The Tutor Handbook · Volume 0195 · Series ID THB-0195
The Tutor Handbook: Complete Series Index
A brilliant CV can still produce the wrong tutor
A tuition programme needs another tutor. One candidate has excellent examination grades, a strong degree and years of subject experience. They speak confidently in the interview and give a polished ten-minute demonstration. A second candidate has a less impressive résumé but notices a learner’s misconception quickly, asks a discriminating question, explains the idea without taking over the thinking, and changes approach calmly when the learner does not follow the first explanation.
Which candidate is stronger?
There is no responsible answer from credentials alone. There is also no responsible answer from one demonstration alone.
Tutor selection is an evidence problem. A programme is not choosing the person who looks most teacher-like. It is deciding whether a candidate has enough of the knowledge, judgement, professional boundaries and learning capacity required for the actual tutoring job, and which remaining capabilities can reasonably be developed through induction and coaching.
The Tutor-Selection Evidence Gate asks how a tuition programme selects tutors from observable evidence instead of charisma, prestige, familiarity or a single performance prepared for the interview. It also protects the opposite mistake: constructing an elaborate selection process that tries to prove everything before employment and quietly makes recruitment impractical.
The useful question is not “Is this person impressive?”
It is “What must this tutor already be able and willing to do before training begins, what can the programme teach, and what evidence would justify believing the candidate can do the job safely and learn the rest?”
Quick answer
Start with the work, not the applicant.
Define the small number of capabilities that are genuinely required at entry. Separate them into three categories: capabilities that must already be present, capabilities that can be trained before live tutoring, and capabilities that can develop through supervised practice after appointment.
Then choose evidence that matches each capability. Subject knowledge may need a content task. Instructional judgement may need a short learner case. Explanation may need a live or recorded teaching simulation. Openness to feedback can be tested by giving feedback and asking the candidate to try again. Professional reliability can be explored through structured examples and references. Relationship-building can be observed in a bounded interaction after basic suitability has already been established. Safety and safeguarding requirements must follow the programme’s applicable legal and organisational procedures rather than being improvised inside an educational interview.
Use common prompts and explicit indicators so candidates are compared on evidence rather than interviewer chemistry. Do not claim that one interview predicts future teaching with certainty. Selection is a risk-reduction decision followed by training, observation and coaching.
A good selection system therefore does two things at once: it keeps obvious mismatches away from learners, and it avoids demanding that a candidate arrive as a finished tutor before the programme has done any development work.
Why this is a new Tutor Handbook job
The Tutor Handbook already owns tutor–learner matching. The Tutor–Learner Matching Gate asks how an existing tutor is paired with a particular learner using expertise, accessibility and working fit. The Tutor Reassignment Gate asks when consistency should yield to a better fit. Those are allocation decisions after a tutor is already inside the programme.
The present gate happens earlier.
It asks whether the programme has enough evidence to appoint the person as a tutor at all, and what conditions should surround that appointment.
It is also different from the Training Need Gate. Training Need asks whether a performance problem should be solved by more training or by changing materials, systems or workload. Tutor selection asks which capabilities the programme deliberately chooses to select for because they are too important, too costly or too uncertain to leave entirely to later training.
The Coaching Receipt and Training-to-Live-Practice Transfer Check remain later owners. A candidate who performs well in selection still has to show that trained practice appears in real lessons.
The selection gate is therefore the entrance to a longer evidence chain, not a final judgement on a person.
What current high-impact tutoring guidance actually says
The National Student Support Accelerator’s current Tutor Selection Strategy makes the core sequence explicit: decide what qualities the tutor role requires, decide which qualities the programme will teach, and select for the important remainder. It also recommends observable behavioural indicators so decisions can be traced to evidence from the application process.
Its current Toolkit for Tutoring Programs likewise recommends competency-based selection with observable measures, structured evaluation and multiple methods such as interviews and scenario tasks. The November 2025 Tutoring Quality Standards classify tutor recruitment and selection as research-informed rather than as a perfectly settled causal science. The standards also make an important point for small programmes: many different tutor types can be effective when they receive appropriate preservice training, ongoing coaching and structured materials, and less experienced tutors may need more support.
That evidence boundary matters.
A programme should not turn one preferred biography into a universal definition of a good tutor. A retired teacher, university student, subject specialist, career-switcher or experienced private educator can each be suitable under different programme designs. The selection system should start from the job the programme needs done.
The Accelerator also recommends reference and background checks as part of programme selection systems. Exact legal screening duties vary by jurisdiction and organisation, so this handbook does not invent a Singapore employment or safeguarding checklist. The principle is narrower: educational evidence from an interview never substitutes for the formal safety procedures that the organisation is obliged to follow.
