Why your child understands math but fails the test

Your daughter worked through her fraction homework last Tuesday. You watched her divide fractions, reduce them, check her work. She asked one or two questions. You gave her a hint when she stalled. She got nine out of ten right. Thursday the test came back. She got four out of twelve right, and three of the four wrong answers were on the exact same skill she had just practiced. You are staring at the test now, wondering what happened between Tuesday night and Thursday morning.

The short answer

• When a child understands math at home but fails the test, the usual cause is not anxiety or effort. It is that the skill at home was leaning on scaffolding the test removed.
• Three mechanisms explain most cases: borrowed performance (the hint, the formula card, the example disappeared), transfer failure (the practice was one narrow version, the test asked a different one), and format dependence (the skill was tethered to how the problem looked on the worksheet).
• Homework accuracy does not tell you whether a skill is stable. Tests under novel conditions do.
• Read on for how to tell which mechanism is happening to your child, and what to do about each one.

What actually happened between the kitchen table and the test

The usual assumption is test anxiety. Or carelessness. Or that your child didn't study, even though you watched them study. Those explanations feel true because the gap is so abrupt. The skill was there Tuesday. It was gone Thursday. Something must have changed in the child.

What changed was not the child. What changed was the conditions. The homework had scaffolding the test did not. The practice problems were narrow. The test asked a slightly different version. The worksheet showed the problem one way. The test showed it another. The skill looked stable because the support structure was invisible. When the test removed the supports, the skill collapsed.

This is not a character flaw. It is not laziness. It is not a sign your child cannot do math. It is a stability problem. The performance at home was real. It just was not entirely your child's. Three mechanisms explain most homework-test gaps: borrowed performance, transfer failure, and format dependence. All three are fixable. But first you have to recognize which one you are looking at.

The skill was borrowed (and the test removed the loan)

Borrowed performance is what happens when the child leans on something during practice and that thing disappears during the test. The hint you gave when they stalled on the first fraction problem. The formula card taped to the edge of their desk. The worked example at the top of the worksheet. The teacher's just-completed explanation still echoing in their head. The performance was genuine. Your child was thinking. They were working. The answer was correct. The skill just was not entirely theirs.

A concrete example. Your child is working on equivalent fractions. They write 3/4 = 6/? and stall. You lean over and point to the numerator. "What did we multiply the 3 by to get 6?" you ask. They say "two," then write 8 in the denominator and move on. They finish the worksheet. Ten problems, nine correct. You both feel good. Thursday they take the test. No parent at the table. No pointing. The first equivalent-fraction problem reads 2/5 = ?/15. They stall. They guess. They write 5. They move on, already unsure.

Most practice tools and most homework measure whether the answer is correct. They do not measure whether the skill survives when the scaffolding is removed.

The practice was too narrow (and the test asked a different version)

Transfer failure is what happens when the student practices one narrow version of a skill and the test asks a slightly different one. The procedural memory is strong. The conceptual model underneath it is missing or shaky. The student can execute the trained version fluently. The novel version collapses.

A concrete example. Your child can compute "3/4 of 20." They know to multiply 20 by 3, then divide by 4. They get 15. The test asks "what fraction of 20 is 15?" They do not recognize it as the same structure. They guess 3/5. They guess 15/20. They write nothing. The procedural path was one-way: given a fraction and a whole, find the part. The conceptual understanding that would let them reverse the path was never stable. This is one face of what we call false mastery: the right answer in one narrow context, the wrong reasoning underneath.

The report card averaged these two states together. Your child got a B because they were fast and accurate on the fifty trained problems. The test exposed the gap because it asked the six untrained ones. A MAP® score tells you whether your child is on grade level. Whether the skill will transfer to a slightly different problem type two weeks later lives at a different resolution than MAP was designed to measure.

The format shifted (and the skill was tethered to the original version)

Format dependence is what happens when a small change in how the problem is presented breaks the skill. The homework showed division as 24 ÷ 6. The test showed it as 24/6 or "how many 6s are in 24?" The student's retrieval pathway was bound to the first format. The notation shifted. The skill collapsed.

A concrete example. Your child can add fractions when the problem is written with fraction bars: 1/4 + 2/4. They know to keep the denominator and add the numerators. The test writes the problem in words: "one-fourth plus two-fourths." They stall. They write 3/8. They were not reading the fraction bar as a symbol with meaning. They were reading it as a shape with a procedure attached.

The number on the report card averaged across these three very different skill states. That is why the grade did not warn you.

What you can do this week

Five concrete steps. Start with one.

Re-do one homework problem with all scaffolding removed. Pick a problem your child got right two nights ago. Give it to them again, cold, with no hints, no formula sheet, no worked example, no "remember we did this on Tuesday?" Watch what happens. If the skill stands, it was stable. If it shakes, you have found borrowed performance. Either way, you know.

Ask the same problem three slightly different ways. If your child can solve 2x + 3 = 11, ask them to solve 11 = 2x + 3. Then ask them "if 2x + 3 = 11, what is x + 1?" Watch where they stall. The stall point is the transfer gap.

Have your child teach the skill back to you. No notes. No looking at the worksheet. "Explain to me how you add fractions with different denominators." If they can teach it in their own words without defaulting to "you just do this, then you do this," the conceptual model is probably stable. If they cannot explain why the procedure works, the skill is still forming.

Check whether the test format matches the practice format. If your child practiced vertical multiplication and the test was horizontal, give them two practice problems in horizontal format tonight. If they practiced "3/4 of 20" and the test asked "what fraction of 20 is 15?", practice the reverse version. Format gaps close quickly once you name them.

If all three mechanisms are present, the issue is not more practice. If the skill collapses when scaffolding is removed, when the format shifts, and when transfer is required, you are not looking at a volume problem. You are looking at a diagnostic problem. You need to know which skills are stable and which are still shaky under testing conditions. That is what Helix Math was built to do. The free Helix diagnostic takes 30 to 40 minutes and maps the skills behind the number. It shows you exactly which rungs are missing and which ones are shaky, so you know where to start.

If your child is not anxious about the gap, do not create anxiety by hovering. Give them one problem with no help and watch. The skill will either stand or it will shake. Either way, you will know.

The number did not tell you this was coming

If your child had a good MAP score or a strong report card, none of those instruments would have flagged borrowed performance, transfer gaps, or format dependence. They measure point-in-time accuracy under one set of conditions. They do not measure stability. The homework measured accuracy with scaffolding present. The test measured accuracy with scaffolding removed. The gap between those two states is where math gaps form. The work now is not to panic. It is to find which rungs are shaky and build them back, one at a time.

The test exposed what the homework hid. The child did not change. The conditions did.

MAP® and RIT® are registered trademarks of NWEA. Helix Math is not affiliated with or endorsed by NWEA.