted/selective classes), based on a parent survey of 590 respondents and interviews with 29 teachers across three low-SES (Western Sydney) and three high-SES (Northern Sydney) public primary schools.
An estimated 17% of students receive private tutoring at some point during their schooling in Australia, with the figure rising to 25% in Sydney... the proportion of students receiving private tutoring ranged from 23% to 59%, with the highest percentages in schools with the classes for gifted students
Christina Ho · Undermining Classroom Teaching? Teachers' Perspectives on the Impact of Private Tutoring on School Education in Sydney, Australia · European Journal of Education, Vol. 61, Issue 2 · 2026
Peer-reviewed / causalVerified 2026-08-23#TUT-08
A large meta-analysis of feedback effects on student learning (435 studies, 994 effect sizes, N>61,000), replicating and expanding Hattie's Visible Learning research, found an overall medium effect of feedback on learning (d=0.48), with high-information/elaborated feedback (d=0.99) substantially outperforming simple reinforcement or praise/punishment feedback (d=0.24).
A meta-analysis (435 studies, k = 994, N > 61,000) of empirical research on the effects of feedback on student learning was conducted with the purpose of replicating and expanding the Visible Learning research... Overall results based on a random-effects model indicate a medium effect (d = 0.48) of feedback on student learning.
Benedikt Wisniewski, Klaus Zierer, John Hattie · Visible Learning Meta-Analysis Revisited: Feedback · Frontiers in Psychology · 2020
Peer-reviewed / causalVerified 2026-08-23#TUT-14
The Education Endowment Foundation's Teaching & Learning Toolkit rates one-to-one tuition as delivering approximately five additional months' progress on average, based on 123 qualifying studies, with 'moderate' security of evidence.
Evidence indicates that one to one tuition can be effective, providing approximately five additional months' progress on average... The security of the evidence around one to one tuition is rated as moderate... 123 studies were identified that meet the inclusion criteria for the Toolkit.
Education Endowment Foundation · One to one tuition — Teaching & Learning Toolkit · Education Endowment Foundation (UK) · 2024
Official / governmentVerified 2026-08-23#TUT-03
Bloom's 1984 'Two Sigma Problem' paper reported that students who received one-to-one tutoring (combined with mastery-learning feedback/correctives) scored, on average, higher than 98% of students in conventionally taught classes -- a two-standard-deviation (2-sigma) advantage -- based on two University of Chicago doctoral studies (Anania; Burke).
the average tutored student was above 98% of the students in the control class
Benjamin S. Bloom · The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as One-to-One Tutoring · Educational Researcher, Vol. 13, No. 6, pp. 4-16 · 1984
Official / governmentVerified 2026-08-23#TUT-05
Modern large-N meta-analyses and re-analyses find that Bloom's 2-sigma effect has never been replicated at scale: a 2020 meta-analysis of 96 tutoring RCTs found no study reached a two-sigma effect, with typical tutoring effects ranging roughly 0.27-0.84 SD depending on test type; the original Anania/Burke studies used narrow, unfamiliar-topic tests and combined tutoring with extensive testing/feedback/retesting, which likely inflated the result beyond what tutoring alone produces.
the two-sigma effects obtained in the 1980s by Anania and Burke were real and remarkable, but they were obtained on a narrow, specialized test, and they weren't obtained by tutoring alone
Paul von Hippel · Two-Sigma Tutoring: Separating Science Fiction from Science Fact · Education Next · 2024
Official / governmentVerified 2026-08-23#TUT-06
A 2025 RCT of Chapter One's high-dosage, one-on-one embedded early-literacy tutoring program found that Kindergarten students (n=818, a large Southeastern US school district) receiving the program were more than twice as likely to reach the program's target reading level by year end (70% vs 32% control) and scored 0.23 SD higher on an oral reading fluency assessment than controls.
Students assigned to the program were over two times more likely to reach the program's target reading level by the end of kindergarten (70% vs. 32%) and scored 0.23-standard deviations higher on an oral reading fluency test than the control group.
