Why Data Science Education?
Data science, artificial intelligence, machine-learning, and related disciplines are rapidly evolving technology fields that carry significant implications for our students, regardless of their career choice. More importantly, navigating daily life in the 21st century now entails several encounters with data, both in personal and professional settings. Creating equal access to data literacy education will be a critical challenge for our economy, our workforce, and our democracy.
IES is committed to supporting the use of evidence-based practices in the design and implementation of educational programs in these areas—while acknowledging that, as a field, we continue to learn together what works best, for whom, and under what conditions.
Intended for PK–12 leaders and educators, this site provides a checklist for guiding design, implementation, and evidence gathering in new programs for data science education and related fields. The site also catalogs a variety of resources, including those created outside of IES, that may be useful references for first implementing data science and data literacy education programming. Keep in mind that while some practices are based on rigorous evidence, many are rooted in theory but have yet to be tested. While some steps may seem obvious, even the most practiced experts benefit from checklists and reminders. As we use and share today's best available evidence on data science and data literacy learning, we must continue to build stronger, more rigorous evidence about "what works" going forward.
Other Resources
Getting Started: Building Common Language
Technology-based fields like data science, artificial intelligence, and machine-learning are changing quickly—including their definitions. These terms can be confusing and interwoven, and are often used to reference different applications interchangeably. While the Department does not endorse any official definitions, we offer an imperfect framework below as a reference to help educators and education leaders decipher literature and program descriptions in this new space. Further clarifying these terms with yourself and others may help in goal-setting, navigating program selection, and making other design choices with common understanding.
Data Science: an inter-disciplinary field often combining statistics, computer code or manipulation, and domain-specific knowledge. Use-cases can include data analysis, data storage or management, data visualization, and data ethics.
Artificial Intelligence: a very broad term referring to any program, and the development thereof, that enables computers to automate tasks resembling human processes. Use-cases can include visual recognition, image categorization, speech processing, smart voice assistants, and language translation. Many artificial intelligence methods may be considered a subfield of data science, since these methods use large or complex data in the process of achieving automation.
Machine-Learning: a subfield of artificial intelligence that leverages large amounts of data and statistical models to improve an algorithm automatically. Current use-cases can include recommendation algorithms (music, TV, products), autonomous vehicle programs, and most other enterprise AI.
Neural Networks: an increasingly popular class of machine-learning algorithms, which uses many individual decision nodes to learn from data to make predictions. Deep Learning, another increasingly popular algorithmic approach, describes a neural network with many layers.
When building a common understanding of these terms with stakeholders, aiming to teach all these domains and techniques by high school graduation may likely be overwhelming. Instead, PK–12 leaders and educators may choose to focus on identifying more universal data acumen as a foundation for all students on which to build, regardless of their academic or career interests, and consider the ways in which new educational pathways can enable progression to more distinct, specialized fields
Using Evidence in Data Science Education Programs
This site is organized around the idea of using evidence to inform the design, implementation, and continuous improvement of data science education programs. While there are many models of evidence-based implementation and improvement, they generally follow similar steps. We use the cycle of evidence use and evidence building outlined in the Department's "Strengthening the Effectiveness of ESEA Investments".
Begin by identifying the specific needs of your students and community, including state requirements, locally demanded skill sets, and opportunities or gaps to educational advancement. We recommend taking the steps below, including:
- Consult existing state standards, post-secondary requirements, and student pathway data:
- Consult current state or local standards in math, science, and other K–12 subjects, and identify where data-related content can service or enrich existing requirements.
- Consult any forthcoming state or local standards in math, science, and other K–12 subjects, and identify where data-related content can service or enrich new requirements.
- Consult admission requirements for local institutes of higher education (IHEs), focusing on where students most often matriculate:
- Review math, science, or other credit requirements for college admissions.
- Contact admissions officers at local IHEs.
- Contact your school or district's guidance counselors.
- Review school or district assessment data to identify areas where students are performing below grade level, especially in STEM subjects like math or science, to identify points of intervention.
- Meaningfully engage stakeholders:
- If using ARP funds, have you talked to the following Stakeholders as required by the American Rescue Plan using focus groups, surveys, advisory boards or other means?
- Students, families, and caregivers
- Teachers, other educators, school staff, principals, and other school leaders
- Advocacy organizations, including those representing the interests of children with disabilities, English learners, children experiencing homelessness, children in foster care, migratory students, and children are involved with the juvenile justice system
- Other community based organizations, such as youth development organizations, STEM Learning Ecosystems, cradle to career networks like Strive Together affiliates and other place-based initiatives.
