Concreteness Fading for Primary School Computing

Objectives . This paper investigates the Theoretical Model of Concreteness Fading in the context of teaching children aged 9 to 10 about computer network structure and routing. Three between-groups, pre-test/post-test experiments were conducted to test the following hypotheses: 1) The Concreteness Fading group will have higher learning gains than the Abstract, Concrete, and Concreteness Introduction groups; 2) The Physical Concrete group will have higher learning gains than the Virtual Concrete group; 3) The Three-Step group will have higher learning gains than the Two-Step group; 4) The Five-Step group will have higher learning gains than the Three-Step group. Participants . Participants for all three experiments were pupils aged 9 to 10 at a primary school in the South of England, with 166 participants overall. Study Method . Three experiments were conducted, each with a between-groups, pre-test/post-test design using random, or blocked-random, assignment. In Experiment One (n = 59), four variations of physical and/or paper-based representational progressions were compared. In Experiment Two (n = 48), the physical and paper-based Concreteness Fading progression from Experiment One was compared against a tablet-based augmented reality (AR) learning environment and paper-based equivalent. In Experiment Three (n = 59), the AR learning environment from Experiment Two was extended to compare three distinct approaches to linking the concrete and abstract representations. Outcomes were measured as learning gains using two (pre/post) versions of an immediate near-transfer test in abstract, but contextualised, form. All data were collected in a primary school; pre-tests were completed in the classroom as a group, and post-tests were completed individually in a shared corridor workspace. Findings . In Experiment One, an ANCOVA test showed that differences between groups were not significant (F(3, 54) = 2.413, p = 0.077, partial η2 = 0.118), but the Concreteness Fading group achieved significantly higher learning gains than the Concrete group (Mdiff = 1.00, 95% CI [0.26, 1.75], p = 0.010). Hypothesis 1 was not supported. In Experiment Two, a Welch t-test showed that the difference in learning gains between the Physical and Virtual Concrete groups was not significant (Mdiff = 0.47, 95% CI [-0.47, 1.41], t(41.725) = 1.015, p = 0.316). Hypothesis 2 was not supported. In Experiment Three, an ANOVA test showed significant differences between groups (F(2,56) = 3.670, p = 0.032, η2 = 0.116). The Three-Step group achieved significantly higher learning gains than the Two-Step group (Mdiff = 0.99, 95% CI [0.61, 0.99] p = 0.037). Hypothesis 3 was accepted. The Five-Step group did not achieve significantly higher learning gains than the Three-Step group (Mdiff = 0.16, 95% CI [-0.78, 1.09], p = 0.738). Hypothesis 4 was not supported. Conclusions . These findings demonstrate that Concreteness Fading can be an effective method for teaching computing concepts to children and provide significant benefits over using concrete materials only. It should not be assumed that a physical manipulative is required during the initial stage of learning. Virtual manipulatives may be equally effective, enabling an increase in flexibility of design and distribution. One crucial design factor is the level of support provided to learners when translating between representations. Including at least one intermediate stage, designed explicitly to help identify and link mutual referents between concrete and abstract representations, can increase learning outcomes, particularly for those with lower prior test scores.