Integrating STEM & Computing in PK-12: Operationalizing Computational Thinking for STEM Learning & Assessment

Symposium at ICLS 2020

June 22, 2020

Organizer/Chair : Shuchi Grover

TABLE OF CONTENTS

Designing for Synergistic Learning of Science and CT in C2STEM

Shuchi Grover, Looking Glass Ventures/ Stanford University

shuchig@cs.stanford.edu | @shuchig | shuchigrover.com�Gautam Biswas, Vanderbilt University gautam.biswas@vanderbilt.edu

Nicole Hutchins, Vanderbilt University nicole.m.hutchins@vanderbilt.edu | @NicoleUSVI

​

​

STEM+Computing in Scientific Modeling

• STEM and CT essential skills for 21st century workforce (NRC, 2011, Wing, 2016)
• “Using Math & CT” -- key science and engineering practices (K-12 Science Education (NRC, 2012); Next Generation Science Standards (NGSS Lead States, 2013); Integrative Science Framework (Honey, et al., 2014)
• Programming & computational modeling can serve as effective vehicles for learning challenging science and math concepts (diSessa 2001; Jona, et al., 2014; Sengupta et al., 2013)
• CT will help students “become creators, and not just consumers of the next wave of computing innovations” (Schnabel, 2011; Wing, 2006)

Operationalizing CT in Computational Modeling

• Algorithmic thinking, Debugging, Problem decomposition, Pattern recognition, Abstraction and generalization (Grover & Pea, 2013, 2018)
• Our analyses focused mainly on �
• Building models, Using data (Weintrop et al. 2016)

​

C2STEM Learning by Modeling environment

• Collaborative, Computational STEM learning environment
• Students learn STEM topics by building computational models
• Incorporates physics DSML blocks with a visual, block-based environment

​

Integrating STEM & CT in C2STEM

• Supporting student learning through... �
• Domain specific Modeling Languages (DSMLs): Modeling language focused on STEM domain, specified at the right level of abstraction to support computational modeling (Hutchins, et al, in press)
• Evidence-centered design of learning and assessment activities (Mislevy and Haertel 2006; Hutchins, et al, 2020)
• Scaffolded exploratory Learning ₋ combining block structure modeling with visualization of dynamic behaviors using animation and plots (Grover et al, 2019; Hutchins, et al, 2020)
• Supporting debugging processes (Emara et al, 2020; Snyder et al, 2020; Snyder et al, 2019)
• Collaborative work in dyads or triads (Grover et al, 2019, Snyder et al, 2019)

Assessment Examples

Example Formative and Summative assessment questions targeting the 3 key CT problem-solving skills (algorithmic thinking, debugging, and problem decomposition)

C2STEM Classroom Studies

C2STEM Studies of Physics and Marine Biology units conducted in middle and high school classrooms in TN, CA, IL, and MA (2017-2019).

​

Data Sources: Pre-Post Assessment, Embedded Assessments, Video Analyses, PFL Assessments

​

Results from Nashville, TN study in Physics classrooms (Hutchins et al. 2020)

Results of PFL Assessment

Acknowledgements & Additional Information

• Primary Researchers: Gautam Biswas, Shuchi Grover, Kevin McElhaney, Dan Schwartz, Luke Conlin
• Website: www.C2STEM.org
• Publications: https://wp0.vanderbilt.edu/oele/publications/
• C2STEM flyer: https://cdn.vanderbilt.edu/vu-wp0/wp-content/uploads/sites/226/2017/08/14230945/PI-C2STEM-Flyer-Final.pdf
• C2STEM in the classroom (video): https://stemforall2019.videohall.com/presentations/1583

NSF Award DRL-1640199

Heterogeneity and Practice: Programming as Expressive Media for K12 STEM

Amanda C. Dickes*, Amy Voss Farris*, Pratim Sengupta* (*Equal contribution)

Operationalizing CT

How can computational thinking support students in the complexity and ill-defined nature of scientific modeling in K-8 classrooms?

​

• Scientific modeling is the key epistemic and representational practice in the sciences (Lehrer & Schauble, 2006; Nersessian, 1999)
• Scientific sensemaking is dialogical: social, interpretive, and involves complex material entanglements (Lehrer & Schauble, 2006; Pickering, 1995)

​

​

Computational Literacy and Heterogeneity

• K12 science is a rich context for integrating computational thinking and modeling (Sengupta et al, 2013; Clark et al, 2015)
• Includes deep engagements with data representation (Lee & Wilkerson, 2018; Weintrop et al, 2016)
• Integration in K12 science affords adoption of new literacies in service of scientific problems and questions (diSessa et al, 1993)
• Participation in computing as a heterogenous and multi-voiced experience (Bakhtin, 1983)

​

Setting

• Two-year, longitudinal design-based research study in 3rd and 4th grade
• 99% African American, high poverty public charter school in an urban school district in the Southeastern US.
• Emma, the classroom teacher, taught both years and co-designed integrated STEM+C lessons which privileged embodiment, mathematization & the “goodness” of scientific models
• 10 Grade 3 students looped to Grade 4

Year 1 (Grade 3)

7 months of activity

15 students: 14 African American, 1 Latino

Year 2 (Grade 4)

