A Beijing dad emailed me last March. His daughter scored 95 on her Chinese school's math final, then sat an AEIS-style word problem and stared at the page for eleven minutes. The problem wasn't hard arithmetically — two unknowns, three sentences, total operations needed: maybe four. What broke her was the format. She had been taught to set variables, write equations, solve. Singapore had taught its kids to draw a rectangle. The AEIS Math model drawing method is not a trick. It's a different operating system, and if your child has never used it, no amount of arithmetic speed will save them in the exam hall.
This is the article I wish I had written two years ago. It's long because the topic deserves length. If you're skimming, jump to the "common bar model archetypes" table — that alone is worth the read.
Why Singapore teaches math through bars instead of equations
Most countries' primary math curricula go: numbers → arithmetic → word problems → algebra (around age 12-13). Singapore inserts a layer between word problems and algebra called the model method — a visual representation system using rectangular bars to stand in for quantities, parts, and relationships.
The official term in MOE syllabus documents is the "concrete-pictorial-abstract" approach, but parents and teachers just call it "model drawing" or "bar models." It was developed in the early 1980s when Singapore's math scores were mediocre, and within fifteen years the country was topping international benchmarks like TIMSS. Whether causation or correlation, the method stuck.
Here's the core idea: instead of writing x + y = 80, x − y = 12, solve, a Primary 4 child draws two stacked rectangles, marks the total as 80 and the difference as 12, and sees that the smaller bar is (80 − 12) ÷ 2. The visual makes the algebraic structure obvious to a 10-year-old who hasn't met algebra yet.
For AEIS, this matters in two ways. First, the test is benchmarked against MOE syllabus, which means roughly 30-40% of word problems are designed assuming bar-model thinking. Second, even when a problem can be solved with arithmetic or early algebra, the fastest path is often a bar drawing — and AEIS Math is timed tightly enough that method efficiency becomes a scoring factor.
What the AEIS Math model drawing method actually looks like
Let me walk through one concrete problem so the abstract becomes tangible. This is roughly Primary 4 level, the kind of thing a child sitting AEIS for P5 placement should solve in under three minutes.
Aisha and Bala had $84 altogether. After Aisha gave $6 to Bala, they had the same amount of money. How much did Aisha have at first?
A child trained on equations writes: A + B = 84, A − 6 = B + 6, solve → A = 48. Fine. About two minutes if she's fluent.
A Singapore-trained child draws this:
Aisha: [====|==] ← total bar with a 6-block at the right
Bala: [====] ← shorter bar
← combined = 84
She sees that the difference between Aisha and Bala is $12 (because moving $6 evens them out — $6 from Aisha plus $6 to Bala). Then she sees that the average is $42, so Aisha = $42 + $6 = $48. About forty seconds, no equations written.
That speed gap, multiplied across 40+ problems on the AEIS Math paper, is the difference between finishing comfortably and running out of time. This is why the AEIS Math model drawing method isn't optional enrichment — for many overseas children it is the single highest-leverage skill to build before the test.
The seven bar model archetypes your child must master
Most Singapore primary math problems collapse into a small number of structural patterns. Once a child recognizes the pattern, the drawing follows automatically. I drilled these into my own children using flashcards — pattern on the front, drawing on the back.
| Archetype | Typical phrasing cue | When it appears |
|---|---|---|
| Part-whole | "altogether", "in total" | P2-P6 |
| Comparison | "more than", "less than" | P3-P6 |
| Before-and-after | "after giving/spending..." | P4-P6 |
| Equal-units | "twice as many", "3 times" | P3-P6 |
| Constant-difference | "both received the same number more" | P5-P6 |
| Constant-total | "exchange", "transfer" | P5-P6 |
| Internal transfer with fractions | "1/3 of A equals 2/5 of B" | P6 |
A child preparing for AEIS at P4 level needs the first four cold. P5 entry needs all six. P6-into-Secondary 1 entry needs all seven plus comfort moving between bar models and early algebra.
The mistake overseas tutors make is teaching these as "topics." They aren't topics. They're patterns of how quantities relate, and the right teaching sequence is to mix them from week one so the child practices recognizing which pattern fits before drawing.
Where overseas children actually struggle with AEIS bar model word problems
I've now reviewed practice scripts from somewhere north of 200 families preparing for AEIS over the past eighteen months. The error patterns are remarkably consistent, regardless of whether the child came from China, Indonesia, India, or the UK.
