Can They Do It in the Real World? Designing for Transfer of Learning

Why do we build instruction? Seems like a “duh” kind of question for those of us who build instructional materials for a living, but it actually deserves more thought than you might expect.

Editor’s Note: Parts of this article may not format well on smartphones and smaller mobile devices. We recommend viewing on larger screens.

There are many reasons for building instruction, but one of the more critical reasons is to impact what people can do in the real world. Students are not taught math so they can solve problems at the end of each chapter. They are taught math so they can handle purchases, understand mortgages and other financing, and complete mathematical tasks required by their jobs.

Management training is not provided so that managers can perform well during instruction. The desired outcome is good performance on the job. As developers of instruction, we often expect learning to be applied in numerous situations, but our expectations and reality are quite different.

Consider the following scenarios:

• Teachers teach how to do percentages. Their students can solve percentages problems in class but they can’t determine the final price of a sale item in the store, given the discount amount.
• A training group provides instruction on legally defensible hiring practices but afterwards, managers’ actual interviewing practices contain critical legal inadequacies. In addition, many of these managers don’t apply these concepts to the wording of job descriptions.

What we have in these (and similar) situations is a problem with transfer of learning. Transfer of learning means that something learned in one situation can be applied in another. In these two examples, transfer has not occurred to the degree needed.

What is transfer?

Unfortunately, situations like these aren’t unusual. Research shows that transfer doesn’t happen just because instruction is provided. Rather, the preponderance of evidence, from Thorndike’s studies in the early 20^th century through more recent research, shows that the desired transfer from instruction to application happens relatively infrequently. People often do not understand how what they know can be applied in various situations because they often don’t make the connections we expect them to make. This is undeniably a problem since transfer is the primary goal of instruction. Although studies have shown that certain instructional strategies can dramatically improve transfer, another problem is that many e-Learning applications aren’t designed to take advantage of them.

Why don’t we regularly design for transfer, if transfer is so critical to our success and the success of our stakeholders? Here are a few reasons. Many folks who design instruction think transfer takes care of itself (it certainly doesn’t). Many don’t know how to create online instruction that makes transfer more likely. Many realize transfer is important, but design for relatively low levels of transfer (what I will be calling near(er) transfer) when higher levels (what I will be calling far(ther) transfer) are needed.

Back to my initial question about why we design instruction — if we don’t design for transfer, why design instruction at all? Let’s look at some of the basics that help us understand how transfer works and then consider how this information can be used to build e-Learning that creates more transferable skills and knowledge.

”Learning” vs. transfer

The student who can solve the percentage problems at the end of the percentages chapter in his math text has “learned” how to do percentages. A manager who can select the unlawful phrases in a simulated interview has “learned” which phrases are unlawful. What does it mean, then, if they cannot solve percentage problems in a store or build a lawfully worded real-life job interview? It means that there is a critical difference between learning during instruction and application of that learning in the real world. Consider this: It takes far more skill to actually do the interview than just discriminate between good and bad wording. The bottom line is that real life application is almost always much more complex than the instruction. Unless we take into account the inevitable complexities of real-world application, the instruction we design is likely to be far too simplistic for transfer to occur to the degree needed. And that means the instruction we build is less effective (often by far) than is needed to justify the time and expense.

Near(er) vs. far(ther) transfer

You may have noticed that I have referred a few times to the degree of transfer needed. Degree of transfer refers to whether application in the real world is a lot like the instructional context or quite different from it. For example, Mary takes a class on how to handle an angry customer. She learns a three-step process in the course that includes very careful listening, repeating back to the customer what she has heard, and then suggesting solutions. In the course, learners practice this process over and over in a variety of cases in order to get practice with the wide range of situations where this may be useful. If she goes back to the job and is able to use the process exactly as she was taught, the degree of transfer isn’t terribly far. Being able to adapt this process when the exact sequence doesn’t make sense means the degree of transfer is far(ther), because she is using the skill in a new context.

This concept is called near and far transfer. That’s a bit of a misnomer because it’s a continuum from near(er) to far(ther) transfer, with potentially higher degrees of transfer along the way. Near(er) transfer is transfer between very similar contexts. For example, a technician is trained to replace the hard drive on a computer with a similar or the same type of hard drive. Or a clerk receives training to use and troubleshoot the mail sorting equipment and then performs these exact same skills on the job. Near(er) transfer is generally what is needed for tasks that are routine and consistent.

Far(ther) transfer refers to learning applied in real life situations that are somewhat to greatly different than the learning contexts. This is most needed for tasks that are executed differently depending on the situation. The hallmark of far(ther) transfer is the need to adapt actions based on judgment. Knowing how to use the knowledge of percentages to determine the discount on a sale is farther transfer than being able to do percentage problems in a math text. Applying project management lessons to home construction projects is farther transfer than applying it on the job, where the learning occurred.

Instructional strategies for transfer

So, how can we help transfer occur to a greater degree? Research shows that the strategies summarized in Table 1 can dramatically improve transfer from instruction to the real world.

David Perkins and Gavriel Salomon synthesized research findings about transfer and concluded that there are two mechanisms for transfer that they called “low road” and “high road.” Low road transfer relates to near transfer situations, where instruction and learning are similar. In that case, teaching learners to perform relatively automatically with extensive practice makes sense. On the other hand, what is needed in high road transfer, where performance requires a fair degree of judgment and skills are performed differently in different circumstances, is varied practice, self-monitoring, and generalizing about how the learning will be applied in the real world.