9 months of activity

21 students: 19 African American, 1 Latino, 1 Somali

Phases of Activity: Year 1

Geometry

2 Months

Kinematics

3 Months

Ecology

2 Months

Phases of Activity: Year 2

Geometry

1 Month

Kinematics

7 Months

Ecology

1 Month

Transformations (Year 1)

Computing expanded beyond the computer across tangible, lived and paper-based mathematical expressions

​

Embodied Modeling

Inventing Measures

Coding

Modeling & Mathematical Formulations

Transformations (Year 2)

set step size 45

pen down

repeat 11

go forward

set step size minus 6

place measure point

​

• Across both years CT integration is a mangle among theory, experiment, and computing (Pickering, 1995)
• Computing is no longer “on the computer”

​

Discussion and Implications

In this work, we’ve investigated how computational abstractions can become contextualized in praxis.

Students did more than learn programming or computational thinking: They engaged in a dialectical relationship between the tangible work of modeling and authored computer programs as explanations of the natural world

Understanding how computational thinking is experienced by students and teachers requires us to broaden and deepen our inquiry into their experiences beyond thin notions of “thinking” and device-level representations of “computational abstractions.”

Supported by the National Science Foundation under a CAREER Grant awarded to Pratim Sengupta

(NSF CAREER OCI #1150230)

Papers available: http://www.m3lab.org/publications

​

Forthcoming book from MIT Press

Voicing Code in STEM

A Dialogical Imagination

By Pratim Sengupta, Amanda Dickes, and Amy Voss Farris

​

Moving from Literal to Principle-Based Computational Reasoning: A Learning Progression for Integrating Computational Thinking with Earth and Environmental Sciences Instruction

Beth A. Covitt, Kristin L. Gunckel, Alan Berkowitz, & John C. Moore

Problem - Addressing the need for:�-Defining discipline-based CT for “competent outsiders,” -Understanding students’ CT sense making and�-Designing and implementing responsive instruction

While frameworks defining K-12 STEM-related CT have been published (e.g., Weintrop et al. 2015), most of these have articulated target concepts and practices for students to learn. Relatively little scholarly work has examined ways (including informal ways) that students make sense of CT and computational modeling (e.g., Wilensky & Reisman 2006).

Further, much CT scholarship has focused on education aimed at supporting students in moving toward further CT-related studies and careers. Less work has considered what CT knowledge and skills might be needed by “competent outsiders” who encounter and need to deal with issues in their lives such as climate change or environmental contamination. Competent outsiders are “nonscientists who can access and make sense of science relevant to their lives” (Feinstein et al., 2013).

Thus, within CT scholarship, there is need to address questions including…

What knowledge and practices of CT are needed to be a competent outsider in responding to socio-environmental problems situated in Earth and environmental systems (EES) science disciplines?

How can we help high school students develop the CT knowledge and practice necessary to be competent outsiders addressing problems like environmental contamination and climate change?

The Comp Hydro Project

The Comp Hydro project addresses the needs described in the previous slide through using a learning progressions (LP) approach to:

1. Integrate ideas from CT, systems thinking, and EES sciences to articulate an empirically-grounded LP framework that spans from informal to target ways students make sense of CT in EES computational modeling contexts.
2. Develop and test instructional approaches designed to respond to students’ informal ways of thinking and help them move toward target CT.
3. Provide evidence of student learning as a result of instruction designed with reference to the LP.

Context

Comp Hydro includes sites in 4 states. Each state site developed a place-based unit integrating CT and computational modeling into high school hydrology instruction.

We report on implementation from 2 sites. In both states, the 2 to 3-week long unit context was a local case of groundwater contamination. Data are from 19 teachers and 1,279 students.

At 1 state site, participants were from 1 urban school district with a student population that is over 90% Persons of Color. The other state site included 8 school districts in a range of communities from rural to small urban areas; most students are White.

LP Research Methods

Our discourse-based LP research involves iterative assessment cycles aimed at developing, refining, and validating a LP framework over multiple years.

Relevant literature including our past research was used to develop an initial upper anchor (target) for integrated hydrologic and CT sense making.

Pre/post assessments with items that elicit constructed responses were used to examine students’ ways of making sense of CT in a hydrologic science context.

An Item Response Theory (IRT) analysis approach was used.

�Operationalization of CT in EES Sciences:�Upper Anchor Framework & Example Item

Competent outsiders can understand and reason about:

• Defining the system of study (e.g., employing abstraction to reduce a system into fundamental parameters)
• Making sense with system data and representations (e.g., interpreting and making inferences from data representations such as graphs and maps)
• Explaining and predicting events with imperfect data and models (e.g., judging the validity and limitations of a computational model and its outputs)

Example Item: Judging Uncertainty in Model Outputs

Methods: Curriculum Approach

Comp Hydro units are designed to be accessible to students with informal ways of making sense of computational modeling, and to scaffold all students toward target sense making. This is accomplished through moving from more concrete to more abstract learning experiences.

​

Students engaged in multiple connected experiences with different groundwater system models, moving from concrete (e.g., with physical models) to more abstract (e.g., with 2D representations like maps and cross-sections and computational models) experiences over time.

