Differentiated STEM Instruction in Informal Settings
Informal STEM environments differ from formal ones in ways that matter instructionally: attendance is voluntary and often intermittent, grade bands are mixed within a single room, there is no summative assessment to anchor pacing, and the session ends whether or not the investigation resolved. Facilitation under those conditions is transdisciplinary and student-directed rather than sequenced by curriculum (Margot & Kettler, 2019; Macalalag et al., 2022).
This module sets out the constructs CMSEE uses to plan and read a Saturday room, how each one changes shape across our four program strands, and the vocabulary we share across the team. Which enactment fits a given group, material set, or moment is a judgment call — the module is meant to inform that call, not replace it. Throughout, differentiation means expanding representation and expression while holding cognitive demand constant (Tomlinson, 2014).
Section 1: What this module covers
- Distributing access across modalities without altering the cognitive demand of the task.
- Using a formative probe as grouping data, and recognizing when the data does not support grouping.
- Selecting materials whose affordances span more than one developmental entry point.
- Working with frustration during design failure as a regulation problem before a cognitive one.
- Structuring participation through role interdependence, including its language dimension.
- How each of these changes shape across robotics, engineering, science, and mathematics.
Section 2: The instructional cycle and the Express Phase
Duran and colleagues (2012) modify the BSCS 5E cycle by inserting a diagnostic pause after Explain. The phases run in this order:
- Engage. Establishes the question and surfaces prior interest.
- Explore. First contact with materials, ahead of formal explanation.
- Explain. Concepts and disciplinary language are attached to the experience.
- Express — the inserted phase. An active pause. A short ungraded probe makes visible the mental models students are actually operating with, before those models are built upon.
- Elaborate. Three-tiered, grouped on what the probe surfaced: Tier I (scaffold), Tier II (on-target), Tier III (extension).
- Evaluate. Evidence of understanding through prototypes, drawings, or oral accounts.
Section 3: Five cross-cutting frameworks
These five constructs apply in every strand and every grade band. Each one below pairs the construct with several ways it can be enacted in a CMSEE session. The options differ in preparation cost, documentation, and how much they interrupt the flow of a room, so choices can be made deliberately rather than by default. Section 4 takes each framework into the four strands, where enactment changes considerably.
Modality differentiation
Differentiation operates on representation and expression rather than on cognitive demand. The investigation question, and what counts as a warranted answer, stay constant; the routes into the material and the forms in which understanding is shown vary (Tomlinson, 2014; Vasquez et al., 2013).
Ways to enact it:
- Vary the input: The same source material as a labeled diagram, an audio version, and print, with students choosing.
- Vary the output: The same claim defended through a prototype, an annotated drawing, or a verbal account.
- Vary the scaffolding, not the task: Partial schematics, sentence frames, or a worked first step for groups that need structure rather than a different problem.
The Express pause
An active pause after Explain, in which a short ungraded probe surfaces conceptual models before students commit them to a physical build. The probe is diagnostic rather than evaluative: its value is the distribution of responses, not any individual score (Duran et al., 2012; Keeley, 2005).
Ways to enact it:
- Written probe: A single conceptual question on cards or whiteboards. Collectable, sortable, slowest.
- Prediction with reasoning: Recorded before materials are distributed. Fast, and reveals reasoning rather than only conclusions.
- Peer explanation round: Students explain to a partner while you circulate and listen. Minimal interruption, but nothing is documented.
Material range tolerance
Open-ended materials with wide affordance ranges — blocks, cardboard, straws, tape — admit multiple developmental entry points from a single kit. The same straws support geometric joint-binding for K–2 and analysis of truss load distribution for 6–8 (Vasquez et al., 2013). In a mixed-band room this is a planning constraint as much as a pedagogical one.
Ways to enact it:
- Tier by constraint: Identical materials, different specification: maximum height for one group, load-bearing minimum for another.
- Tier by representation: Identical build, with older students additionally modeling or calculating what the structure is doing.
- Hold a second set in reserve for extension, rather than replacing the first set for advanced groups.
Pause–Name–Return
Design failure is intrinsic to engineering work and, for younger students, frequently produces physiological arousal that precedes any reasoning about the failure. The sequence separates regulation from diagnosis: Pause to let arousal settle, Name the specific failure (a social conflict and a joint failure call for different responses), and Return to the cycle with a revised hypothesis. The reframe that carries it is treating the failed prototype as data.
Ways to enact it:
- Individually: With the student away from the build, when arousal is personal.
- As a team reset: When failure is entangled with a social conflict and separating them is the diagnostic work.
