Education Research

What is Learning Engineering?

What is Learning Engineering3-01 (1).jpg

Although the term “learning engineering” was coined more than 50 years ago, it is still an emerging field of study. Learning engineering is a process that applies the learning sciences using engineering design methodologies and data-informed decision-making to support learners and learning. In the past 50 years, discoveries about how people learn have influenced how people teach and learn, and there are many new findings still to be applied.

While engineering is the application of creativity and science to solve problems, learning engineering is the creative application of learning sciences and engineering principles to solve problems for learners and learning. Learning engineers create and iteratively improve the conditions and experiences for learning.

Learning engineering applies the science of learning. The sciences of learning are informed by what we know about how the brain works. According to a report published by the Massachusetts Institute of Technology, the learning sciences include

  • neuroscience, a branch of biology that studies the nervous system, neurons, and the behavior of the brain;

  • cognitive psychology, the study of the human mind as understood through observable behavior; and

  • education research, which is focused on models and interventions at the classroom level and above.

Learning engineering is about people. Human-centered design focuses on the learner. Data informs design decisions. The engineering process guides design choices that promote robust student learning. According to a paper published in Cognitive Science, learning scientists say that learning is robust when the student retains information over time, can apply the learning to new situations, or can draw on the learning to accelerate learning in other areas.

Learning engineering is a team sport. The problems that learning engineers try to solve often are too big to be handled by the skills of any one person, so education agencies, institutions, and organizations that support schools compile multidisciplinary learning engineering teams. The composition of the teams varies based on the problem to be solved.

Learning engineering is data driven. Unlike traditional instructional design, curriculum development, or lesson planning, learning engineering emphasizes data use to inform an iterative design, development, and improvement process. This is different from having researchers evaluate the efficacy of a curriculum or methodology after it is fully developed and deployed. Continuous access to data is integral to the process and products of learning engineering.

A Practical Application of Learning Engineering

Duolingo, the award-winning language learning app, is an example of a learning environment and set of learning experiences developed using learning engineering. One of the ways in which Duolingo is innovative is in its use of game dynamics to motivate learners.

“One of the tricks is figuring out how to marry things that are pedagogically sound with things that are reinforcing motivators,” said Burr Settles, who leads Duolingo’s research group.

To figure these things out they use engineering techniques rooted in scientific research.

“We run controlled experiments for almost everything we change,” he said.

The spacing effect, a human learning characteristic, was first discovered over a century ago. This term refers to the finding that people are more likely to remember something if they spread out the learning over time, rather than cramming it in all at once. More recently, learning scientists discovered the lag effect: people commit information to long-term memory better if the time between study or practice increases gradually. Still other research showed that the best time for people to practice is right before they’re about to forget something.

Settles worked with Brendan Meeder, a former Duolingo employee who now works at Uber’s Advanced Technologies Group, to develop a “trainable spaced repetition model for language learning.”

Their model, called half-life regression (HLR), predicts when someone will forget a word. HLR takes in data about a student’s practice history (when the student practiced and whether or not the student remembered correctly) and data that indicate how difficult a word is to memorize, compared to other words. Given data about the times a learner practiced a word, the app predicts the best time to practice next.

2019 Conference on Learning Engineering

Join QIP at the IEEE IC Industry Consortium on Learning Engineering (ICICLE) 2019 Conference on Learning Engineering to be held May 20 to 23, 2019, at George Mason University Arlington Campus, Arlington, Virginia. The conference will bring together professionals from industry, academia, and government to contribute to the development of this emerging field, share ideas with key stakeholders, and help further define the discipline of learning engineering.

QIP is one of 71 sponsoring organizations of ICICLE, an organization aimed at advancing learning engineering as a professional practice and academic discipline. QIP is supporting the conference by producing a video.

jimg.jpeg

Jim Goodell (@jgoodell2) is Senior Analyst at QIP. He works on connections between education sciences, policy, practice, and personalized/optimized learning. Learn more about Jim here.

Big Data in Education: Researchers’ Responsibilities

bigdataimage.png

While big data’s growing influence has impacted our lives across a spectrum of issues, it also has created many questions and concerns, particularly among education researchers.

Big data allows researchers to uncover patterns in data that might be otherwise invisible. This has led to several powerful advances, such as better treatments for disease, improvements in agriculture, and more timely and effective responses to natural disasters. The benefits of big data have even been highlighted in popular media, such as in the movie Moneyball, which dramatizes how the pioneering use of large datasets helped a general manager assemble a winning baseball team.

