Reading the Profile: When a Score Gap Rules Out the Obvious Explanation
Jordan, a fifth grader, has these reading results: Word Reading stanine 7 (77th percentile), Reading Vocabulary stanine 6 (60th percentile), Reading Comprehension stanine 3 (12th percentile), Fluency stanine 4 (25th percentile). The scaled-score standard error of measurement is about 3 points, and the Word Reading–Comprehension difference is 14 scaled points. Jordan reads aloud smoothly in class, volunteers answers about stories the class has read aloud, and his teacher says, "He must not be able to decode-that's why comprehension is low." The test does not measure listening comprehension, and no oral-language data are on file.
1. Evaluate the teacher's claim using the score data and the standard error of measurement.
Read the worked answer
The claim does not survive the data.
- Word Reading at the 77th percentile is above average. A decoding problem severe enough to drag comprehension to the 12th percentile would ordinarily depress word reading first; it does not.
- The Word Reading–Comprehension gap is 14 scaled points against an SEM of about 3. That is roughly 4–5 SEMs, far beyond measurement noise, so the gap is real-but its direction points away from decoding.
- The classroom observation corroborates the scores: smooth oral reading and strong performance on text read aloud are inconsistent with a word-level bottleneck.
Caveat worth stating: the Fluency stanine of 4 (25th percentile) is below average, so rate cannot be dismissed entirely. It is a contributing factor to check, not the explanation the teacher proposed.
2. What is the most probable source of the gap, and what two pieces of data would test that hypothesis?
Read the worked answer
Most probable source: language comprehension-specifically inference and integration across text-rather than word recognition. Supporting logic: vocabulary is adequate (60th percentile), so single-word meaning is not the obvious weak link; word reading is strong; and comprehension falls only when the reader must build meaning beyond the sentence. Note the honest limit: the test does not measure listening comprehension, so this is a well-supported hypothesis, not a conclusion.
Two data points to collect:
- Listening comprehension on a passage of comparable difficulty read aloud to him, with the same question types. Strong listening comprehension with weak reading comprehension isolates the problem to reading-specific integration; weak listening comprehension points to general oral-language comprehension.
- An unaided retell and inference probe on a silently read grade-level passage, scored for cross-sentence links. This confirms the failure pattern the scores predict and shows whether the difficulty appears when support is removed.
If those two conflict, a fluency check (rate on grade-level text plus comprehension after a rereading) would separate "too slow to integrate" from "does not integrate."
3. Should Jordan be placed in a decoding intervention? Justify the decision and name the risk of getting it wrong in each direction.
Read the worked answer
No. Word reading at the 77th percentile leaves little room for a decoding intervention to produce comprehension gains, and intervening where the data show strength uses instructional time that the actual weakness needs. The target should be comprehension: text-based inference, cohesive links between paragraphs, retell with motive and cause, and teacher modeling of integration-while monitoring fluency as a secondary factor.
Risk if placed anyway: he practices a skill he already has, shows no comprehension growth, and the team concludes the intervention "didn't work" rather than that it targeted the wrong skill-delaying appropriate help by a semester or more.
Risk in the other direction: assuming comprehension is unteachable, or that the profile is a testing artifact, and providing no intervention at all. Mitigate this by setting a short progress-monitoring window (four to six weeks) with a specific measure and a decision rule written in advance. If comprehension does not respond, revisit fluency and oral language rather than defending the original hypothesis.