- Named in advance: Introduced at the start as part of the engineering cycle, so the sequence is anticipated rather than remedial.
Role interdependence
In teams of four to six, unstructured participation tends to concentrate: students with more confidence or prior exposure take the materials. Assigned roles — Supplier, Reader, Recorder, Builder — distribute participation structurally instead of requesting it, and establish individual accountability within a shared product (Brooks, 2024). Roles also carry unequal linguistic demand, making assignment consequential for multilingual learners.
Ways to enact it:
- Assign at the start and rotate between phases, so no student holds the same participation form all session.
- Introduce mid-task when participation has already collapsed, as a structural response rather than an individual correction.
- Choose the enforcement mechanism: material gating, where only the Supplier handles components, or product accountability, where each role signs a distinct artifact.
Section 4: The frameworks by program strand
Robotics and Computer Science
Highest-load frameworks: the Express pause, role interdependence.
- Failure is invisible here: Faulty code runs without complaint and produces wrong behavior. The Name step needs debugging vocabulary: is the plan wrong, or is the plan right and sequence mistyped?
- Probe for sequence and causality: Address persistent misconceptions that blocks execute simultaneously or that robots infer goals. Ask students to predict what a three-block sequence will do before running it.
- Scarce resource is screen access: Use paired programming with Driver and Navigator roles rotated on a timer.
- Range tolerance in code: The same Finch task carries K–2 tracing a square block-by-block and 6–8 producing the path with repeat loops and variables.
- Modality without screens: Unplugged routes (acting out sequences, command cards) offer entry points independent of reading interfaces.
- AI activities note: Surface misconceptions that models “understand” rather than match patterns during the Express pause.
Engineering
Highest-load frameworks: material range tolerance, Pause–Name–Return.
- Native range tolerance: Cardboard, straws, and tape support every band. Differentiation happens through group specifications.
- Failure is physical/public: Pause–Name–Return runs often. Naming it in advance frames structural collapse as routine rather than failure.
- Probe material properties: Address beliefs that stronger means heavier or more material means more stable before building begins.
- Tier by constraint: Cost ceilings, mass limits, and budgets convert builds into optimization problems without new supplies.
- Material gating: Supplier role holds real authority because physical components are countable, serving as an inclusion lever.
Science
Highest-load frameworks: the Express pause, modality differentiation.
- Express pause: Science features dense misconception literature (heat transfer, particulate matter). Probes can often be adapted directly from Keeley (2005).
- Expression as differentiation: Notebooking supports UDL via labeled drawings, photo captions, or written prose.
- Range tolerance via questions: Use single specimens or setups across bands (K–2 sorting/describing; 6–8 quantifying/explaining mechanisms).
- Safety constraints: Where handling is restricted, Reader, Recorder, and Observer roles distribute participation safely.
- Trauma-informed hazards note: Signal wildfire/earthquake topics early, keep imagery non-graphic, and offer participation via engineering protection design.
Mathematics
Highest-load frameworks: Pause–Name–Return, role interdependence.
- Cognitive failure: Distress presents as withdrawal or stopping work. Pause–Name–Return should address affective states (“Does this feel like a problem you’re not good at?”).
- Probe reasoning, not recall: Predictions with justification separate students who apply relationships from those executing procedures.
- Deliberate roles: Use Checker (tests against alternate cases) and Presenter (explains reasoning aloud) alongside Recorder.
- Open middles: Use single datasets (e.g., sports stats) for K–2 counting and 6–8 ratios/predictions.
- Tier III openness: Extension means removing constraints to let students justify their own parameters rather than adding tedious steps.
Section 5: Shared vocabulary
| Operational Term | Theoretical Paradigm | Saturday Enactment |
|---|---|---|
| Differentiated instruction | Modality autonomy & equity (Tomlinson, 2014) | Multiple entry points for content, process, and product, with cognitive demand held constant. |
| Express Phase | Formative assessment & misconceptions (Duran et al., 2012) | A short, systematic check-in after Explain that surfaces conceptual models before the build phase. |
| Three-tiered Elaborate | Zone of proximal development (Duran et al., 2012) | Grouping on Express Phase data into Tier I (scaffold), Tier II (on-target), or Tier III (extension). |
| Range tolerance | Parallel usability & material affordance | Open-ended materials supporting simple stacking for K–2 and load analysis for 6–8 from one kit. |
| Pause–Name–Return | Self-regulation & SEL | A three-step sequence separating regulation from diagnosis when a model collapses. |
| Cooperative roles | Positive interdependence | Assigned team functions (Supplier, Reader, Recorder, Builder) that distribute participation structurally. |
Section 6: Applied analysis
One kit, two grade bands
Situation: K–2 and 6–8 students share straws and tape. Younger students practice joint-binding while older students analyze truss loads.