​

In an example lesson, students used data from monitoring wells collected through a virtual Google Earth tour to model a selenium contamination plume (first by hand, then with a NetLogo model). They developed understanding that inputting additional data is one approach to reducing uncertainty in computational modeling.

Results: LP Framework for CT in EES Contexts

Less sophisticated literal model users explain models through the lens of a game player and are skeptical that computer models can represent the real world.

Proficient principle-based model users can explain how computational operations are used to define systems, generate & interpret data outputs, and calibrate and judge models.

Nascent principle-based model users can explain how models represent the real world and are just beginning to learn how that works.

Results: Transitions in CT Sense Making

Less sophisticated literal model users explain models through the lens of a game player and are skeptical that computer models can represent the real world.

Proficient principle-based model users can explain how computational operations are used to define systems, generate & interpret data outputs, and calibrate and judge models.

Nascent principle-based model users can explain how models represent the real world and are just beginning to learn how that works.

The transition from nascent to proficient principle-based model user represents a shift in depth of knowledge and practice rather than a shift in world view. For example, nascent principle-based model users can explain that model accuracy is important for addressing real world problems but it is not until they shift to proficient principle-based model users that they can explain how data-based operations such as calibration may be used to refine a computational model.

The transition from literal model user to nascent principle-based model user represents a significant shift in how students make sense of the world. While literal model users can manipulate and interact with models, it is only after they shift to nascent principle-based model users that they see the potential power that computer models built using fundamental principles defining how systems work have for helping to explain and predict what happens in the real world.

Results: Evidence of Student Learning

For all three progress variables, the effect size for pre to post change was medium. On average, students moved from low to high “literal model users” LP level range for defining the system items. They moved, on average, from “literal model users” to “nascent principle-based model users” range on data sense making and explaining and predicting with models items.

For all three progress variables, distributions of students’ proficiency scores included students who performed at the level of ”proficient model-based user,” suggesting that this target level is achievable for high school students.

Change from pre to post for Weighted Likelihood mean Estimates (WLEs)

Discussion & Implications

The Comp Hydro project adds to understanding of the less and more formal ways high school students make sense of CT and computational modeling in the disciplinary context of EES sciences.

Knowing how students make sense of computational modeling provided us with the opportunity to design and implement instruction that is responsive to students’ informal sense making approaches and that can support students in moving along a trajectory toward knowledge and practice needed as competent outsiders who can make sense of and judge computational models and their outputs as they participate in responding to real world problems such as groundwater contamination.

Consequently, the Comp Hydro project has provided evidence that through short 2 to 3-week units that move from more concrete to more abstract learning experiences, and without engaging in coding, high school students can begin to develop CT knowledge and practices needed to be competent outsiders participating in addressing socio-environmental problems.

Questions we are interested in pursuing in the future include… (1) How this work may serve as a model for developing LPs and responsive instructional design in other CT-related disciplines? And (2) how we can support teachers in understanding and responding to the ways their students make sense of computational modeling?

​

​

​

​

Project Website and Contact Information

Visit the Comp Hydro website at http://ibis.colostate.edu/CompHydro/Index.php for more information about our project, curriculum, and research.

Beth A. Covitt, University of Montana, beth.covitt@umontana.edu

Kristin L. Gunckel, University of Arizona, kgunckel@email.arizona.edu

Alan Berkowitz, Cary Institute of Ecosystem Studies, berkowitza@caryinstitute.org

John C. Moore, Colorado State University, jcmoore@nrel.colostate.edu

This work is supported by NSF STEM+C: (#1543228) Research on Effects of Integrating Computational Science and Model Building in Water Systems Teaching and Learning. Any opinions, findings, and conclusions or recommendations expressed here are those of the authors and do not necessarily reflect the views of the National Science Foundation.

​

CT-ifying STEM Education: Co-designing with teachers to integrate computational thinking into high-school math and science curricula

Golnaz Arastoopour Irgens, Michael Horn, and Uri Wilensky

Operationalizing CT: CT-STEM Taxonomy

(Weintrop, D., Beheshti, E., Horn, M. et al., 2016)

Designing with Teachers to “CT-ify” Curricula

• CT Summer Institute (CTSI) took place in 2019 and will take place in 2020 virtually for 4 weeks.
• Successes: Curricula were co-designed and implemented by high school STEM teachers
• Challenges: How do we choose which concepts to CT-ify? And what are the roles of the researcher and teacher in this partnership?

(Bain, C., & Wilensky, U., 2020)

“If I didn't have a co-design partner, I would probably spend like a whole day on three lines of code and figuring out how that works. I felt I didn't have to struggle through that aspect of it and it was more thinking about: here's what I want my students to take away.”

- High school mathematics teacher

​

“It was really helpful to have people with a different background who could offer insight on what would work best and be most practical”

- High school physics teacher

​

“[My co-design partner] made something that seemed unobtainable and foreign to me -not- and he found a way that it worked for me, which might be different for everyone [else]”

- High school biology teacher

​

​

Measuring Learning

• Surveys to measure attitudes towards computational tools and thinking
• Validated Pre-Post tests to measure science knowledge and CT-STEM practices based on taxonomy, learning objectives, and tools used in unit
• Epistemic Network Analysis to dynamically model and measure CT-STEM practices as students engage with curricular unit.

