Tuition Guides

What the Data Says About Tuition in Singapore — A Review of the Research

A review of Singapore household tuition spending and experimental tutoring research: what the figures mean, what studies establish, and how parents can use the evidence in a family decision.

Updated 17 Nov 2025

By DeepThink Teaching Team · Originally published 25 May 2025 · 5 min read

Checked against SingStat HES 2023 analysis, experimental tutoring research and the EEF small-group tuition review

Singapore's spending data and tutoring studies answer different questions. Household surveys tell us what people spend. Experimental studies can test whether a defined tutoring programme helped the students who took part. Neither establishes that a particular child needs tuition or that a commercial programme will deliver a promised grade increase.

For parents deciding whether to enrol, the useful evidence is specific: which population was studied, what teaching they received, what it was compared with and which results were measured. This review separates those questions from claims about the size of the market.

Read household spending figures with their denominator

The Department of Statistics' 2025 analysis of households with young children uses Household Expenditure Survey 2023 data. Its Table 5 reports the following rounded monthly private-tuition expenditure:

Household group in 2023Average private-tuition spending per household per month
All resident householdsS$105
Married couple-based households with children below 16S$389
Within that group, youngest child aged 6 or belowS$235
Within that group, youngest child aged 7–15S$548

These averages include households in each group whether or not they incurred that spending. They are not the bill among tuition users, a per-child price or a recommended family budget. The family rows use the publication's married couple-based definition and should not be generalised to every household structure.

The HES 2023 publication describes expenditure during the survey period. Calling it a 2026 tuition fee would confuse the survey year with the date somebody is reading it.

Spending alone also cannot show how many marks tuition produced. Higher expenditure can reflect a different mix of households, services, hours and prices. A larger bill does not isolate better teaching.

Why a simple tuition-versus-no-tuition comparison is difficult

Families choose tuition for different reasons. One child may be rebuilding an earlier topic; another may already be doing well and seeking harder work. Available time, household resources, prior attainment and motivation can also differ between those receiving tuition and those who do not.

If the groups later have different scores, tuition is one possible contributor among several. An observational association does not by itself show what tuition caused. Statistical adjustments can help, but their validity depends on the data and assumptions.

Randomised evaluations address this problem by comparing groups assigned to an offer of support or a comparison condition. Such studies exist in tutoring research. It would be wrong to say the evidence is entirely observational or that randomisation is impossible.

What experimental tutoring studies find

Nickow, Oreopoulos and Quan's review of experimental PreK–12 tutoring studies reports a pooled learning effect of 0.37 standard deviations. The review includes one-to-one and small-group programmes, with different tutors and delivery arrangements. It finds differences by programme characteristics, including who teaches and when support is provided.

That pooled effect is evidence that tutoring interventions can improve measured learning. It is not a Singapore PSLE estimate, a rating of commercial tuition centres or a promise about every student.

A standard deviation describes a score distribution. Converting 0.37 standard deviations into “a few PSLE marks” or an AL-band change would require an appropriate score distribution and a justified link between the study tests and the examination. The review supplies no universal conversion of that kind.

Average effects also conceal variation. A programme may work well overall without helping every participant equally; a useful intervention may differ from the one a parent can actually buy.

What online tutoring adds to the evidence

The Italian Tutoring Online Program evaluation found improved test performance for disadvantaged middle-school students offered online support during school closures. It tested a particular trained-volunteer programme against a comparison group.

This supports the possibility of useful online tutoring. It does not establish that a weekly Singapore online class matches a physical class, and it does not test DeepThink. The target population, school circumstances and amount of support matter when applying the finding elsewhere.

Our PSLE online-tuition evidence guide considers what a family can check directly when making that narrower decision.

What the evidence means for choosing support

The Education Endowment Foundation's small-group tuition review emphasises teaching targeted at pupils' needs, staff preparation and the quality of feedback and interaction. Its findings concern the school interventions reviewed, rather than a guarantee for a particular commercial class.

For a family, that suggests questions worth asking: what difficulty has the teacher identified, how will they explain it, and what later work will show whether the child can now manage it independently? These are practical implications, not a formula that predicts an examination result.

A branded method, a tutor's credentials and a class-size cap can provide information about an offer. They cannot substitute for evidence about how the child responds to the teaching.

Keep workload and wellbeing in the decision

A tutoring study's academic result does not tell you whether an extra class fits your child's current week. Compare the lesson, travel if any, independent practice and time left for sleep and other activities.

Ask your child which difficulty they want help with. Agree on a manageable task and a point to discuss progress. If work is causing persistent distress or affecting daily life, seek appropriate support rather than assuming more tuition is the answer.

This review does not establish a universal maximum number of tuition hours, nor that tuition caused national trends in anxiety. Those claims need their own evidence. A family's timetable and the child's experience still deserve attention even when a national causal estimate is unavailable.

A practical decision from limited evidence

Before enrolling, write down the learning need and what you hope the child will be able to do more independently. Bring a recent attempt to the teacher and ask how it would be addressed. Read the actual fees and terms, and agree what you will revisit after several lessons.

Then use fresh work to assess the plan: a related question, an explanation of a method or a repeated error that is becoming less frequent. School results can add context, but differences between papers make them an imperfect measure of a tuition programme's effect.

We run a tuition programme, so we have a commercial interest in this question. DeepThink's tutors teach live online, supported by homework and extra practice between classes. If you would like to discuss whether that arrangement addresses a particular difficulty, talk to our team.

Found this helpful? Share it with someone who needs it.