But the rise of big data has also prompted many to note its potential negative consequences. Within education, researchers have identified not only benefits to using big data, but also legitimate concerns. As they do with all data, education researchers have a responsibility to focus on both the integrity of their research using big data and on clear communications about this research to the public. Further, their communications with the public should focus not just on the research itself and its useful possibilities, but also on the precautions they are taking to ensure that the rise of big data does not negatively affect the education community.

Big data, defined

The term “big data” refers to very large and complex datasets—those datasets that have been described as “defying traditional data-processing applications” (National Academy of Education, 2017). Modern technologies allow us to capture information in previously unforeseen ways and transform it into digital data. This has resulted in datasets that are much larger and more complicated than anything seen before. From a research standpoint, big data changes data collection from an often lengthy and painstaking process to one that can happen nearly automatically, given the right connections to sources.

Big data in education: improving teaching and learning

Big data in education tends to fall into two major categories: administrative data and learning process data. Combining digital data from these two areas in innovative ways can allow researchers to identify patterns or correlations that may otherwise go unnoticed.

  • Administrative data can be demographic, behavioral, and achievement data and may include items such as attendance records, transcripts, and test scores.
  • Learning process data are continuous records of students’ behaviors and may include online assessments, keystrokes, or time latencies (e.g., the time it takes a student to respond to a question).

Innovative data analyses can lead to useful solutions to problems in schools and classrooms, uncover potential inequities in learning opportunities, and zero in on students’ needs in ways that reveal how to personalize learning more effectively. The overarching goal of this data collection and analysis is to expand possibilities for teaching and learning—including how to meet individual students’ needs.

Big data in education: legitimate concerns

Education researchers have raised some legitimate concerns about big data. While they recognize that big data has many exciting possibilities, researchers have also identified some potential problems with its use—or misuse. These concerns tend to fall into three main categories: misinterpretation, inappropriate use, and data privacy and security.

  • Misinterpretation concerns center on the possibility that studies using big data may be misunderstood by readers—especially if the studies are distilled or simplified before reaching the public—and that these misinterpretations could lead to inaccurate decisionmaking.
  • Inappropriate use concerns suggest that the public nature and accessibility of some big data may lead to people using the data in ways that were not intended and that defy accepted research standards.
  • Data privacy and security concerns are based on concerns that individuals’ personal information may not be properly protected, which could lead to data breaches or other inadvertent disclosures of private information.

As the education field continues to move toward greater use of big data, each of these issues should be specifically and consistently addressed. This can be accomplished through strong data governance, research standards, and other precautionary measures.

Researchers’ responsibilities: communication with the public

Education researchers must think not just about the research on big data, but also about how the public is receiving and reacting to this research. Public discussion of big data is frequently negative and inaccurate. Unlike the measured considerations of big data presented in academic articles, much of the communication about education-related big data to the public has encouraged skepticism and fear. It is not surprising that many parents and other stakeholders have developed negative views, given the frequent headlines that tout the “big dangers” of big data. The public less frequently encounters news that describes the potentially positive aspects of this education information or the clear standards that are in place to protect the privacy of personal information.

At the same time, researchers should work to ensure that members of the education community understand the legitimate concerns about big data and what we can all do to avoid or mitigate problems that may arise from misinterpretation, inappropriate use, and data privacy and security issues. Walking the fine line between explaining the intricacies of this difficult topic and communicating concisely and clearly is something education researchers must strive to master.

Big data is indeed a problem if it is used ineffectively, inappropriately, or by individuals without a requisite level of comprehension of the complexities of the subject. But that is true of all research data. Data, in various forms, can reveal that something has happened, that a phenomenon exists, or that variables appear to have a relationship, but data cannot on their own reveal why. It is the responsibility of researchers—especially those in the public sphere—to provide the lenses that make research relevant and comprehensible to varied audiences, from parents and teachers to administrators and elected officials.

It is important for education researchers to make clear that they are using the same stringent research standards for big data analysis that they have adhered to with previous types of data. Additionally, they must communicate to the public that they are regularly discussing the potential hazards of big data and routinely updating methodologies and security protocols as projects and analyses become increasingly complex. The clearest path to public trust in the research process is via straightforward and detailed communication.

Bridget Thomas (@DrBridgeQIP) is Senior Education Researcher at QIP and Adjunct Professor at George Mason University. Her work focuses on early childhood policy and translating research for multiple audiences.