(Arastoopour Irgens, G., Dabholkar, S., Bain, C. et al. , 2020)

“Plants indirectly affect the wolf population because when there are a lot of plants, the animals that eat and are hunted by wolves increase, giving wolves more food to hunt… This ecosystem is not stable because the moose population lived past the wolves.”

-Written reflections from a student whose discourse network is shown above

Computational thinking and modeling for elementary science education via

immersive virtual worlds

Shari Metcalf, Soobin Jeon, Amanda Dickes, and Chris Dede

Operationalizing CT

• The NGSS (2013) identifies CT and scientific modeling as essential STEM education practices.
• Computational modeling can be considered a subset of computational thinking as applied to STEM education (Sengupta et al., 2013; Weintrop et al., 2016).
• Computational thinking can augment problem solving and sense-making regarding data (Waterman, Goldsmith, & Pasquale, 2019) and computational modeling is vital in expanding knowledge that are not available otherwise through understanding the phenomena being modeled (Sanford & Naidu, 2016).
• Construction of concept maps can provide students opportunities to express their understanding in a representation that can be shared and discussed. Active construction can promote deeper learning and improve learning processes and outcomes of inquiry learning and problem solving (Janssen, Erkens, Kirschner, & Kanselarr, 2010).
• Programming and CT are deeply intertwined. Studying a phenomenon by constructing a model provides opportunities for students to organize and test their knowledge of science concepts by converting concepts and relationships into computational structures, that then can be executed to generate model behaviors (Sengupta et al 2013).

​

EcoMOD

• EcoMOD is a 3rd grade inquiry-based curriculum that blends ecosystem science learning and computational modeling.
• The virtual 3D and 2D worlds help students learn how the ecosystem and organisms within it are affected by beavers building a dam. (figures to the right).
• The curriculum also includes paper-based activities. On days 3, 9 and 14, students draw paper-based causal concept maps that represent the complex relationships between the different factors of the ecosystem.
• During March to June 2019, the EcoMOD curriculum was implemented in eight classrooms at five different schools, each in a different district, with seven teachers and about 150 students.

Students learn about a 3D forest ecosystem by using various tools such as a calendar, population, and beaver point-of-view tool. “Being” a beaver helps students learn the steps involved in building a dam.

​

During days 4 to 7, students use the 2D modeling tool to construct an agent-based computational model of a beaver building a dam, using customized programming blocks. A 2D sandbox lets students see model outcomes and changes in the ecosystem.

​

Measuring Learning of Concept Mapping

• Construction of concept maps exhibit growing depth of knowledge during the curriculum, especially around cause-effect relationships.
• Data from 4 classrooms at 2 different schools, with 3 teachers
• 153 complete sets of concept maps generated by 51 3rd grade students

​

Example of a student’s concept map during three days in the curriculum.

​

Concept Map Assessment Rubric

• A concept map assessment rubric was developed to evaluate both student understanding of key ecosystems ideas and their use of concept mapping as a representational model of the system, comparatively assessing student performance over time.
• We recognize 26 core relationships.
• There are 3 categories of systemic relationships: beaver, pond, and long-term
• Each relationship was coded as a 0, 1, or 2, based on an evaluation of partial or full representation of the concept (Yin et al., 2005).

Completeness of Concept Maps

by Types of Claims by Day

• Beaver claims remained unchanged across the three days, but claims about the Pond and Long Term both increased significantly (F = 19.26, p < 0.0001; F = 114.5, p < 0.0001).
• Claims about the Pond increased significantly between days three and nine (diff = 0.26, p < 0.001), then did not change significantly between days nine and fourteen.
• Claims about Long Term effects did not change significantly between days three and nine, but increased significantly between days nine and fourteen (diff = 0.60, p < 0.001).

​

Discussion: Concept Map Analysis

• Concept maps were significantly more detailed and included more complex pathways and relationships, including more nodes, connections and articulations.
• Scores of completeness of claims increased in newer categories over time, and did not change significantly for claims in older categories, suggesting retention of knowledge that was previously learned.
• Both individual and whole-class construction of concept maps are valuable as assessments:
• Being able to construct concept maps individually prompted students to consider the relationships among factors within the ecosystem, and to reflect and synthesis their understanding of interrelationships.
• Whole-class constructions allowed teachers to elicit their students’ ideas, identify and correct misconceptions, prompt for evidence and reasoning, and foster discussions valuable for developing students’ understanding.

Measuring Learning of Programming

• A custom block-based programming tool for agent-based programming, where student pairs construct agent-based models of a beaver building a dam, was designed.
• Data from five classes from three schools of different school districts, led by four different teachers. Data from 47 pairs were analyzed.
• We designed two assessment processes
• First, we designed a rubric-based assessment to determine how far each pair got to successfully complete the programming task, based on the pair’s final programming product. This was a measure of functionality.
• We found limitations in our findings so, we developed an additional analysis.
• Second, we designed a rubric-based assessment to capture developing fluency in the use of CT constructs of sequences, loops, and conditionals. This was a measure of conceptual fluency.