Define entry capability before choosing an interview question
Selection becomes noisy when the programme begins with fashionable interview activities.
“Let’s ask for a demo lesson.”
“Let’s give them a personality test.”
“Let’s ask what their teaching philosophy is.”
These may produce information, but only after the programme knows what decision the information is meant to support.
Suppose the role is small-group Secondary Mathematics tutoring. The programme may decide that entry-level capabilities include sufficient subject knowledge, the ability to detect a wrong first step, the ability to explain without simply doing the question for the learner, professional communication, and willingness to use evidence and feedback.
A polished slide deck is not itself one of those capabilities.
Neither is extroversion.
Neither is having attended a famous school.
The programme can now design a lean evidence route. Subject task for content. Learner-work interpretation for diagnosis. Five-minute explanation for communication. Changed learner response for adaptation. Feedback-and-retry for coachability. Structured questions for professional judgement.
Every selection activity should earn its place by informing a defined capability.
If an interview component cannot change the appointment decision or the support plan, remove it.
Credentials are priors, not receipts
Qualifications and experience matter. They can change what is plausible before direct evidence arrives.
A degree in the subject may make strong content knowledge more likely. Years of teaching may make classroom routines more familiar. Previous tutoring may mean the candidate has already experienced the intimacy and rapid adaptation of small-group work.
But credentials are not the same as live evidence of the current job.
A highly qualified candidate can over-explain, miss learner misconceptions, respond defensively to feedback or struggle to adapt from whole-class teaching to three learners with different needs.
A less experienced candidate can show strong subject thinking, careful listening, clear explanation and unusually good uptake of feedback, while still needing structured materials and coaching.
Selection should therefore use credentials as one evidence stream, not as a replacement for job-relevant samples.
The decision language should stay proportionate.
“Has relevant subject preparation and prior teaching experience” is evidence.
“Therefore will be an excellent tutor” is prediction beyond the evidence.
“Has no formal teaching qualification” is evidence.
“Therefore cannot become a strong tutor” is also prediction beyond the evidence.
A programme that knows what it can train becomes better at reading credentials without worshipping them.
Composite case: the candidate who explains beautifully and diagnoses poorly
This case is fictional.
Daniel is applying to teach Secondary Mathematics. His explanation of completing the square is elegant. He uses clear notation, a useful visual relationship and confident pacing.
Then the selector shows him a learner solution:
x² + 6x + 5 = 0
x² + 6x = -5
x² + 6x + 9 = -5 + 9
(x + 3)² = 4
x + 3 = 4
x = 1
Daniel says, “They understand completing the square but made a careless mistake at the end.”
Another candidate, Maya, pauses.
She says the learner correctly formed the square but treated the square-root step as though √4 had only one value. She would ask one fresh discriminating question before deciding whether this is a local omission or a broader weakness with solving equations containing squares.
The selector has learned something important.
Daniel may be a strong explainer. The sample does not yet show strong diagnosis.
This does not mean Daniel should be rejected automatically. The programme must decide whether diagnostic interpretation is an entry requirement or a capability it can build under supervision.
If the programme’s work is mainly Class 0 Homework Helper and Class 1 Explainer support with tightly structured materials, Daniel may be viable with training.
If the role immediately requires Class 3 Diagnostic Tutor judgement on consequential learner routes, the gap has greater weight.
Selection is meaningful only when the evidence is interpreted against the actual function.
Do not let the demo lesson become theatre
Demo lessons are attractive because they look like teaching.
They can also be misleading.
A candidate can rehearse one favourite explanation, use a topic they know deeply and perform confidently for adults pretending to be students. The setting may reward presentation more than observation. There may be no authentic misconception, no need to change course and no evidence of what the candidate does when their explanation fails.
A demo becomes stronger when it contains uncertainty.
Give the candidate a small target, a realistic learner artefact or live simulated response, and enough room to make a decision.
Then change something.
The learner gives an unexpected wrong answer.
The learner can imitate the procedure but cannot explain why.
One of three learners finishes immediately while another is lost.
The first explanation does not land.
Now watch what the candidate does.
Do they notice?
Do they ask a useful question?
Do they repeat the same explanation louder?
Do they take over the task?
Do they preserve the learner’s thinking?
Do they invent a diagnosis too quickly?
The point is not to trick the candidate. It is to sample adaptation.
Real tutoring is not a speech. A selection task should not reward speech alone.
A feedback-and-retry is more informative than a feedback question
Many interviews ask, “How do you respond to feedback?”
Almost every candidate knows the desirable answer.
“I welcome feedback.”
“I am always learning.”
