How to Make Progress in Computing Education

Education is the economic issue of the 21st century. Driven by global trade and a technologically connected, always-on global work force, countries understand they must innovate to succeed in the new business environment. A winning innovation policy is tricky to define, but it is clear it starts with education—and it starts early.

In the U.S., policymakers seem to have heard the message. There is a national urgency to improve K–12 education, and, in particular, ensure students have a strong grasp of science, technology, engineering, and mathematics (STEM) education. The Department of Education is pouring an unprecedented hundreds of billions of dollars into states to improve schools, help teachers, and support students. They want to know this money is helping. If you listen closely, you hear leaders from the Secretary of the Education to members of Congress talking about the need for “evidence-based” reforms. Where does this evidence come from? Largely, it comes from measurement tools developed by education researchers.

At the same time, the computing community sees a national urgency to reform K–12 computer science education. As computing transforms society for the digital age, students need to be able to think computationally about the world to succeed in life. How do students really learn rigorous computing concepts? We need research to tell us.

Computing is a relatively new discipline with a small education research base and limited assessments. Those responsible for making policy decisions in K–12 are interested in adopting curriculum in schools where you can assess how it is improving student learning. They are also interested in focusing resources on the “core” that students must know. Rigorous computing courses, if they exist, aren’t typically in the “core.” This leads to a chicken-and-egg problem for K–12 computer science, where you can’t really measure how students learn without putting it in schools, but schools aren’t interested in it until you can measure it.

We need to break this cycle and one aspect is improving the research base for computing education.

It isn’t enough to rely on general education research. We need research specific to computing—a form of domain-specific education research. General education research helps us understand (for example) how students learn and how schools best facilitate learning. Domain-specific education research answers questions that are unique to the domain. Mathematics education researchers help us determine what preschoolers ought to know so they succeed later at multidigit arithmetic (and how to remediate missing skills early, before they impede students’ progress). Physics education researchers know why students have trouble understanding velocity and acceleration, and they have identified the visualizations and activities that can enhance learning.

Computing education research is necessary for us to improve our teaching of computer science. Researchers in computing education can tell us how students understand parallel algorithms, what kind of visualizations help with understanding data structures (and how to use them), and how to measure understanding about computing that goes beyond any single language. Computing education researchers help us understand why students do not pursue computing as a career, and how to recruit, engage, and motivate more (and more diverse) students.

But we are the new kids on the school block. The National Council of Teachers of Mathematics was founded in 1920. The National Association for Research in Science Teaching started in 1928. In comparison, ACM’s Special Interest Group in CS Education (SIGCSE) is only 40 years old, and ACM started the Computer Science Teachers Association (CSTA) six years ago. SIGCSE’s research conference, International Computing Education Research (ICER) Workshop, is only in its fifth year.

Being relatively new puts us at a disadvantage when seeking competitive funding. Imagine that you are seeking funding in a general education research program, in competition with proposals in mathematics education and science education.

• Which proposals have great evaluation plans that will demonstrate the value for the investment? Mathematics and science education can point to scores of reliable, valid measures of learning that they can use. In computing education, there is no reliable, valid measure of introductory learning that isn’t tied to a specific programming language. Overall, there are few standard measures of computing learning.
• Which proposals will lead to more students achieving learning objectives, identified by standards at the state or national level? Such standards and objectives exist for mathematics and science, but rarely for computer science in the U.S.

Some countries do fund research in computing education. There are strong research programs in computing education in Israel (Technion and Open University), England (at the University of Kent at Canterbury, for example), Germany (Free University Berlin), Sweden (Uppsala University), and Finland (at University of Joensuu). These research programs are investments in IT competitiveness in those countries.

Back to Top

The State of Computing Education Research Funding in the U.S.

How about in the U.S.? Things are much more dismal, particularly for the K–12 level. The National Science Foundation (NSF) is primarily responsible for funding education research,^a which comes two directorates: Computer and Information Sciences and Engineering (CISE) and Education and Human Resources (EHR). We examine CISE first.

CISE has had two programs—CISE Pathways to Revitalized Undergraduate Computing Education (CPATH) and Broadening Participation in Computing (BPC)—with a focus on education. However, as of this writing CISE announced that it is combining these programs into a broader program. This new vision would look at the entire pipeline but with special focus in two areas:

• moving earlier into the pipeline with specific engagements in middle/ high school to bring computational thinking/computer science concepts into this space; and
• widening the program to be inclusive for all populations, built around a theme that “computing is for everyone.”