Functionality Rubric

Conceptual Fluency Rubric

With no prior instruction, 15 of the 47 pairs were able to construct a fully functional computational model, and an additional four pairs were able to complete the task except the final step of building a lodge. Using the functionality rubric, we found that pairs had an average score of 3.38, with significant hurdles moving to multiple conditionals and loops that often resulted in a score of 0 for partially complete models.

​

Pairs that were not able to finish in the limited amount of time allotted still demonstrated learning of CT concepts such as sequence, loops and conditionals. All pairs showed an understanding of sequence, and almost all pairs were able to at least engage in use of multiple or nested conditionals, even when some encountered hurdles in getting them to work together correctly.

​

The rubric scores how well the final model achieved the programming task: beaver builds a dam and lodge while avoiding a predator. It was based on 7 required steps.

​

The rubric scores development of students’ fluency by assessing complexity in the use of sequences, loops, and conditionals.

​

​

Discussion: Program Analysis

• We found that each rubric provided a different lens on the data. The functionality rubric assessed how well the program achieved the task, but it was less successful at correctly identifying CT progress in programs that were partially correct. A rubric for conceptual fluency, in contrast, was more able to recognize stages of development in use of CT concepts.
• We found that with no direct instruction on programming, 3^rd grade student pairs with minimal prior programming experience were able to make progress in using computational concepts of sequencing, loops, and conditionals.
• Our research explores how the design of the intervention, using a visual, block-based interface and a small set of custom, domain-specific building blocks, made the task more accessible.

For More Information

For more information about our work, including videos, resources, and publications, visit:

https://ecolearn.gse.harvard.edu/projects/ecomod

​

This work is supported by the National Science Foundation grant DRL-1639545 to Chris Dede, Karen Brennan and Tina Grotzer at the Harvard Graduate School of Education. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors, and do not necessarily reflect the views of the National Science Foundation.

​

Computational Thinking Practices in an Interdisciplinary Middle School Curriculum

Gilly Puttick ¹ , Debra Bernstein ¹, Kristen Wendell ², Ethan Danahy ², Michael Cassidy ¹, and Fay Shaw ²

¹ TERC

² Center for Engineering Education and Outreach, Tufts University

Designing Biomimetic Robots

• 4-week middle school curriculum
• Design challenge: make a search and rescue robot that can dig through rubble
• Explore animal structure-function relationships
• Engage students in engineering design and CT practices

CT, Science and Engineering Practices Mapped to Curriculum Activities

Operationalizing CT

*Drawn from: International Society for Technology in Education (2011). Operational definition of computational thinking for K-12 education. Retrieved from: http://www.iste.org/docs/ct-documents/computational-thinking-operational-definition-flyer.pdf; K-12 Computer Science Framework Steering Committee (2016). K12 computer science framework. Retrieved from: www.k12cs.org

Measuring Learning in a Focus Group

Decomposition (Biology): Students identify specific body parts that they think help the pangolin dig – claws, hands, head and nose – and document how each part contributes to the animal’s ability to dig.

​

Abstraction (Biology): Students represent pangolin’s motion as a 3-step process in a storyboard, including pertinent details only: extends arm, scoops dirt, pulls back dirt, curves its hand, scoops with claws, directional arrows

Abstraction (Engineering): Students align their understanding of digging movements to an engineering mechanism, a cable drive that will close the pangolin’s claws, and represent it in a design sketch

Conclusions

The robotics design task provides an opportunity for students to use CT practices (decomposition, abstraction, algorithmic thinking, and iteration) as they describe, discuss, and argue about structure-function relationships in animals and robots.

​

Thus, CT supports interdisciplinary learning by providing a common set of practices that can be used across multiple disciplines.

​

Contact for more information - Debra Bernstein <debra_bernstein@terc.edu>

​

This material is based upon work supported by the National Science Foundation under Grant Number 1742127. Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Situating Computational Thinking in the Context of Systems Modeling Using an

Approach to Expand Equitable Access

Daniel Damelin, Steve Roderick, Lynn Stephens, and Namsoo Shin

Systems Thinking (ST) and Computational Thinking (CT) are Critical

• Many complex phenomena are best understood and complex problems addressed using both ST and CT
• Climate change
• Economic policy
• Urban planning
• Epidemics

​

​

Defining Systems and Computational Thinking

Systems Thinking

• Defining a system (boundaries and structure)
• Engaging in causal reasoning
• Recognizing interconnections and identifying feedback
• Framing problems or phenomena in terms of behavior over time
• Predicting system behavior based on system structure.

Computational Thinking

• Decomposing problems such that they are computationally solvable
• Creating artifacts using algorithmic thinking
• Generating, organizing, and interpreting data
• Testing and debugging
• Making iterative refinements

Aspects of ST and CT culled from multiple resources, filtered by common inclusion in literature, specificity to that field, and ability to operationalize for student engagement and measurement.

Computational Modeling Naturally Integrates Systems Thinking and Computational Thinking.

• Systems thinking and computational thinking are essential for reaching the educational and societal goals.
• Modeling engages ST and CT for exploring a complex world on a manageable scale.
• STEM educational standards moving toward greater inclusion of modeling, ST, and CT.

