Presenting multiple cases from different perspectives supports transfer in ill-structured domains¶

Claim · All claims
Evidence · 2 studies · 1 causal r2, 1 quant-synthesis r? · q3–q4 · i2 medium
Currency · newest study 2013 · not yet checked against its sources · not reviewed by a person

Cognitive Flexibility Theory holds that ill-structured domains — where concepts vary unpredictably across situations — require learners to encounter the same content in multiple cases and from multiple conceptual perspectives, so knowledge is assembled flexibly rather than stored as a single oversimplified schema.

Subclaims¶

q3 i? In a small randomised experiment in an ill-structured domain (the social impact of technology), undergraduates who revisited the same cases under several conceptual themes wrote better transfer essays than a drill control, though the drill group scored higher on factual recall. → Jacobson & Spiro 1995

q4 i2 Across 57 experiments, having learners compare cases produced more learning than studying single or sequential cases or conventional instruction (d = .50). The analysis did not isolate ill-structured domains or multiple perspectives, and the benefit was larger at immediate testing than after a delay. → Alfieri et al. 2013

Evidence¶

Jacobson & Spiro 1995¶

Jacobson, M. J., & Spiro, R. J. (1995). Hypertext learning environments, cognitive flexibility, and the transfer of complex knowledge: An empirical investigation. Journal of Educational Computing Research, 12(4), 301–333. doi:10.2190/4T1B-HBP0-3F7E-J4PN

q3 · peer-reviewed randomised experiment · i? · no standardised effect size reported; Session 4 essays F(1,32) = 7.03, p < .01 · n=34 · causal · r2

First- and second-year university students were randomly assigned to a treatment based on Cognitive Flexibility Theory or to one of two computer drill controls. The CFT treatment, "Thematic Criss-Crossing Hypertext", had students reread short cases under different combinations of themes. The two drill groups did not differ and were pooled, leaving 17 per group. The drill group scored higher on short-answer factual questions. By the fourth session, however, the criss-crossing group scored higher on problem-solving transfer essays (mean 9.47 vs 7.24 on a 15-point scale), and the difference held after adjusting for verbal ability and the extra study time. The sample is small, all participants were paid volunteers, and the treatment bundled several CFT features, so the study cannot show which feature produced the effect. The numbers here come from the 1993 technical report (ERIC ED355508), which is the version read. The published 1995 article reports the same study, but its text was not read.

Alfieri et al. 2013¶

Alfieri, L., Nokes-Malach, T. J., & Schunn, C. D. (2013). Learning through case comparisons: A meta-analytic review. Educational Psychologist, 48(2), 87–113. doi:10.1080/00461520.2013.775712

q4 · meta-analysis (random effects) · i2 · medium effect, d = .50, 95% CI [.44, .56] · n=57 experiments (336 tests) · quant-synthesis · r?

This random-effects meta-analysis covers 57 laboratory and classroom experiments. It compared case-comparison activities with sequential, single-case and non-analogous case study, and with traditional instruction. Comparison produced greater learning overall. Four of 15 candidate moderators held up: the objective of the comparison, the presentation of a principle, the content, and the lag before testing. Benefits were larger when learners looked for similarities, when a principle was given after the comparison, with perceptual content, and with immediate testing. The review supports the multiple-case half of the claim but does not test ill-structured domains or multiple conceptual perspectives specifically. Its finding that benefits are larger at immediate testing cuts against CFT's prediction that the advantage should be largest on delayed transfer. Only the abstract was read.

Studies bearing on this claim¶

Does revisiting concepts across varied, compared cases improve transfer to a new case? This table gathers the studies behind this claim and the 4 narrower claims that bear on it: 6 distinct studies (4 causal, 1 quant-synthesis, 1 review), q2–q4, 1 of 6 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Scope. The claim is deliberately restricted to ill-structured domains such as medicine, law, history, and complex management problems. In well-structured domains (e.g., introductory physics problem sets), a single well-formed schema may suffice, and the overhead of criss-crossing multiple cases may not pay off [~M]. The theory's core mechanisms — multiple knowledge representations, criss-crossing the landscape, and avoiding oversimplification — predict that single-case instruction produces knowledge that is inert when the situation changes [-M].