Reading: Shared materials carry two developmental entry points. Differentiation rests entirely in framing each group’s task.
Consideration: Visibility across bands can inspire students or create perceived hierarchies of competence depending on room setup.
Three ways in, one discussion
Situation: Text is offered via audio, diagram, and print. An audio user struggles during group discussion of the visual diagram.
Reading: Input access was differentiated, but the shared reasoning referent was not.
Options: Stagger modalities so all students converge on a shared diagram before building, or use the diagram as a shared team artifact.
Role and language demand
Situation: Two students dominate circuit parts; an instructor assigns a quiet multilingual learner to draw schematics as Recorder.
Reading: Role assignment successfully structures inclusion, but Recorder carries high written-language demands that may limit participation.
Alternative: Assign Supplier (low language, high procedural authority) or allow recording via photographs/diagrams (Brooks, 2024).
When the probe comes back flat
Situation: Express probe returns nearly uniform correct understanding across all students.
Reading: Either students share the concept (proceed undifferentiated at higher demand), or the probe lacked conceptual difficulty.
Indicator: Probes asking for reasoning separate application from recall. Tiering should respond to genuine variation, not custom (Keeley, 2005).
Section 7: Session scenarios
Grades K–2, Engineering: Shake table stability
Moment: A skyscraper collapses on test 1. The student kicks the table and quits.
Response: Apply Pause–Name–Return (pause away from build, help name failure cause, return with one change). Frame collapse as normal data.
Grades 3–5, Robotics & CS: Alarm circuits
Moment: Two students hoard circuit parts while a multilingual teammate disengages.
Response: Assign roles mid-task. Move Supplier (handles components) to the disengaged student and provide bilingual circuit diagrams.
Grades 6–8, Science: Watershed filtration
Moment: Three students miss key concepts on probes while another team finishes in 10 minutes.
Response: Tier I gets a quick gravel vs. sand demonstration. Tier III gets material price tags ($20/oz charcoal) to create an optimization problem.
Section 8: Evidence on repeated exposure
Differentiation in informal settings carries planning costs, but evidence connects repeated exposure across sessions to durable gains in STEM identity and self-efficacy (Barajas-Salazar et al., 2025).
| Sessions Attended | Students (n) | Median Score |
|---|---|---|
| 3 to 4 sessions | 6 | 3.13 |
| 1 to 2 sessions | 8 | 1.84 |
Mann–Whitney U = 41.50, p = 0.020, r = 0.61 (Barajas-Salazar et al., 2025, Frontiers in Education, 10:1534452).
Section 9: Reading list and the rest of the series
Barajas-Salazar, B. E., Almeida, M., Aguirre-Muñoz, Z., & Viveros, M. (2025). Culturally relevant informal STEM learning for underserved students: effects of repeated exposure to the engineering design process. Frontiers in Education, 10, 1534452.
Brooks, S. (2024). Case study: Implementing strategies to meet the needs of English language learners (Applied dissertation). Nova Southeastern University.
Duran, E., Duran, L., Haney, J., & Scheuermann, A. (2012). A learning cycle for all students: Modifying the 5E instructional model to address the needs of all learners. Science & Children.
Johnson, J., Zielinski, M., Essary, J. N., Dean, K., Bartynski, K., & Macalalag, A. Z. (2022). Encouraging STEMpathy: A review of literature addressing STEM learning for students with special education services in inclusive learning environments. In Internalization of STEM Education (pp. 1–40). ISTES.
Kartal, B., Kartal, T., & Tasdemir, A. (2022). How and why teachers implement STEM? A journey to teacher beliefs and teaching practices. In Internalization of STEM Education (pp. 41–74). ISTES.
Keeley, P. (2005). Uncovering student ideas in science: 25 formative assessment probes. NSTA Press.
Macalalag, A. Z., Sahin, I., Johnson, J., & Bicer, A. (Eds.). (2022). Internalization of STEM education. ISTES.
Margot, K. C., & Kettler, T. (2019). Teachers’ perception of STEM integration and education: A systematic literature review. International Journal of STEM Education, 6(1), 2.
Tomlinson, C. A. (2014). The differentiated classroom: Responding to the needs of all learners (2nd ed.). ASCD.
Vasquez, J. A., Sneider, C., & Comer, M. (2013). STEM lesson essentials, grades 3–8: Integrating science, technology, engineering, and mathematics. Heinemann.