​

​

​

Solving Problems and Understanding Phenomena

Computational Thinking

Systems Thinking

Modeling Practices

A Framework for Integrating ST and CT

Computational Modeling tool: SageModeler

• Engages students’ ST and CT through building, testing, sharing, evaluating, and revising models.
• Lowers barriers to engagement by facilitating student-constructed computational models without the need for coding or writing equations.

​

SageModeler

Computational Thinking

Systems Thinking

Modeling Practices

Curricular Design Implications

Project Based Learning (PBL) approach

​

​

SageModeler

Computational Thinking

Systems Thinking

Modeling Practices

PBL

• Exploring phenomena
• Student-centered and equity-oriented
• Iterative approach to building knowledge
• Using the tools of scientists to engage in collaborative sensemaking of the world (experimentation, modeling, systems, and computational applications).

Measures of CT and ST: Evidential indicators

Using the Framework, one can develop indicators of engagement in ST and CT.

Design and Construct Model Structure

Example modeling practice

Selected relevant ST and CT practices

ST: Engaging in causal reasoning

ST: Recognizing interconnections and identifying feedback

CT: Decomposing problems such that they are computationally solvable

CT: Creating artifacts using algorithmic thinking

Selected data sources

Whole class discussion

Student writing

Modeling artifact

Screencast recording

Selected indicators

Model rubric for characterizing quality of problem decomposition and causal reasoning

Coding classroom discussion, screencast recordings, interviews, and student writing for identification of feedback mechanisms

Teacher and student interviews

Questions?

​

• If you have any further questions about the project, contact at
• Daniel Damelin <ddamelin@concord.org>
• Steve Roderick <sroderick@concord.org>
• Lynn Stephens <lstephens@concord.org>
• Namsoo Shin <namsoo@msu.edu>

​

• Thank you for MCM team members, teachers, and students

​

​

Designing Teacher Professional Development to Support CT Integration in Middle School Science

Irene Lee and Emma Anderson

Thanks to the generous support of NSF AYS #0639637

NSF DRK12 #1503383 / 1639069

T

Teachers with GUTS

(Growing Up Thinking Scientifically)

Between 2016-2019, 68 teachers from 7 districts participated in the program.

​

Teachers with GUTS’ research studied teachers’ learning and enactment of Project GUTS’ CS in Science curriculum and student learning.

Research Question:

“How can we enhance the ability of middle school science teachers to provide high quality CT experiences for students within regular school day science classes?”

Project GUTS’ CS in Science is a middle school curriculum that integrates CS/CT in Science through the use, modification and creation of computer models of scientific phenomena. The Teachers with GUTS PD prepares teachers to implement 4 modules in Earth, Life, and Physical science during the regular school day science classroom.

​

The PD program includes a 1-week summer intensive workshop, quarterly face-to-face 1-day follow up workshops, monthly webinars, and participation on the TeacherswithGUTS.org online network.

Operationalizing CT

CT as “formulating problems and their solutions so that the solutions are represented in a form that can be effectively carried out by a computer.”

Wing (2006, 2011)

Computational Thinking (CT)

Operationalized In Computer Modeling & Simulation context

ABSTRACTION

​

What should I include in the model?

​

AUTOMATION

ANALYSIS

​

How should I encode processes?

How should I make the model run?

​

What data will I collect from runs?

What do the data tell me?

In what ways is the model valid?

​

​

“What does it mean to be a computational thinker?”

Operationalized for Teachers

A computational thinker can…

• Decode and understand computer models.
• Modify and create models using abstraction and automation.
• Use models for scientific inquiry.
• Think of real-world phenomena in terms of how they can be modeled studied.
• Analyze models.

Teachers engaged in these practices as they experienced the curriculum as learners during PD then supported students in these practices.

Integrating CT and Science in Project GUTS

Through learning how to code simple movement, then interactions with the environment, and finally interactions between agents, learners gain the skills needed to create a first model of the spread of disease.

​

This Earth Science module considers how humans are impacting the environment and how resources are being used and managed (or not managed).

In this Life Science module learners modify a simple predator prey model to explore how population dynamics, ecosystems dynamics, and feedback loops.

This Physical Science module explores chemical reactions: the conditions under which they occur, limiting reactants / reactants in excess, and when chemical reactions stop.

Linking models, CT, and inquiry in Module 1

What are models? Students learn what computer models are and what they are good for by comparing and contrasting real world “participatory simulations” and the same activities as computer models.

​

CS/CT skills: Students build CS/CT skills incrementally by coding simple movement, then interactions with the environment, and finally, interactions between agents. With these skills students create a first model of the spread of disease.

​

Scientific inquiry: Students study the spread of disease under different conditions using their model as an experimental testbed. They analyze data generated by their model to understand epidemics.

​

Linking models, CT, and inquiry in Modules 2, 3, 4

Lee, I., Martin, F. Denner, J., Coulter, B., Allan, W., Erickson, J., Malyn-Smith, J., & Werner, L. (2011). Computational Thinking for Youth in Practice. ACM Inroads 2 (1): 32-37.

Modules 2, 3, and 4 follow a Use-Modify-Create progression wherein students:

​

Use pre-existing models to run simulation experiments. �

Decode models to assess the validity of abstractions made by others.

​

Modify models to be able to answer new questions or

refine models.

​

Create new models that reify students’ scientific understandings.

Assessment Measures and Methods.