National Student Support Accelerator (Stanford University) / Chapter One program evaluators · A Scalable Approach to High-Impact Tutoring for Young Readers: Results of a Randomized Controlled Trial · National Student Support Accelerator, Stanford University · 2025
Official / governmentVerified 2026-08-23#TUT-10
In Year 2 of the same Chapter One RCT (students followed into first grade, 2022-23), one-on-one embedded tutoring produced a 35% of a standard deviation increase in oral reading fluency and a 16% reduction in the likelihood of being classified as at-risk on the district literacy exam, relative to controls.
a 35 percent of a standard deviation increase in oral reading fluency... a 16 percent reduction in being classified as at risk on the district literacy exam
National Student Support Accelerator (Stanford University) · Year Two Results: Assessing the Effects of a Scalable Approach to High-Impact Tutoring for Young Readers · National Student Support Accelerator, Stanford University · 2023
Official / governmentVerified 2026-08-23#TUT-11
A 2024/2025 RCT-based comparison (Robinson et al. 2024 established the effectiveness gap; Hsieh et al. 2025 analyzed the mechanism via transcript analysis) of virtual early-literacy tutoring formats found one-on-one tutoring was nearly twice as effective as two-on-one tutoring for improving student learning, in a K-2 sample of 510 (1:1) and 570 (2:1) US students; two-on-one tutors spent 63% more time managing behaviour/transitions than one-on-one tutors, reducing instructional engagement.
one-on-one virtual early literacy tutoring was nearly twice as effective as two-on-one tutoring for improving student learning... Tutors in two-on-one settings also spent 63 percent more time managing behavior and transitions
Robinson et al. (2024) / Hsieh et al. (2025), summarized by National Student Support Accelerator · The Power of Personalized Attention: Comparing Pedagogical Approaches in Small Group and One-on-One Early Literacy Tutoring · National Student Support Accelerator, Stanford University · 2025
Official / governmentVerified 2026-08-23#TUT-12
A 2024 RCT of Saga Education's high-dosage math tutoring (in-person tutor alternating daily with an adaptive ed-tech platform, average 4:1 student-tutor ratio, one full class period daily for the school year) with 2,065 ninth-graders across six high-poverty Chicago/NYC public high schools (91% economically disadvantaged, 57% Hispanic, 24% Black) found a statistically significant 0.19 SD increase in end-of-year math achievement on the district test.
Saga Technology increased end-of-year math achievement on the district test by 0.19 standard deviations
Study underlying NBER Working Paper w32510, summarized by Arnold Ventures · A Randomized Controlled Trial (RCT) of Saga Technology Providing High-Dosage Math Tutoring and Engagement with an Education Technology Platform to Ninth Graders in Public Schools · Arnold Ventures · 2024
Official / governmentVerified 2026-08-23#TUT-13
A meta-analysis of 33 primary studies (22 published, 11 dissertations) on corrective feedback in second-language acquisition found a medium overall effect size for corrective feedback that was maintained over time, with implicit feedback effects better maintained than explicit feedback, and larger effects in foreign-language than second-language classroom contexts.
there was a medium overall effect for corrective feedback and the effect was maintained over time
Shaofeng Li · The Effectiveness of Corrective Feedback in SLA: A Meta-Analysis · Language Learning, Vol. 60, No. 2, pp. 309-365 · 2010
Official / governmentVerified 2026-08-23#TUT-15
An independent systematic review of mastery learning, tutoring, and direct-instruction studies cites VanLehn (2011)'s estimate that modern human one-to-one tutoring achieves an average effect of d≈0.79 -- large, but well below Bloom's original d=2.0 claim, which the review attributes largely to Bloom's tutees being held to a stricter mastery standard (90% vs 80%) rather than tutoring alone. The same review finds broader direct-instruction effects for the general (non-disadvantaged) population are closer to a quarter of a standard deviation.
the effects that do seem to emerge are weaker, perhaps of a quarter of a standard deviation at best
José Luis Ricón · On Bloom's two sigma problem: A systematic review of the effectiveness of mastery learning, tutoring, and direct instruction · Nintil (independent research blog) · 2020
Large-N industryVerified 2026-08-23#TUT-07
The Australian Tutoring Association's CEO has publicly estimated that almost a quarter of Australian families used private tutoring in 2019, and that the industry is conservatively valued at around $1.3 billion.
Almost a quarter of Australian families used private tutoring in 2019... As an industry, it is conservatively valued at around $1.3 billion and growing.
Mohan Dhall (quoted), EducationDaily staff · Tutoring is a huge $1.3 billion industry. How do you choose a good tutor? · EducationDaily · 2024
Expert commentaryVerified 2026-08-23#TUT-09
Cite any finding by its anchor — for example #LRX-01 — and the link resolves to the claim, quote and source above. The verification protocol lives on the register’s main page.