Moderators and boundary conditions. The theory was developed by Spiro and colleagues in the late 1980s partly as a critique of reductive schemas and analogies applied to complex content; it predicts that oversimplified representations actively mislead learners in ill-structured domains [-M]. The approach presupposes substantial prior knowledge to compare across cases, so it is better suited to intermediate and advanced learners than to novices, who may lack the working-memory resources to process several cases simultaneously — consistent with the broader expertise-reversal pattern described in Expertise reversal effect [~M]. Case quality matters: cases must genuinely differ in how the target concept manifests, or learners extract a single rigid interpretation anyway [-M]. Superficially similar cases also impose comparison costs without flexibility gains, so case selection — not just case quantity — is the primary design lever.

Design implications. Practitioners implementing this claim typically sequence several case studies that share a core concept but differ in surface features and in how the concept plays out, then prompt learners to articulate what changes and what stays constant across cases — the comparison mechanism described in Analogical reasoning improves transfer. Deliberately conflicting cases can also induce the productive tension described in Conflict-based instruction improves science conceptual learning, though staged contradictions helped only learners who reported being confused, and no study isolates disequilibrium as the mechanism. The Cognitive Flexibility Theory pattern operationalizes this as "criss-crossing" the case landscape: revisiting the same concept from different cases and perspectives rather than covering each case once. Hypertext and other non-linear formats were the original delivery vehicles for this criss-crossing, but the mechanism transfers to any format that lets learners revisit a concept from multiple case entry points. Because processing several cases at once taxes working memory, designers should scaffold comparisons — highlighting the shared concept and the dimension along which cases diverge — consistent with Cognitive load management.

Open questions. Two studies are recorded: one small randomised experiment (Jacobson & Spiro 1995, n=34) in which revisiting cases under several themes improved transfer essays but not factual recall, and a meta-analysis of case comparison (Alfieri et al. 2013, d = .50) that does not isolate ill-structured domains or multiple perspectives. Spiro and colleagues' foundational studies and the KANE hypertext experiments are not yet recorded, so the multiple-perspective mechanism rests on one small study, and practitioners should treat the design guidance as largely theory-driven [~W]. The closest classroom-level evidence is the multiple-case instruction literature summarized in Case-based learning improves exam performance, but that evidence does not isolate the multiple-perspective manipulation this claim specifies. A second open question is dosage: how many cases, and how much variation between them, are needed before flexibility gains appear — the literature offers no established minimum. A third is whether the hypothesized advantage over well-structured single-case instruction shows up on delayed transfer measures rather than immediate post-tests, which is where the theory predicts its effects should be largest; Alfieri et al. (2013) found case-comparison benefits larger at immediate than at delayed testing, which cuts against that prediction.

Related Claims¶

• Case-based learning improves exam performance — the closest empirical sibling: multiple-case instruction in classroom settings
• Analogical reasoning improves transfer — comparison across cases is the mechanism CFT prescribes, but CFT warns against single-source analogies
• Conflict-based instruction improves science conceptual learning, though staged contradictions helped only learners who reported being confused, and no study isolates disequilibrium as the mechanism — conflicting cases create the productive disequilibrium CFT exploits
• Cognitive Flexibility Theory — the parent instructional pattern this claim underwrites
• Cognitive flexibility — the underlying learner capability
• Expertise reversal effect — bounds the claim: novices may be harmed by multi-case complexity that benefits advanced learners
• Case studies — the primary instructional element for implementing multiple-case sequences
• Productive Failure Improves Conceptual Learning — related
• A flexible hypertext course group averaged 76% on achievement tests against 59% for a direct-design group that studied different software, so the contrast confounds design with content — a narrower finding that bears on this claim
• Drawing on the positional reflexivity of two researchers with distinct perspectives can function as a strength for cogenerating themes and theory in policy evaluation — related
• Deep understanding is characterized by coherent, interconnected mental schemas that make knowledge transferable — related
• Experimental-group students held more positive attitudes toward flexible hypertext course design than control-group students toward direct design (t = 4.723) — a narrower finding that bears on this claim
• Encoding variability across varied example contexts produces decontextualization supporting transfer (review reports DiVesta and Peverly) — a broader claim this one bears on