Data sources

• Knowledge and skills in CT (KS-CT) pre/post
• Classroom observations
• Post-observation interviews
• Artifact based interviews

Measuring

Teacher learning

The KS-CT consists of four scales relating to Weintrop et al’s (2016) Computational Thinking in Mathematics & Science Taxonomy.

​

The KS-CT assesses knowledge across three categories of the taxonomy (Systems thinking, Modeling and simulation, Computational problem solving) and skills aligned with one of the categories (Programming and debugging).

​

​

Cohort 3 change in KS score from pre to post.

In cohorts 1 & 2, Science teachers (in purple) benefitted most from the PD program compared to math (in cross hatch), technology teachers (in boxes).

In cohort 3 (science teachers only), teachers on average showed a 10% improvement in scores (with statistically significant improvement on the CS and Trace and debug scales).

Teacher Enactment of the curriculum

Classroom observations illuminated a variety of enactments of integrating CS/CT in Science. In analysis, we distinguished three approaches:

​

​

A) Coding centric (emphasized learning to program);

​

B) Modeling centric (emphasized abstractions and assumptions in models)

​

C) Experimentation centric

​

​

​

​

​

​

​

In post-observation interviews, we found teachers’ beliefs about the fit of the curriculum, beliefs in students’ capabilities, preparation in CT, and epistemic beliefs about science came into play in their enactments.

​

Used Module 1 to introduce coding then some stopped implementing.

​

Focused on analyzing models (for components and mechanisms) rather than on programming of experimentation.

​

Emphasized the use of models as experimental testbeds (with varied amounts of decoding / examining the models themselves.) Implemented more modules than teachers with other approaches.

​

​

​

​

​

Is deep CS knowledge necessary to integrate CT with Science?

Case studies of teachers’ knowledge and skills in computational thinking and their enactment of a CT-rich curriculum within science classrooms.

Two teachers start with low-mid range of CT knowledge and skills at pre-. (7 out of 17 pts).

Both teachers were observed implementing CS in Science Module 4: Chemical reactions.

Teacher A:

Large KS-CT gains, 8 pts

Focus of instruction was on coding.

Engaged students coding to build a model.

Linked chemical equation to code.

Very little discussion of abstraction.

Emphasized coding the stages of reaction.

+Provided Intensive coding experience.

- lacked connection to “why model?”

​

​

Teacher B

Small KS-CT gains, 1 pt

Focus of instruction was on modeling.

Presented impetus for modeling.

Connected the model to real world.

Discussed abstractions in the model daily.

Emphasized concept of conservation of mass.

+ Shared epistemic ideas about models.

- lacked student preparation in CS/coding.

​

* Provided an exemplar for integrating CT without

teacher strengths in areas of CS and programming.

Cohort 3 Teachers (sw region)

KS change vs participation in PD (hours)

Cohort 3 Students (sw region)

KS change with 1, 2, and 3 modules experienced

Dosage response in KS-CT scores.

How teachers used modeling to support mechanistic reasoning

Ling Hsiao, Irene Lee, and Eric Klopfer Making Sense of Models. British Journal of Education Technology (2019)

​

During artifact based interviews, we observed teachers’ different patterns of StarLogo Nova usage that had varying outcomes with respect to mechanistic reasoning. Observations of the simulation led to Level 1 explanations (details that only focused on visible aspects of the phenomenon) and Level 2 explanations (addressing “how it is happening”). Observations of the graph of data combined with experimentation (running the model with different variable settings) led to Level 2 explanations. Only when observations of the simulation were combined with examining code did Level 3 explanations emerge (a partial or full mechanism based explanatory process explaining “why something happened”).

Teachers’ evolving understanding of CT

Analysis of teachers’ response to the question “What does it mean to you to be a computational thinker?” at the midpoint and end of the PD program shows a movement from CT as being about “reading chart & graphs” and “having good number sense” to “problem solving” and “building a model and using it to test scientific ideas.”

​

The most common conception of CT at the end of the PD was that it was related to “problem solving” (3 responses) and more nuanced understandings such as “abstracting and thinking in loops” and “thinking about processes” were evidenced in teachers who had taught only module 1.

​

One respondent focused on pedagogy and shifted from “demonstrating concepts with models” to “open-ended inquiry and problem solving” reflecting a shift in pedagogy afforded through CT integration.

Discussion:

• We found that our assessment tools provided a different angles on the teachers’ learning and experience. The KS-CT assessed how well the PD provided experiences that supported teacher learning. Observations provided rich data on teachers’ enactment of the curriculum, and artifact based interviews evidenced teachers’ thought processes and actions that promoted mechanistic reasoning as they made sense of a new model. KS-CT score changes were correlated with “dosage” for both teachers and students.
• We found that teacher learning (change in KS) did not necessarily correlate with implementation. Deep CS knowledge was not necessary to integrate CT with Science. Conversely, teachers who believed their students were not capable of programming did not implement the curriculum regardless of their gains in KS.
• Our hypothesis about teachers’ development of conceptions of CT through the PD were only partially supported by teachers’ reponses at midpoint and end of the PD .