- School and district administrators, including superintendents
- Charter school leaders, if applicable
- Education organizations and advocacy groups, including teacher and staff unions
- Community and elected leaders, including tribal leaders, school boards, and leaders representing business and industry
- Other potential partners, including other community-based organizations, university faculty and staff, and faith communities
- If using ARP funds, have you talked to the following Stakeholders as required by the American Rescue Plan using focus groups, surveys, advisory boards or other means?
- Seek systematic alignment with postsecondary pathways:
- Review postsecondary pathways available to your students, using any data on student matriculation, input from guidance counselors, and other stakeholder engagement. Examples may include:
- 4-year Colleges or Universities
- 2-year Community Colleges
- Apprenticeships or other Work-Based Learning Programs
- Local Employers (small and large)
- Federal, state, or local government, including the Armed Forces
- Create a working group, advisory board, or otherwise consult a diverse sample of representatives from each pathway to identify gaps, make them aware of your proposed changes, and seek aligned credit opportunities (e.g. recognized credentials, dual-enrollment, coursework credit, etc.). Examples may include:
- Department leaders in Statistics, Computer Science, and/or Mathematics
- Provost or other academic directors
- Admissions officers and/or directors
- Student academic advisors
- Employer Human Resources officers and/or directors
- Nonprofit or Educational Program directors
- Community programs (e. g. AmeriCorps, TeachForAmerica, etc.)
- Collaboratively identify opportunities to create more equitable and modern post-secondary pathways through the intervention or pilot
- Review postsecondary pathways available to your students, using any data on student matriculation, input from guidance counselors, and other stakeholder engagement. Examples may include:
- Consider piloting first, and design with iterative goals in mind:
- Consider testing your course or other intervention with a small number of students — testing first will help identify remaining gaps, unanticipated issues, or other resources that may be needed.
- Consider your pilot as a mechanism to demonstrate promise and build confidence with educators, parents, and other stakeholders, and as an opportunity to improve for future school years.
- Design your pilot with specific goals in mind — identify an attribute or set of attributes to test, demonstrate your intervention with either a representative sample or a targeted of your student population, or focus on identifying challenges for a particular issue of interest that will aid with future improvement or scaling.
- Prioritize relevant outcomes to focus your program over time:
- List all possible outcome goals identified through stakeholder engagement, collaboration with post-secondary programs, and student data review.
- Consider other potential outcomes that have been trialed in comparable data science programs
- Prioritize and select one or more outcomes for focus of your intervention based on local priorities and locally in-demand skillsets.
Additional Resources
Well-designed and well-implemented data science and data literacy instructional programs have the potential to support a variety of student outcomes. While acknowledging the field is new, you should design or select programs that, whenever possible, use evidence-based practices in program design and that draw upon existing evidence-based practice and resources in other domains. We discuss strategies and resources both below.
For more information about the Department's definition of evidence-based practices, visit What is An Evidence-Based Practice?
Guidelines for selecting interventions that Demonstrate a Rationale:
- Check if the program used evidence-based practices or theory for design:
- Does the program build on existing peer-reviewed education research?
- Does the program have a clear and cohesive pedagogical approach for instruction?
- Check if the program uses age-appropriate and subject-appropriate technology tools:
- Does the program use a software that builds real-world data acumen? Is it authentic?
- Does the tool or software appropriately challenge students? Does it present accessible learning opportunities, rather than overwhelm them?
- Does the program use technology and datasets that align to required content, or builds upon existing knowledge, in the educator's subject (e.g., math, science, social studies)?
Making Intentional Choices: Data Science Software
What software do "real" data scientists use? A professional data scientist is often comfortable with many different software tools — should students be as well, or should technology barriers be minimized? How should an educator pick a technology tool for the classroom? Early research suggests exposure and practice with an age-appropriate data science software can be beneficial for building student confidence. Educators may consider using tinker-based exploration tools (CODAP, Tuva, Cognimates), popular data science software (R / RStudio, Python, SAS, Stata, Tableau PowerBI), or even spreadsheets (Microsoft Excel, Google Sheets). Some programs have also created a progression of multiple software, introducing students to different tools across a learning trajectory, beginning with education-focused software. Regardless of approach, building a common understanding of the terms, differences between tools, and their real-world and educational use-cases should be an intentional process involving multiple stakeholders when designing your program or intervention.
- Check if the program offers professional development resources:
- Does the program offer synchronous or asynchronous learning resources for educators to implement the program as designed?
- Does the program offer continuing support after initial professional development programming?
- If the program has been evaluated before, check the following:
- Was the student population similar to the students in your school or district?
- Did the pilot use a validated assessment or measure?
- Did the evaluation use a validated assessment or measurement tool?