Pattern 1: Drawing the bars to scale instead of structurally. A child reads "John has $80 and Mary has $20" and tries to draw a bar four times longer. Then a problem comes along where the ratio isn't given numerically and she's paralyzed. The fix: bars represent structure, not measured length. Train this with problems where ratios are unknown.
Pattern 2: Starting to draw before finishing reading. Common in kids from exam cultures where speed is rewarded. They sketch the first sentence, then have to redraw when sentence three changes the picture. Fix: read the whole problem twice before pencil touches paper.
Pattern 3: Forgetting "before-and-after" needs two diagrams. A transfer or spending problem usually needs a "before" model stacked above an "after" model, with arrows or labels showing what changed. Children who try to cram both into one bar lose accuracy fast.
Pattern 4: Algebra-trained children refusing to draw. This is the hardest one to break. A 12-year-old who can already do simple algebra resists bar models because they "feel babyish." Then she hits a P6 problem with internal-transfer fractions and her two-variable equation system takes nine minutes to set up correctly. The fix is showing her — with a stopwatch — that on Singapore-style problems, the bar model is genuinely faster. Once she sees the time data, resistance collapses.
Pattern 5: Labeling that doesn't match the question. A child draws the model perfectly, calculates the value of one unit, and then writes that as the answer — when the question asked for the difference or the total after. Always re-read the question line right before writing the final answer. I have my students literally underline the question word.
A realistic 12-week training plan for the model method
If your child is sitting AEIS in roughly three months and has never seen bar models, here's the pacing I've found works. This assumes 45 minutes a day, six days a week — about 32-35 hours total, which is genuinely the floor for going from zero to AEIS-ready on this skill alone.
| Weeks | Focus | Daily structure |
|---|---|---|
| 1-2 | Part-whole + comparison, simple numbers | 10 min explanation, 25 min guided practice, 10 min review |
| 3-4 | Equal-units (multiplication-style bars) | 5 min recap, 30 min mixed practice, 10 min error log |
| 5-6 | Before-and-after, single transfer | Introduce two-stage diagrams, mixed pattern recognition |
| 7-8 | Constant-difference and constant-total | This is where most kids hit a wall — slow down |
| 9-10 | Fractions and ratios with bars | The hardest pattern; expect a regression in confidence |
| 11 | Mixed timed practice, full word problem sets | Simulate test conditions, 40 problems in 75 minutes |
| 12 | Past-paper-style mock and error analysis | Don't introduce new content; consolidate |
Two non-obvious points. First, the "error log" — a notebook where every wrong problem gets re-drawn correctly with a one-line note on what went wrong — is more important than doing more problems. Second, week 9-10 confidence dips are normal. Don't panic and switch tutors.
Families who are still figuring out the broader exam path before drilling math should read AEIS Complete Guide 2026: Timeline, Test, and Real Costs first to confirm which level their child is even sitting — drilling P5-level bar models for a child who'll actually be tested at P3 wastes weeks. And if you haven't sorted the practical logistics of the move yet, Moving to Singapore with School-Age Kids — A Checklist covers the parallel timeline so academic prep and visa paperwork don't collide in the final month.
How to teach the Singapore model method math at home if you can't find a tutor
About half the families I talk to are still abroad and don't have access to a Singapore-trained tutor. The good news: the model method is teachable by a parent who is willing to learn it themselves first. The bad news: most parents who try without preparing first end up confusing their child by mixing equation-thinking into the explanation.
Here's the protocol I give parents:
-
Spend two evenings learning the method yourself, before teaching anything. Work through 30-40 problems at your child's level. If you can't comfortably solve them with bars, you can't teach them. MOE's website (moe.gov.sg) has the official syllabus document; the textbook series most schools use is "My Pals Are Here" or "Targeting Mathematics" — readily available as PDFs from legitimate Singapore book retailers.
-
Start one level below your child's actual level. A P5-aspiring child should begin with P3 bar models. The drop in difficulty lets her focus on the drawing technique without arithmetic load. Move up when she can solve 8 of 10 problems unaided.
-
Do the problems alongside her, not over her shoulder. Both of you draw. Compare drawings. This removes the power dynamic that makes kids stop attempting when they're unsure.
-
Forbid algebra for the first six weeks, even if she already knows it. Yes, even if it's "faster" for her on a particular problem. The point is to build a different muscle, and that muscle only grows if it's the only available tool.
-
Photograph her drawings, especially the wrong ones. Reviewing them weekly together catches systematic errors faster than any tutor can.
