Flat MAZE Scores in a Group of Twelve

Do not interpret a group mean as if it described every student. A mean slope of 0.05 hides at least three different stories.

Plausible explanations, in the order worth checking:

1. Measurement. MAZE is a comprehension-oriented but decoding-sensitive measure. If students cannot read the passage fluently, MAZE is partly measuring word reading, not comprehension. Also check passage difficulty, administration conditions, and whether forms were equivalent.
2. Instructional mismatch for a subgroup. Four students with word reading accuracy under 90% and weak nonsense word fluency cannot access grade-level text. A main-idea strategy lesson delivered in grade-level passages does not address their proximal need; their flatness is expected and is not evidence the program is bad.
3. Dosage. 60 minutes per week in groups of eight yields few opportunities to respond. Recommended Tier 2 comprehension dosage is generally higher.
4. Fidelity. The interventionist openly reports skipping guided practice, which is likely the active ingredient. This alone can account for the outcome.
5. Attendance. Two students missing a third of sessions have not received the intervention.
6. Only then, program effectiveness for the students who actually received it as designed.

Conclusion: the data do not support a wholesale program change. They support disaggregation, fidelity repair, and targeted redesign for a subgroup.

Subgroup A: the four with weak word reading (accuracy below 90%)

• Change the target to word reading/decoding: explicit phonics and advanced word study, with connected text at their instructional level.
• Progress monitor with a measure aligned to the target -- nonsense word fluency or word reading fluency -- not MAZE. Measuring a decoding intervention with a comprehension measure guarantees ambiguous results.
• Consider a smaller group (three to five) for more practice opportunities.

Subgroup B: students with adequate decoding but weak comprehension

• Keep the comprehension focus but restore the omitted guided practice; add text-appropriate vocabulary and background knowledge work; consider text at instructional rather than frustrational level.
• Increase active responding (choral, partner retell, writing) and keep MAZE plus a curriculum-based comprehension measure.

Subgroup C: the two chronically absent

• Do not judge the intervention on their data. Problem-solve attendance with the family and the office; restart their clock when attendance stabilizes.

System level: raise dosage if the schedule allows, keep the program for the students it fits, and document the decision rule for each subgroup so the principal sees an evidence-based plan rather than program-hopping.

• Build a checklist of the program's non-negotiable components taken from the publisher's protocol: modeling, guided practice, independent application, corrective feedback, pacing. Keep it to 8-12 observable items.
• Observe directly, at least twice and at varied times, rather than asking for self-report. Score percentage of components delivered and note dosage (actual minutes, sessions held, student attendance).
• Interview the interventionist about barriers -- the report that guided practice is skipped for time is an implementation problem with a solution (restructure the lesson, extend to 30 minutes, or reduce the group size), not evidence of a flawed program.
• Set a threshold in advance (commonly 80% of components) and retrain, coach, and re-observe if it is not met before any conclusion about student response.
• Report integrity data alongside student outcome data in the team report, because a flat slope with 50% integrity is uninterpretable, and a flat slope with 95% integrity is genuinely informative.
• Protect against the common error of concluding 'the program doesn't work' or 'the child doesn't respond' when what actually occurred was low fidelity or low dosage.