It would also add a specific education research component that would seek to build education research capacity at the university level and to provide a greater understanding of how children come to understand computing concepts. No one knows exactly what this new program will look like until the solicitation comes out, which CISE is saying will happen in the summer of 2010. It is expected the new program will be funded at about $20 million, which is similar to the combined amount for CPATH and BPC.

These are likely positive steps toward addressing clear gaps in the field. But it will likely be a series of small steps until the community can start leveraging other parts of NSF. Compared to the relatively small CISE budget for education, EHR has over $850 million for education research, which is where we need to turn our attention. Not all of this funding goes into education research, but in looking where Congress is investing federal education research money, it is clear they are looking to EHR for those answers. EHR funds both higher education and K–12 research programs through various programs.

The program that probably does the most for higher-education computer science is the Course, Curriculum, and Laboratory Improvement (CCLI) program. It seeks to improve undergraduate education across STEM through proposals on how interventions work and how they get disseminated, and funds the development of new assessment techniques and instruments. It funds applied research that informs interventions, and doesn’t fund computing education research that helps us develop new theory about how people come to understand computing.

The state of computing education research and teacher support at the K–12 level is more complicated. There are several relevant EHR funding programs. ACM staff analyzed abstract summaries from NSF’s “Fast-lane” within EHR to better understand where computer science education research is funded or where computer science teacher support existed. The scope of EHR programs was limited to: funded proposals that had a significant K–12 focus, or those that prepared or continually trained K–12 teachers. Abstracts are only a brief representation of the plan, so the analysis tended to be more inclusive—”close” was good enough. However, the analysis specifically excluded grants that were primarily focused on getting computing into the classroom or getting teachers prepared to use computing in the classroom.

The results of the analysis appear in the table here. Of the 1,839 proposals funded across seven programs, only 67 (4%) had an explicit computer science component. Our analysis of abstracts could not tell us which of these projects had any kind of research component, nor where the research informed our understanding of learning computing specifically.

---

It isn’t enough to rely on general education research. We need research specific to computing—a form of domain-specific education research.

---

Regardless of the limitation of the analysis, it is clear—there is far too little computing research or teacher support being done by the key agency charged by the federal government for understanding how to improve STEM education.

Back to Top

Making Progress in Computing Education

Funding is important. Funding allows researchers to make progress on problems that are a priority. Funding is recognition that a particular discipline or strategy is worthwhile. Funding can create a virtuous circle, where funded work attracts more funded work. Lack of funding creates a vicious circle, when the lack of theory and of assessment prevents other projects from being funded.

The computing education research is concerned with how to sustain interest and progress in the research community. Few of the U.S.-based presenters at the International Computing Education Research Workshop have NSF funding to work in computing education research. Those that have received NSF funding do the computing education research component on the side. Few Ph.D. students focus on computing education, and those that do focus on this area rarely obtain faculty slots in research institutions. Working in an area without explicit funding programs is dangerous for an untenured assistant professor at a research institution.

Funding is particularly important to bootstrap a field. Computing education research seems to be in a vicious cycle. As a community we need to take some basic steps to break the cycle:

• Learn what NSF programs are available and aggressively go after funding. NSF CCLI Program Officers regularly offer SIGCSE workshops walking through the various NSF programs that are available for education research to catalyze proposals.
• Sit on NSF review panels when asked, particularly in EHR. There should be a computing voice at these review panels. The best way to learn what gets funded (and how to write a fundable proposal) is to sit on these panels.
• Encourage fundamental research in computing education. As teachers of computing, we want to know what language, IDE, book, and curriculum works best in our classroom. We also need the theory that helps us make these choices, and the assessment that gives us data on what works best. We want students to start successfully and to develop expertise and skill.
• Look for opportunities to work with other domain-specific education groups. Mathematics education, for example, has funding sources, and computing education research could grow in combined projects.
• We must stand together. Reform proposals supported by many institutions carry weight. Shared goals and strategies lead to proposals that reviewers can support.

Computing education research is an important investment in innovation and competitiveness. If the U.S. wants to remain the world leader in computing, it ought to be making that investment and the community needs to aggressively go after funding. Other countries are making that investment. Growing computing education research in the U.S. would improve teaching and learning in computing nationally and inform the research community internationally.

Back to Top

Back to Top

Back to Top

Back to top