Against

This file lists null results, controversies, costs of the proposal, and practice that contradicts a principle.

Null or weak results

Claim at risk Evidence against Source
Overview first helps organizer effect small on immediate tests (d = 0.21, Luiten; delayed 0.19 to 0.38), median 0.48 (mean 0.66) on tests a week or more later (Stone); the unfamiliar-material condition rests on Mayer's vote count (Mayer 1979) and is not confirmed by Stone's effect sizes (very small k) Luiten 1980; Stone 1983; Mayer 1979
Any overview helps a network map did worse than a hierarchical one, most for novices DeStefano and LeFevre 2007; Amadieu 2009
Zoom-lens and spiral sequencing the few tests are null; spiral evidence is single-site and quasi-experimental Van Patten 1986; Harden and Stamper 1999
Skim first skimming gives only moderate comprehension Rayner 2016
Concrete before abstract no pooled effect Fyfe 2014
SQ3R as a whole only the read-recite-review core is tested McDaniel 2009
Prerequisite order transfer between levels is hard to show White 1973
Structured proofs mostly not significant, slightly worse justification Fuller 2014

Controversies

  • Problem first or instruction first. Problem first g = 0.36 on concepts and transfer, none on procedures (g = -0.03) (Sinha and Kapur 2021), but reversed when many elements interact (Ashman 2020). Loibl 2017 says it works only with contrasting cases or built-on instruction.
  • Just-in-time timing. The model predicts theory first, how-to during practice; its own experiment found the reverse most efficient (Kester 2001, 2004).
  • Depth versus breadth. Depth went with better grades (Schwartz 2009), but the data are correlational.
  • Expertise reversal. The same scaffold helps novices (+0.51) and hurts the advanced (-0.43) (Tetzlaff 2025), so a single procedure for all readers is suspect. The authors themselves advise providing assistance rather than withholding it when in doubt, which supports a default for an unknown reader.

Costs of the proposal

  • Goal-bounded depth lowers learning of the non-goal part (Klauer 1984; McCrudden 2005): if the later test asks outside the task set, the rule loses.
  • Stop thresholds come from word-pair tasks (Metcalfe and Kornell 2005; Rawson and Dunlosky 2011; Nelson and Leonesio 1988).
  • Several effect sizes are read in summaries only (marked secondary).
  • Each followed link is a decision that loads working memory; the rule tries to remove the decision.

Failed replications

  • None was located for the organizer, relevance or hypertext-order findings.
  • That is a statement about this search, not about the literature.

Practice against a principle

Practice Principle at risk Evidence
Inverted pyramid cited as support for overview first overview first it is importance order, not a map; its two tests, both theses, disagree (Nildam 1998; Sternadori 2008)
Conclusion first for every message overview first explanation first rated higher for bad news (Jansen 2010)