Baseline
Named baseline
Linear reading: take a document front to back, in its given order, with no overview and no stop rule.
- Every item is read at the same depth.
- Reading stops at the end or when interest runs out.
- Links are followed as they appear, or not at all.
How each rule is compared
Measures, in order of weight: delayed comprehension, transfer to new problems, time to a passed self-test, share of material read.
| Rule | Difference from the baseline | Evidence of the gain | Grade |
|---|---|---|---|
| Overview or map first | adds step 1 | g = 0.70 memory for maps; d = 0.21 for organizers | measured |
| Goal-set depth | adds a task set | prequestions g = 0.54; relevant part up, rest down | measured |
| Related-next order | replaces given order inside the task set | helps low-knowledge readers | measured |
| Self-test stop | replaces reading to the end | overconfident readers stop early; tests correct it | measured |
| Interleaved review | replaces blocked practice | g = 0.42 | measured |
| Proof plan and self-explanation | adds steps | d = 0.95 (self-explanation) | measured |
| Whole procedure | all of the above | no direct comparison | convention, untested |
A test that would settle it
- Arms: linear reading; overview first only; iterative deepening with the stop rule.
- Material: one hierarchical course, same exercises in every arm.
- Outcomes: delayed test on the task set, delayed test outside it, time on task.
- The outside-the-task-set test measures the cost named in against.md.