“I reflect on my practice.”
Those statements may be sincere. They are still self-report.
A stronger selection move is to give one precise piece of feedback and let the candidate try the teaching move again.
For example: “Your explanation was accurate, but you did most of the reasoning. On the second attempt, keep the same target and give the learner more of the decision.”
Now observe the retry.
Does the candidate understand the feedback?
Can they translate it into behaviour?
Do they make a small adjustment or completely abandon what was working?
Do they ask a clarifying question?
Do they become defensive?
One retry cannot prove long-term coachability. It does produce better evidence than asking whether the candidate considers themselves coachable.
It also respects the programme’s responsibility. If ongoing coaching is central to the tutor model, selection should sample the candidate’s ability to participate in coaching.
Subject knowledge should be sampled at the level the tutor must actually use
A content test can be too easy or too academic.
If it only asks routine questions, it may show that the candidate can execute but not that they can explain, compare methods, detect misconceptions or judge prerequisite structure.
If it becomes a university-level contest, it may screen for advanced mathematics unrelated to the tutoring role.
The content sample should match the real educational responsibility.
For Primary Science, that may include explaining causal mechanisms in age-appropriate language, identifying a misconception in a pupil answer and distinguishing acceptable simplification from scientific inaccuracy.
For English, it may include close reading, sentence-level language control, interpretation of a learner paragraph and the ability to discuss why one revision improves meaning rather than merely sounding nicer.
For Mathematics, it may include solving correctly, identifying the first consequential error, offering an alternative representation and knowing which prerequisite to check.
The candidate need not perform every Class 0–6 tutor function at entry.
The content sample should test enough knowledge to make the intended tutor role safe and learnable.
Relationship evidence should not collapse into “chemistry”
Tutoring is relational. That does not justify hiring by vibe.
A selector may feel that one candidate is warm and natural while another is quieter. Such impressions can be informative, but they are vulnerable to similarity bias, accent bias, cultural expectations and interviewer preference.
Translate relationship quality into observable professional behaviour.
Does the candidate listen to the learner’s full response?
Do they maintain high expectations without humiliation?
Do they respond to uncertainty respectfully?
Can they acknowledge effort without pretending an incorrect answer is correct?
Do they ask about the learner’s reasoning rather than only telling?
Can they establish warmth while keeping professional boundaries?
Do they adapt their language without performing a caricature of the learner’s age or background?
These behaviours are more defensible than “great personality”.
The programme should also avoid treating one short interaction as destiny. Tutor–learner matching remains a later gate because fit emerges in context. Selection only needs evidence that the candidate can build positive professional relationships, not proof that every learner will connect with them.
Structured comparison protects both the candidate and the programme
Unstructured interviews feel natural because conversation flows.
They also make candidates difficult to compare.
Candidate A is asked three diagnostic questions because the interviewer finds the topic interesting. Candidate B spends half the interview discussing previous employment. Candidate C shares a personal story that creates instant rapport. Afterwards, the panel remembers impressions rather than comparable evidence.
A structured process does not require robotic interviewing.
Use a common core.
Ask every candidate a small set of role-relevant questions or tasks. Use the same evidence dimensions. Let follow-up questions clarify rather than create an entirely different test. Record concise evidence before discussing overall impressions.
This helps the programme separate “I liked this person” from “the candidate demonstrated the required capability”.
It also allows later learning.
If new hires repeatedly struggle with one capability, the programme can ask whether selection failed to sample it, training failed to build it, or the job design is unrealistic.
Without a structured record, every hire becomes an anecdote.
The programme should select for fewer things than it wants eventually
An excellent tutor needs many capabilities.
Trying to select all of them before appointment creates an impossible entrance test.
The programme should identify the minimum viable professional starting point.
Some qualities are expensive to train quickly. Baseline subject knowledge, basic professional judgement, safety-mindedness and willingness to learn may deserve strong entry evidence.
Other capabilities can be built.
A new tutor can learn the programme’s questioning routine.
They can learn how the programme records learner evidence.
They can rehearse a three-learner orchestration pattern.
They can learn the exact boundary for AI-generated materials.
They can deepen curriculum familiarity.
They can receive coaching on pacing.
This distinction also improves equity. Selection should not quietly reward applicants who have already had privileged access to the programme’s preferred language and routines when those routines can be taught explicitly.
Ask: “Does this person need to arrive knowing this, or are we using prior exposure as a shortcut because we have not designed training?”
That question often reveals unnecessary barriers.
Composite case: the polished veteran and the adaptive novice
This case is fictional.
A programme interviews two candidates for Primary English.
