Drawing improves STEM learning by a small-to-moderate amount over not drawing and over other active study strategies, in one meta-analysis¶

Claim · All claims
Evidence · 1 study · quant-synthesis r? · q3 · i1 small · n=3635 (160 effects, 35 studies)
Currency · newest study 2021 · checked against its sources 2026-10-07 (automated, 1 of 1 entry) · not reviewed by a person

Learners who create drawings of to-be-learned material during study learn more than learners who do not draw (g = 0.362) and than learners using other active strategies such as imaging, explaining or re-watching (g = 0.400), in one meta-analysis of 35 STEM studies read from its abstract. No recorded study compares drawing with summarizing, and none covers subjects outside STEM.

Subclaims¶

q3 i1 A meta-analysis of 35 studies (160 effects, N=3635) in STEM learning found that drawing produces small-to-medium gains over both no-strategy study and other active strategies (imaging, explaining, re-watching). → Zhang et al. 2021

Evidence¶

Zhang et al. 2021¶

Zhang, Y., Guo, X., Pi, Z., & Yang, J. (2021). Learning by drawing in STEM: A meta-analysis. In Proceedings of the Tenth International Conference of Educational Innovation through Technology (EITT 2021) (pp. 83–87). IEEE. doi:10.1109/EITT53287.2021.00025

q3 · meta-analysis in conference proceedings (abstract only) · i1 · small effect vs. not drawing, g=0.362 · n=3635 (160 effects, 35 studies) · quant-synthesis · r?

A meta-analysis of STEM-learning studies published 2005–2020 (35 articles, 160 effects, 3,635 participants) compared drawing to two kinds of alternatives: not drawing at all, and other active study strategies. Drawing outperformed not-drawing (g = 0.362) and outperformed other strategies such as imaging, explaining, and re-watching (g = 0.400); adding scaffolds did not significantly improve outcomes further but did save learning time (g = 1.693 for time saved). The benefit was larger for middle/high-school students than for primary-age students, and online drawing benefited more from instructor-provided scaffolds than paper-based drawing did.

Discussion¶

Mechanism. Drawing is typically explained as a generative learning strategy: constructing a pictorial representation forces learners to select key ideas, organize them into a coherent structure, and integrate verbal and pictorial representations — consistent with Dual Coding Theory and generative activities frameworks. Drawing also imposes cognitive load, so its benefit depends on the drawing not overwhelming working memory — the same trade-off that governs chunking and other load-management strategies.

Moderators and boundary conditions. The benefit should be strongest when learners are supported in drawing accurately — for example, when source material is provided, drawings are compared with model illustrations, or drawing is scaffolded rather than free-form. Poorly constructed or off-target drawings may encode misconceptions, echoing the coherence principle: pictorial work must represent core content, not decorative detail. Learners with low spatial or graphic skill may spend capacity on the mechanics of drawing rather than on sense-making, which predicts an expertise-reversal pattern analogous to that documented for worked examples and consistent with the expertise reversal effect. Drawing appears most valuable as a complement to, not a replacement for, verbal processing of the material — pairing drawing with annotating or self-explanation may distribute the generative work across verbal and pictorial channels.

What the recorded evidence says, and does not. The one meta-analysis above (read as an abstract only) partly answers two of these questions. Drawing beat other active strategies (imaging, explaining, re-watching), not just no strategy. Adding scaffolds did not improve learning outcomes, though it saved learning time. That second finding cuts against the moderator suggested above: in that corpus, support for drawing changed efficiency, not the size of the gain. It covers STEM only, and it found larger benefits in secondary than in primary school.

Open questions. Whether drawing outperforms summarizing or self-explaining specifically, how much the effect depends on the accuracy of the drawing, and whether it holds outside STEM remain open. The primary studies behind the meta-analysis should be recorded here before this page is relied on.

Related Claims¶

• Prompting learners to self-explain, one generative strategy, improves learning by a moderate average amount; other generative activities are not tested by the evidence recorded here — the general claim this finding bears on, with the studies behind it

• Learner-constructed graphic organizers are not shown to outperform provided ones: the one direct test, with college readers, favoured provided organizers on transfer — the general claim this finding bears on, with the studies behind it

• Chunking reduces working memory load. — drawing must fit within the same working-memory limits it is meant to exploit

• Coherence principle: irrelevant material hurts learning. — drawings must represent core content, not decorative detail
• Activation improves learning. — drawing activates and externalizes prior knowledge during study
• Prompting learners to self-explain improves understanding and problem solving on immediate tests by a moderate average amount, with little evidence yet on delayed or classroom gains. — a competing generative strategy; comparisons with drawing are a key open question
• Annotating improves learning. — a lower-cost generative alternative that pairs naturally with drawing
• Worked examples reduce unnecessary search for novices. — model illustrations paired with drawing scaffold accurate pictorial representation, paralleling example-based support
• Generative (Constructive) engagement by young children produces better learning than attentive (Passive) engagement — related
• Decorative Illustrations Do Not Improve Learning — related
• Teaching Others Generative Strategy — related
• Concept mapping improves learning over most comparison conditions by a moderate average amount, but not over time-matched retrieval practice — related
• Teaching reflective questioning or using graphical representations improves procedural and conceptual knowledge across diverse populations — related
• Empirical evidence on learner-generated drawing is inconsistent, and the review organizes it around hypotheses grounded in cognitive and strategic processing — related
• No evidence for a mapping between representational format and enhancement of a specific kind of understanding — related
• Active-processing focus-support techniques such as testing, self-explaining, concept mapping, and drawing show positive effects on learning — a broader claim this one bears on
• Generating drawings is effective for university students but largely ineffective for elementary-school children unless instructional support is extremely high — related
• Misalignment between learning-strategy modality and test modality can mask or distort the detected benefits of drawing — related