- If the program has not been evaluated before, check the following:
- Have you prepared to build evidence associated with the implementation of the program, including identifying a validated assessment or measure for an outcome relevant to your circumstances?
Explore existing evidence-based practices in data science education and related education literature:
Charles A. Dana Center Data Science Course Framework, the IDSSP Framework for Introductory Data Science for a list of suggested best-practices in high school data science programs, or the NCTM-ASA GAISE II Report (see PDF below) for suggested guidelines in statistics and data education.
GAISEIIPreK-12_Full (9.23 MB)- Look in the What Works Clearinghouse (WWC) for a list of programs that have evidence aligned to the outcomes you designing for. Helpful categories in the WWC include Literacy, Math, Science, Early Childhood, Behavior and High School graduation. (Not all these programs have been designed for use in data literacy programs, but they could be strong places to start in considering how to incorporate new content).
- Consult other resources that reviewed the evidence for related learning science domains including:
- Project Based Learning — general overview, in science education, in math education, and in social studies education.
Community Based Learning — general overview (see PDF below), early evidence for effects on learning outcomes (civic engagement, capstone projects), and assessment approaches.
ED580907 (43.97 KB)Computational Thinking — potential strategies, professional development best practices, suggestions for early math (see PDF below) and science integration, and in the context of problem-solving with datasets across subjects
Microsoft Word - 7733 Naidu (760.54 KB)
Explore pilot data literacy programs from peer schools or districts:
- Consult subject specialists (math, science, social studies, etc.) at your state education agency (SEA) to learn of any nearby data science or data literacy pilots, evaluation studies, or new instructional programs.
- Consult nearby educators or education leaders at neighboring districts.
- Explore programs in states and districts like mine, and consider the following questions:
- How does their program relate to mine? Is the student community similar or different?
- What steps has their program taken to seek local stakeholder engagement?
- What steps has their program taken to collect data, evaluations, or other practice-based learnings that may be useful for mine?
Explore Example Programs!
Planning for implementation requires dedicated time and capacity. Note there are several suggested factors, based on the italicized sections below, that require planning. Doing more now will save time later and allow your program to contribute to broader collective learning.
Given the early stage of research in data literacy education, it is critical that inaugural programs have a plan to build evidence about what works best, for whom, and under what conditions. This means your implementation should include plans for tracking challenge and lessons learned, initial student outcomes, and other useful data to improve your program over time.
Build a logic model for your approach
- Use your favorite logic model design or use REL Pacific's Logic Model tool or the template provided here (See Figure 2).
- Ensure that your inputs and outputs included specific, quantitative measures, related to the data that you intend and are able to collect.
Logic Model: Project Name | |||||
Problem Statement: | Goal (s): | ||||
Inputs | Activities | Outputs | Short-Term Outcomes (Date to Date) | Short-Term Outcomes (Date to Date) | Short-Term Outcomes (Date to Date) |
|
Figure 2: Logic Model Template
Why build a logic model?
The development of a logic model is important for meeting evidence tiers designated under the Every Student Succeeds Act (ESSA) of 2015. In addition to helping specify the goals and outcomes of your intervention, a documented logic model can help a program meet the minimum standard of evidence: Tier 4, Demonstrates a Rationale. Meeting evidence tiers will enable a program to be eligible for grant opportunities with evidence requirements, and help the field collectively gather more evidence on data science education. If local staff and technical capacity allows, meeting higher evidence tiers is always a good idea. You can read more about ESSA evidence tiers here. (PDF: 652 KB)
Develop detailed implementation plans, timelines and workplans
- See example guides for STEM programming such as those created by the Arizona STEM Network, the State of Ohio, and the State of Indiana.
- If you are planning a summer program, use the Summer Learning Toolkit which contains examples, templates and more to plan your program. You can also see how the State of Texas has adapted the toolkit to fit their needs.
- If you are planning an expanded or out-of-school learning program, start with the 21st Century Community Learning Center's New Director's Toolkit
Ensure sufficient time and resources for educator professional development
- Consult with your schools' educators and other staff to create awareness for data-related education across subjects, including but not limited to: mathematics and statistics, biological and physical sciences, and social studies.
- Identity enthusiastic candidates, seeking representation within the chosen team that can encourage all students, including those historically underrepresented in STEM fields, to participate.
- Allow time for professional development opportunities for interested educators before introducing courses or other intensive programs.
- Consider selecting curriculum-aligned professional development programming for candidate educators to ensure relevance.
- Reach out to local Institutes of Higher Education (IHEs) Data Science, Statistics, or Computer Science departments to facilitate partnerships for professional development.
- Reach out to local employers, local governments, or other organizations using data analytics to scope the possibility of short-term "externships" for educators to gain practice-based professional development.