Enriching mathematics and science with computational thinking:

Co-designing preschool activities with educators and parents

Ximena Dominguez¹, Shuchi Grover², and Phil Vahey³

​

¹ Digital Promise ² Looking Glass Ventures ³ SRI Education

​

Background & Approach

• Build a knowledge base
• which/how CT skills can be promoted in early childhood
• whether/how CT can be integrated with STEM

​

• Develop learning blueprints to guide the development of resources�
• Design and pilot test prototype activities, identifying design principles useful to future efforts integrating CT and mathematics/science in early childhood

​

• Focus on underserved communities

​

• Connect school and home learning

​

• Integrate hands-on experiences with digital apps

​

• Co-design process involving a multidisciplinary team & inclusion of teachers/families

​

​

​

Operationalizing CT

Guided by Grover & Pea (2013, 2018) and aligned to early learning standards; identified an initial set of skills & practices to explore/investigate with preschool children:

• Problem Decomposition

Resonated with teachers/parents; adults usually decomposed in current activities

• Algorithmic Thinking

Teachers identified looping (vs sequencing) as possible entry point

• Abstraction

Teachers and parents identified as a possible area that could be integrated easily

• Debugging

Identified as skill that could be embedded in most activities

​

CT & Math Integration - Synergies identified

CT & Science Integration

Sample Activities and Alignment

Field Study

• Sample:
• 5 classrooms implemented the curricular program (all children engaged in classroom activities and 2 families per classroom implemented the home–school program)
• 2 comparison classrooms (business as usual)
• Program includes:
• Teacher PD and workshops
• Hands on- activities for school and home
• 2 digital apps
• Data Sources:
• Surveys and interviews
• Observations in homes and classrooms
• One on one assessments

​

​

1-1 CT Assessment

• Flipbook style assessment
• flipbook
• manipulatives
• scoring sheet

​

• Time: ~20-30 minutes

​

​

​

Sample Assessment Items

Emerging Findings - Assessment

Sample:

• n= 57 preschool children (served in public preschool programs serving predominantly low-income communities)

Item Properties:

• The overall difficulty of the items was relatively high (most items had proportion correct less than .50)
• Discrimination values were adequate.

Reliability:

• The total summated score was found to be reliable, as estimated by Cronbach’s alpha: 0.72 and 0.78 at pre and post respectively.

​

​

Acknowledgements & Additional Information

• Contact:
• xdominguez@digitalpromise.org
• shuchig@cs.stanford.edu
• philip.vahey@sri.com
• NSF STEMForAll Video: https://stemforall2020.videohall.com/presentations/1884

​

NSF Award DRL-1639850

Leveraging computational thinking to teach elementary mathematics and science

Aman Yadav, Katie Rich, Christina Schwarz, & Rachel Larimore

�Michigan State University

This work is supported by the National Science Foundation under grant number 1738677. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Researcher Practitioner Partnership project with Oakland Schools, an intermediate school district serving Oakland County, Michigan in Metro Detroit

Goal: Integrate computational thinking in elementary mathematics and science instruction

​

Operationalizing CT: 4 CT practices

Teacher Professional Learning

• Introducing the 4 CT practices via unplugged and plugged activities
• Explore CT integration into math and science lesson
• Talk moves in the classroom for equitable CT
• Co-design lesson

​

Classroom Implementation of CT

Teacher used CT either explicitly or implicitly to frame a lesson at the beginning, prompt students to engage in the CT practice during the lesson, or invite them to reflect after the lesson.

Source: Rich, K. M., Yadav, A., & Larimore, R. (2020). Teacher implementation profiles for integrating computational thinking into elementary mathematics and science instruction. Education and Information Technologies. DOI: 10.1007/s10639-020-10115-5

How do teachers see the role of CT?

[CT] gives me a better framework for the thinking aspects. It just, it's much more focused on. Alright, now I really want to hear about your thinking here. {3rd grade teacher}

​

I've always been a very math-oriented person, but it [CT] gave me some language and some ... Kind of like a focus lens that I could take and actually teach with, and demonstrate for my students and then help them to apply. And I definitely feel that it [CT] strengthened my ability to actually teach problem solving skills, and to understand what the kids are thinking and how they're working through problems, and to kind of give them some of those tools myself and actually be able to capture something versus just trying to, you know, do lots of examples together. {5th grade teacher}

​

​

Teachers’ views on the impact of CT on students

Because often times they are like I can't do this. And when they know that they have these [CT] tools that they can use, these [CT] strategies that they can use, they're more confident and they're going to persevere longer. They're going to at least attempt it a little bit longer, because hey I can do this. I just need to look for this, I just need to look for that. {3rd grade teacher}

​

They had to code along on a map to get the [inaudible] like go a certain path to, like, hit certain destinations and collect renewable resources by avoiding the non renewable resources. And so they had to figure out how to get it to, like, turn different ways and u-turn around. And it was actually really cool, and it tied the science and math together, and they were really into it. {5th grade teacher}

​

​

​

​

Conclusion

Teachers see CT as a valuable thinking tool to engage their students in problem-solving within math and science instruction

Teachers view CT as enhancing their instruction

Starting with unplugged CT gave teachers more confidence to implement plugged CT

Students transferred CT vocabulary from unplugged activities to plugged activities

Contact

For questions contact Aman Yadav : ayadav[at]msu.edu