In one experiment, students who generated their own underlining did not score significantly higher on tests than students given experimenter-generated underlining¶

Claim · All claims
Evidence · 1 study · causal r? · q3 · n=not reported in what was read
Currency · newest study 1985 · not yet checked against its sources · not reviewed by a person
Evidence quality · No study behind this claim is a peer-reviewed article with a DOI: it rests on an ERIC document (a report, conference paper or thesis). It is kept because 13 design pages cite it; treat it as a lead to check, not an established finding.

One experiment found no significant advantage for students who generated their own underlining over students given experimenter-generated underlining. The entry records no sample size and no effect size, so the result is a single non-significant difference, not a demonstration that the two are equivalent; the entry does not report the result of the study's relevance manipulation.

Subclaims¶

q3 i? In one experiment, students who generated their own underlining did not perform significantly better on tests than students given experimenter-generated underlining; no effect size or sample size is recorded. → Nist 1985

Evidence¶

Primary evidence link: https://eric.ed.gov/?id=ED265520

Nist 1985¶

Nist, S. L., & Hogrebe, M. C. (1985). The Effects of High and Low Relevant Text Underlining on Test Performance. ERIC ED265520

q3 · peer-reviewed experiment, not pre-registered · i? · not significant; no effect size reported in what was read · n=not reported in what was read · causal · r?

In a controlled experiment comparing student-generated underlining with experimenter-generated underlining, the authors found that "subjects who generated their own underlining did not perform significantly better" than those given experimenter-generated underlining. The study also manipulated the relevance of the underlined content (high vs. low relevant text), suggesting that the selection of what to mark — rather than the generative act of marking — drives any effect on test performance.

Discussion¶

Generation is not the active ingredient. A common intuition holds that annotating text oneself is a generative activity that deepens processing relative to passively reviewing someone else's marks. This finding is in tension with that intuition for underlining specifically, though a single non-significant difference with no recorded sample size cannot show that generating the underlining adds nothing. This aligns with the broader literature on text marking, where underlining is among the weaker learning strategies and its value depends heavily on the quality of selection.

Selection quality is the moderator. The study also manipulated the relevance of underlined content, but the entry does not report what that manipulation found. If relevance matters as the authors' design suggests, poorly chosen marks — whether made by students or experimenters — can be useless or harmful, while well-chosen marks help regardless of who makes them. Novices often select poorly when marking on their own, which may explain why self-generated underlining frequently fails to beat provided marks.

Scope limits. The claim concerns underlining specifically and immediate test performance; it does not establish that all forms of annotating are equivalent to provided marks. Richer annotation (marginal notes, elaborative comments, self-questioning) involves different cognitive processes than drawing lines and may show genuine generation benefits. Single-study support in one text domain also warrants replication before strong design prescriptions are drawn.

Related Claims¶

• Relevancy of emphasized text directs attention and influences test performance — the general claim this finding bears on, with the studies behind it

• Annotating improves learning outcomes. — broader annotation evidence; underlining is the weakest form

• Coherence principle: irrelevant material hurts learning. — poorly selected marks act as irrelevant material that diverts attention
• Chunking reduces working memory load. — well-placed marks can segment text, a mechanism independent of who marks
• Highlighting shows low utility for improving learning outcomes — related
• Students who experienced annotation subjectivity firsthand most frequently requested ways to reduce disagreement, indicating they had not internalised disagreement as meaningful signal — related
• Engagement via annotation (underlining, highlighting, notes) shows no significant effect on comprehension — related
• Ten-year-old readers predominantly underline subordinate passage details rather than superordinate conceptual material, even when conceptual postquestions focus attention on topic sentences — related
• Learning by drawing outperforms text-based strategies but is often less effective than studying pre-made drawings, with guidance and drawing training as key determinants — related