25 January 2017 in Journal of Research in Science Teaching
23 October 2017 in Journal of Science Education and Technology
12 August 2024 in Science Education
24 June 2026 in SSRN Electronic Journal
Block-based programming has revolutionized programming for kids, and more recently it has also been employed by scientific modeling platforms such as StarLogo and NetTango. Domain-specific block-based modeling is an even newer development that has made computational modeling more accessible to novice learners. However, these environments may also constrain how students express and revise ideas because of the limited set of blocks that are made available. This paper addresses this design trade-off by examining unpacking as a mid-level design strategy for addressing this tension. We designed a modeling environment in which unpackable blocks are domain-specific blocks that can be expanded into an editable internal definition while remaining in block-based form. Drawing on four classroom implementations of a high school biology unit on eutrophication, we analyze how students modified the unpackable blocks while constructing agent-based models of eutrophication. Our analysis focuses on 59 modification episodes based on screen recordings, student artifacts, and artifact-based interviews. Findings show that modifications were organized around two recurring orientations: (1) performance-oriented changes aimed at making models behave as intended, and (2) phenomenon-oriented changes aimed at representing ecological relationships or biological processes more coherently. These orientations were patterned but not deterministic: parameter changes could support phenomenon-oriented reasoning, while code-structure changes could support performance-oriented goals. We argue that as a mid-level design strategy, unpacking could address the tension between accessibility and expressiveness since it turns domain-specific primitives into revisable procedural resources, supporting plural forms of engagement with computational modeling while preserving accessibility.
24 August 2026 in SSRN Electronic Journal
Computational modeling can be a powerful method for scientific inquiry for both researchers and students alike. However, for K-12 students, the underlying rules of a model can remain elusive, potentially limiting opportunities for learning. This manuscript elaborates on the design of a ”code-first” modeling environment intended to help middle school students learn about the biological concept of evolution through natural selection. We present a case study of a pair of 7th grade students in science class constructing computer code to model the effects of natural selection pressures on a population of fictional organisms. Using an emergent complex systems perspective, we examine video and computer log data of this pair working with the learning environment as part of a larger curriculum unit. Our analysis identifies three ways in which the students’ code became a resource for reasoning about natural selection: 1) Over time, students developed a template structure that we call anchor code, on which their subsequent programs built in a modular way; 2) Students’ focus broadened from the behavior of individual organisms to changes in population as a whole; 3) Students shifted from reasoning about how encoded rules impact groups in isolation to reasoning how encoded rules impact groups in relation to each other. In the discussion, we reflect on how the design of this environment supported students’ reasoning about natural selection and how code-first environments might support reasoning about complex systems across a range of scientific phenomena.