Mr Lim has taught for twenty years. In the demo, he is authoritative, fluent and efficient. A learner gives a vague comprehension answer. He immediately supplies a stronger version and explains why it is better.
Nadia is a university graduate with much less teaching experience. Her explanation is less polished. When the learner gives the vague answer, she asks, “Which word in the passage makes you think that?” The learner points to an irrelevant sentence. Nadia changes course and works on evidence selection before answer phrasing.
The selectors should not turn this into “experience versus youth”.
Both candidates have evidence and uncertainty.
Mr Lim may possess deep subject knowledge and classroom management. His sample raises a question about whether his help replaces too much learner thinking.
Nadia shows promising diagnostic responsiveness. Her sample raises questions about depth of curriculum knowledge, pacing and whether the good move survives outside a prepared scenario.
A sensible selection process gathers one more discriminating sample rather than making the story larger than the evidence.
Mr Lim receives feedback to preserve the learner’s first reasoning and retries.
Nadia receives a more demanding text and must explain the curriculum expectation.
Now the programme learns more.
Selection should create the next useful question, not rush toward a character verdict.
References answer different questions from demonstrations
A reference cannot watch the candidate teach your programme’s curriculum.
A demo cannot tell you whether the candidate reliably arrives prepared over months.
Different evidence sources have different jobs.
A reference may help with patterns of reliability, collaboration, response to feedback and professional conduct in a previous setting. It may also be incomplete or shaped by the referee’s context.
A structured interview can test reasoning and values.
A content task can sample subject knowledge.
A demo can sample enactment.
A feedback retry can sample immediate adaptation.
Formal background and safeguarding checks address organisational safety requirements.
The selection decision becomes stronger when the sources complement one another rather than pretending one source is comprehensive.
This is the same logic that protects learner assessment from one overextended test.
Use the smallest set of evidence sources that covers the consequential risks.
Do not collect extra personal data just because it might be interesting
Tutor selection creates records about adults.
The Record Minimum principle still applies.
Collect information that serves legitimate recruitment, safety, legal and programme decisions. Avoid irrelevant personal probing. Do not store every interview note indefinitely simply because storage is cheap. Do not infer clinical or personality diagnoses from casual behaviour.
The selection record needs enough information to support the decision and later accountability: role, entry criteria, evidence observed, support conditions and the appointment outcome.
The exact employment and retention rules belong to the organisation’s applicable policies and law, not to this handbook.
The educational principle is data minimisation with decision usefulness.
Bias can enter through “standards” as well as through intuition
A structured process is not automatically fair.
If the indicators are poorly chosen, structure can formalise bias.
Suppose “professional communication” is defined as sounding like the existing leadership team. Suppose “confidence” is rewarded even when it becomes premature diagnosis. Suppose candidates are penalised for pausing to think. Suppose only one cultural style of warmth is recognised.
The programme must inspect whether each criterion belongs to the work.
Observable is not enough. The observation must be job-relevant.
This is why NSSA pairs structured indicators with attention to anti-bias and local context.
A useful review question is: “If two people behaved differently here, which difference would actually change the learner’s educational opportunity?”
If the answer is unclear, the criterion may be preference masquerading as quality.
Use rejection reasons that point to evidence, not identity
A programme may decide not to appoint a candidate.
Internally, the reason should be tied to the job.
“Content sample contained consequential inaccuracies.”
“Could not maintain professional boundaries in the scenario.”
“Repeatedly replaced learner thinking after feedback-and-retry.”
“Required availability does not match the role.”
“Did not complete required formal screening.”
These are different from global labels such as “not tutor material”, “not a good fit” or “not our type”.
Evidence-based reasons improve future selection. They also reduce the temptation to turn one bounded performance into a permanent judgement about the person.
A candidate can be unsuitable for one role and suitable for another.
A strong Class 1 Explainer may not yet be ready for an autonomous Class 4 Route Designer function.
The programme should appoint to the job it actually has.
Probation and supervised entry are part of honest selection
No pre-employment process can reproduce normal tutoring perfectly.
The learner is real. The term becomes busy. The tutor sees unfamiliar errors. Parents ask questions. Materials change. Fatigue appears.
A programme therefore needs a supervised entry period appropriate to the role.
This is not an excuse for weak selection. It is recognition that live-practice evidence has higher ecological validity than interview evidence.
The tutor can begin with structured materials, bounded groups and coaching. The programme observes whether subject knowledge, relationship quality, evidence use and professional routines survive ordinary work.
If the candidate needed support in one area, that support can be explicit.
The selection decision becomes: “sufficient evidence to enter supervised practice”, not “proven excellent forever”.
That language is more accurate and more humane.
When should a programme stop the process early?