Ensure sufficient time and resources for acquiring necessary hardware and software for your intervention
- Have you developed a plan, in collaboration with a technology officer or other procurement staff, to ensure in-school access to relevant hardware, software, and internet connectivity?
- Have you a developed a plan, in collaboration with a technology officer or other procurement manager, to ensure proper maintenance, replacement accessories, and staggered refresh cycles for equipment?
- Have you surveyed or otherwise assessed access to relevant hardware, software, and internet connectivity for students at home or elsewhere?
- Have you created and documented planned adaptations of your instructional intervention, relative to technology constraints?
Determine if your American Recovery Plan (ARP) Elementary and Secondary School Relief Fund (ESSER) allocations should support investments in hardware, software, cloud storage, or internet connectivity, and consider long-term agreements with providers to extend investments. See the Department's COVID Handbook Series (see PDF below).
ED COVID-19 Handbook, Volume 2 (PDF) (5.09 MB)- Determine if you are eligible for the Federal Communications Commission (FCC) E-Rate Program for Category 1 (telecommunications, telecommunications services and Internet access) or Category 2 (internal connections, basic maintenance of internal connections, and managed internal broadband services) discounts.
Ensure sufficient time for student recruitment and family engagement
- Do you have partnerships with your school board, parent-teacher associations, or other parent organizations?
- Consider holding information sessions to preview these programmatic changes with parents and families, outlining opportunities for students.
- Consider holding consultation sessions with guidance counselors and other staff to ensure alignment with post-graduation planning services.
Develop plans for continuous improvement, data collection and evaluation
- Check if your district has a research and/or evaluation team, and if so, engage them early to help navigate district protocol for pilot studies or new programming and overall design.
- Check if your district has a research board or IRB, and if so, engage those offices early to understand the process and timeline for program review.
- Many districts are often constrained in supporting interventions which do not focus on math or literacy; while fit to support many subjects, data science programs are frequently articulated as mathematics courses.
Read and use the templates provided in the continuous improvement toolkit (see PDF below) that IES and REL Northeast and Island published for schools and districts to use continuous improvement in education.
- Review A Program Director's Guide to Evaluating STEM Education Programs, a National Science Foundation (NSF) supported TA tool.
Additional Technology Procurement Resources
A well-run implementation process can help build confidence for future programs and contribute to filling research gaps. A "well-run" process does not necessarily entail improved student success, nor should that be a guaranteed outcome. Rather, clear communication, pre-designed support networks, and transparency can help build confidence among educators, parents, and other stakeholders—critical to long-term success for any intervention, especially when using new technology.
Implement your plans
- Have you designed a process for troubleshooting hardware, software, and internet issues that may arise during implementation?
- Have you ensured there is active support and mentorship for educators trialing new content, including mentors or other PD staff?
- Have you created weekly or monthly communication materials to relevant stakeholders, including families or other primary care-givers?
Collecting data on implementation and outcomes
- Have you established your pre and post assessment tools?
- Have you designed surveys to understand student, teacher, and parent experiences?
- Have you involved students in survey design? It is a great way to get to know what matters most and is relevant to them.
- Have you designed a process and obtained necessary approvals to collect and/or merge data of interest, including assessment data, survey data, or descriptive statistics on students, teachers, or other stakeholders?
- Have you designed a process to monitor survey and/or assessment data collection and its accuracy during the program?
Additional Resources on Data Collection:
Examining and reviewing data from your pilot study or programmatic intervention will help identify potential improvements for future programs—including gaps that may be missed by anecdotal or personal experience. Moreover, honest examination of successes and failures may help other educators, schools, and educational agencies in spreading equal opportunity to data literacy education.
Review data and refine your approach
- Read and use the checklists and templates in the continuous improvement toolkit (see PDF below) that IES and REL Northeast and Island published for schools and districts to use continuous improvement in education.
- At minimum, write out your plan for reviewing data addressing:
- How frequently?
- With whom?
- What questions are we trying to answer?
- What will we do with what we learn?
Begin continuous improvement cycle again as appropriate
- Using your logic model and continuous improvement plan, refine your approach and begin the continuous improvement cycle again starting with Step 3 — Plan for Implementation
Finally, the Department, as part of the Education General Administrative Regulations (EDGAR; see 34 CFR 77.1), further defines "well-designed and well-implemented" experimental, quasi-experimental, and correlational studies as part of its definitions of strong, moderate, and promising evidence.
Note: Some ESEA programs, including school improvement programs described in Section 6303 of the ESEA, allow evidence-based programs at the strong, moderate, or promising levels only. For more information, see Section 8101(21)(B) of the ESEA.