A reasonable budget: about $60-100 for textbooks and workbooks, zero for software. The temptation to spend on apps and online courses is strong, but I've seen kids progress fastest with paper, pencil, and a parent who's done the homework themselves.
When bar models stop being enough — the bridge to algebra
For children sitting AEIS at the Secondary 1 level (entry into Sec 2 or Sec 3), pure bar-model technique is no longer sufficient. The Sec-level math paper introduces algebraic manipulation, simultaneous equations, and problem types where bars become unwieldy.
The right approach for these older students is to start every problem with a bar model — to clarify the relationships — and then transition to algebra for the actual solving. The bar becomes a thinking tool; the equation becomes the executing tool. Children who skip the bar and jump straight to equations make twice as many setup errors. Children who refuse to leave bars and try to solve Sec 2 problems pictorially run out of time.
This bilingualism between bar and algebra is the actual end goal of the Singapore model method math journey. A P6 child who can fluently translate "1/3 of A equals 2/5 of B" into either a bar diagram or the equation 5A = 6B and pick whichever is faster — that's the level Singapore-schooled kids reach by the end of P6, and that's what AEIS Sec-level papers implicitly assume.
Common questions overseas parents ask me
"My child is strong in math already. Does she really need this?"
Yes, with caveats. Strong arithmetic doesn't guarantee strong word problem performance on Singapore-style tests. The format is different enough that I've watched kids ranked top-5 in their previous school score middling on AEIS practice. Three to four weeks of bar-model focus is usually enough for a strong child to close the gap.
"Can we just memorize problem types and solutions?"
No. AEIS specifically writes problems to defeat memorization — they recombine patterns, change which quantity is unknown, and add irrelevant information. The model method works because it teaches structural recognition, not memorized solutions.
"Should we use Chinese-language Singapore math materials, or English?"
English. AEIS is in English. Word problem comprehension is half the challenge, and a child who has only practiced in Chinese will lose 10-20 seconds per problem just parsing the English phrasing. If your child's English is weak, that's a separate (and equally urgent) problem.
"How much improvement can we realistically expect?"
From the families I've tracked: a child who starts at roughly 40-50% on AEIS-style word problem sets, with twelve weeks of focused model-method work, typically reaches 65-80%. That's enough to convert a likely fail into a likely pass. Children who start above 60% can usually reach 85%+ with the same effort.
How model drawing fits into the broader AEIS preparation picture
It's tempting, when you discover the model method matters, to make it the entire focus. Don't. AEIS Math also tests:
- Pure computation (about 25% of marks) — speed and accuracy on multi-digit operations
- Geometry and measurement (about 15-20%)
- Data interpretation (about 10%)
- Word problems where bar models help (about 35-45%)
- Heuristic problems requiring multiple strategies (about 10%)
So the model method addresses a third to a half of the paper, depending on level. The remaining content needs separate attention. And of course AEIS Math is only one of two papers — English is the other, and for many overseas children the English paper is actually the harder hurdle.
For families preparing the full application, the document and timeline side is covered in AEIS Registration Document Checklist 2026: What Overseas Families Actually Need, and a broader walkthrough specifically for non-Singaporean families is in AEIS for Overseas Families: The 2026 Complete Guide. Don't skip those — I've seen families do beautiful math preparation and then miss the test because a document expired.
A note on what good practice materials look like
Three signals that a workbook or practice set is worth using:
- Problems are mixed across archetypes within each chapter, not segregated. Real exams mix; practice should too.
- Solutions show the bar drawing, not just the arithmetic. If the answer key just shows "84 − 12 = 72; 72 ÷ 2 = 36" without a diagram, it's teaching arithmetic, not the method.
- Difficulty scales within problem types, not just across them. A good workbook has easy comparison problems early and hard comparison problems later, rather than dumping all comparison problems at one difficulty.
Avoid materials that promise "AEIS Math in 30 days" or that brand themselves as exam-cracking shortcuts. The model method is a thinking framework; you can't compress it without hollowing it out.
What to do this week
- If your child has never drawn a bar model, sit with her tonight, find any Primary 3 word problem online, and draw one together. Don't lecture — just draw.
- Order or download two workbooks: one at her current level, one a level below. Plan to start with the lower one.
- Block out 45 minutes a day in the family calendar for the next twelve weeks. If you can't protect that time, scale your AEIS expectations accordingly — there is no shortcut on this skill.
- If your child is sitting AEIS this September or October, start the bar-model work no later than this month. Beyond twelve weeks of runway is comfortable; under eight is tight.