Some evidence should end the selection process.
A consequential subject-knowledge failure may make the person unsafe for the proposed role without substantial remediation.
A professional-boundary breach in a scenario may create unacceptable risk.
Failure to satisfy required screening cannot be repaired by a strong demo lesson.
Dishonesty about qualifications or experience changes the trust problem.
But not every weak moment deserves elimination.
A candidate who needs a clearer explanation prompt can retry.
A candidate unfamiliar with the programme’s specific notation can learn it if their underlying knowledge is sound.
A candidate who is nervous in the first two minutes can still demonstrate careful teaching.
The programme should distinguish disqualifying evidence from developable weakness before the interview begins.
Otherwise selectors tend to make the distinction after they know which candidate they like.
A small-centre selection protocol
A small tuition centre does not need a human-resources laboratory.
A defensible process can be compact.
First, define the role and its entry capabilities.
Second, screen basic eligibility and complete the organisation’s required safety processes.
Third, use one structured content-and-judgement task.
Fourth, use a short teaching or learner-response simulation.
Fifth, give one piece of feedback and observe a retry.
Sixth, use structured reference evidence where appropriate.
Seventh, make the appointment decision with an explicit support plan for anything that remains developable.
Finally, verify in live supervised practice.
The process should be shorter for tightly scripted entry roles and deeper for tutors expected to diagnose, redesign routes or work with high ambiguity from day one.
Complexity should follow decision risk.
Failure modes
The credential halo. Qualifications are treated as sufficient proof of tutoring skill.
The charisma hire. Confidence, humour or conversational ease becomes the main selection criterion.
The demo-theatre trap. A rehearsed mini-lesson is treated as representative normal practice.
The interview philosophy trap. Candidates receive credit for saying good educational ideas without showing them in action.
The over-selection problem. The programme tries to test every future capability before hiring and creates an unnecessary recruitment barrier.
The under-selection problem. Anyone with subject knowledge is appointed and the programme discovers professional judgement problems only after learners are exposed.
The hidden-bias rubric. A structured score gives apparent objectivity to criteria that mostly encode interviewer preference.
The one-source decision. One reference, one test or one demo dominates despite important blind spots.
The training confusion. The programme rejects candidates for not already knowing routines it could teach efficiently.
The permanent-label failure. A bounded selection result becomes a global judgement on the person rather than a decision about one role under current evidence.
The finished-tutor fantasy. Appointment is treated as the end of professional development rather than the beginning of live-practice verification.
What parents should infer from a programme’s tutor-selection claim
Parents are often told that tutors are “carefully selected”, “highly qualified” or “experienced”.
Those phrases are not meaningless, but they are incomplete.
A stronger programme should be able to explain the job-relevant selection logic without exposing private personnel information.
What does the programme require before a tutor begins?
How does it check subject knowledge and teaching judgement?
What is trained after appointment?
How are new tutors observed and supported?
What happens if a tutor is strong in one function but not yet ready for another?
The goal is not to publish employee files.
It is to show that tutor quality is treated as a system of selection, preparation, coaching and evidence rather than a marketing adjective.
Evidence boundaries
The National Student Support Accelerator’s Tutor Selection Strategy recommends defining required tutor qualities, distinguishing what will be trained from what must be selected, and tying decisions to observable indicators. Its current Toolkit recommends competency-based selection with structured evaluation and multiple assessment methods. These are research-informed programme-design resources.
The Accelerator’s Tutoring Quality Standards, dated 12 November 2025, classify tutor recruitment and selection as research-informed. The standards explicitly note that a range of tutor types can be effective with sufficient training, coaching and structured materials, and that less experienced tutors may require more support. This does not validate any one interview question, score or hiring threshold.
The examples and decision protocol in this article are professional design proposals for tuition programmes. They should be adapted to role, subject, age group, applicable law and organisational safeguarding procedures.
No selection process can guarantee future performance. The strongest test remains what the tutor actually does with learners over time under appropriate supervision.
The end state
The best tutor-selection system is neither prestige-driven nor suspiciously elaborate.
It knows what the role requires.
It knows what the programme can teach.
It samples the capabilities that matter before learners carry the risk.
It uses more than one relevant source where the consequences justify it.
It gives candidates a fair chance to show adaptation rather than rewarding only rehearsed polish.
It records evidence rather than vibes.
Then it treats appointment as permission to enter a supported professional-learning system, not as certification that the person has finished developing.
That is the Tutor-Selection Evidence Gate.
Do not hire the best performance of “being a tutor”.
Select the person who has enough of the right starting capabilities, can learn the rest, and keeps producing stronger evidence when the work becomes real.
