Baseline
The named baseline
A plain linear text: prose or bullets in the order of the source. It has no retrieval prompts, no worked examples with fading, no pre-training of terms, no concept map, split placement of words and figures, and no cues.
This is the control condition of the cited meta-analyses.
| Rule or feature | Control it is compared against | Measure |
|---|---|---|
| Folded check question with feedback | restudy of the same text | g = 0.33 (0.73 with feedback) |
| Spaced review of checks | massed review | g = 0.74 |
| Worked example, then faded | problem solving without examples | about g = 0.47 |
| Reader draws the map | reader studies a given map | 0.72 vs 0.43 |
| Words placed in the figure or code | words separate from it | d about 0.72 to 0.85 (secondary) |
| Bold, headings, arrows | no cues | g = 0.53 retention |
| Terms defined before use | terms defined in the flow | median d = 0.75 (one lab) |
| Compare two cases with a prompt | the same cases studied one after another | d = 0.37 |
| Skeleton before the lecture | no skeleton, free notes | recall up, writing down |
| Prequestions on key terms | reading cold | g = 0.54 on questioned items |
How to compare
- Same content, same reader group, same time on task.
- Measure delayed recall and transfer, not only immediate recall.
- Report the effect size and the control, as the sources do.
- Separate novices from advanced readers (expertise reversal).
Not baselined
Rules with no controlled study are marked "convention, untested" and have no baseline.