Transfer and learning objectives

Because e-Learning designers are accustomed to thinking about outcomes of instruction in terms of learning objectives, it may be helpful to see how these concepts tie into learning objectives. Learning objectives can be divided into two types: procedural and declarative. Procedural objectives involve a sequence of steps. An example of a procedural objective is: The learner will be able to replace the hard drive in the XYZ computer chassis in less than 15 minutes. Declarative objectives, on the other hand, involve conceptual knowledge and fundamental principles — for example: The learner will determine what course of action to take if the circumstances violate company ethics policies.

Procedural objectives generally involve near(er) transfer. The role of instructional activities and practice is to help learners gain skill in utilizing a fairly consistent set of rules. Similar contexts, seeing connections, extensive practice, and self monitoring make sense as instructional strategies.

Declarative objectives generally involve far(ther) transfer. The role of instructional activities and practice is to help learners gain skill with determining the best course of action. What is desired, for declarative objectives, is flexible adaptation of skills to a variety of contexts and situations. Similar contexts, seeing connections. extensive and varied practice, extracting underlying principles, and self monitoring make sense as instructional strategies.

In general, research shows that behavioral training methods tend to better promote procedural objectives and near(er) transfer, while more cognitive methods tend to better promote declarative objectives and far(ther) transfer. (Note: Bill Brandon’s June 1, 2004 article in "Learning Solutions Magazine, How Do People Learn? Some New Ideas for e-Learning Designers, " provides a good overview of these learning theories.)

Applying transfer theory

So how does all this apply to our initial scenarios? Let’s first consider learning how to calculate percentages. Being able to solve a simple percentage problem, say 20% of 136, might be considered near(er) transfer, as there is a routine and consistent series of steps to follow (convert the % to a decimal, then multiply the decimal by...). But what if we go into a store and see a jacket we’d like to buy, and the sign says all jackets are on sale for 25% off? Not only that, today there’s an extra 10% off! The rote steps don’t work as well here. The price tag says $245. Do we start by multiplying .25 or .75 times $245? And where does the extra 10% come off? Add to the 25% (.35) or multiply the first result by .90? This is a far(ther) transfer situation. If you’re a math expert who has been called in to assist with development of an e-Learning application to teach percentages, you realize people will, no doubt, regularly come upon this challenge in real life. You’ll also realize that learners will need to do extensive practice with a variety of near and far transfer contexts. You might use a pie analogy at first to help people see that 1/4 of a pie (make mine pecan!) is 25%. And it would be helpful to demonstrate a variety of examples where percentages are used (sales, mortgages, etc.) and help learners use strategies for solving all sorts of percentage problems.

What about the legally defensible hiring practices scenario? This is a far(ther) transfer situation because a great deal of judgment is needed and we can’t put a rote process into place that will take into account every circumstance. We can’t know everything that a potential hire might say or ask. What we do will change depending on the specific circumstances, even though the underlying principles remain the same. In this case, realistic practice with a wide range of situations will always be needed. Learners will need to establish and then utilize underlying principles (such as which types of questions are a definite no-no).

Activities for transfer

If we want to improve transfer, we need instructional activities that involve learners in the types of activities that allow them to practice using the content as it is used in real-life situations. And they need to deal with increasingly complex uses of the content as a whole (not just the parts), and get meaningful feedback and necessary support along the way. That’s a big charge, but that’s what it takes.

Although many folks are charged with developing self-paced online learning because it’s supposed to be less expensive, how cheap is it really if it doesn’t transfer? Research shows that self paced and collaborative activities have different best uses, as Table 2 shows. (They can be used together for maximum effectiveness.)

What I hope you see in Table 2 is that while self-paced activities can be very useful, collaborative activities are often needed as well for instruction where far(ther) transfer is desirable. That doesn’t mean that these collaborative activities have to happen online (many certainly can), but they absolutely need to happen in order for the types of transfer desired to occur.

Example: The time-off policy

Just to make sure this makes sense, let’s reflect on the transfer needs of a situation that I regularly use when I help e-Learning designers learn to support transfer in their online instruction. I like this example because it comes from a real training project. It provides great food for thought, and since the objectives are both procedural and declarative. Here’s the situation:

Company executives want your department to create an e-Learning application for managers who need to appropriately apply Human Resource policies related to time off (unpaid leave, earned vacation time, paid sick leave). The company has experienced myriad problems over the last few years because some managers have applied these policies in problematic ways. Executives fear that lawsuits, and productivity and morale problems, will result if managers don’t apply these polices more appropriately.

What kind of transfer is needed? There may be some rote application (looking up the policy, documenting actions and conversations) but the skill as a whole is pretty complex, requiring myriad decision points and practice application in a wide variety of situations. It is critical that learners extract the fundamental principles and then apply these principles across multiple contexts. Instructional activities will likely revolve around cases, simulations, and problems to solve. Collaborative activities would be appropriate for providing complex feedback, and there will likely be a need to extend the training beyond a single “event,” because mastery of complex skills occurs over time and this is indeed a complex skill.

To give you a feel for the end result of this exercise in designing for transfer, Table 3 illustrates the second of four terminal objectives (and corresponding enabling objectives) that resulted from this design process, the activities selected, information and support tools used during instruction (and later), and graphics and multimedia to support the activities.

What I hope you see here is that the activities (and supporting information, graphics, and multimedia) are designed specifically to involve learners in gaining skill with the wide range of contexts in which judgments need to be made. (For another view of complex design, see “How to Build Composite Learning Progressions Using Approximations” in the December 16, 2002 issue of Learning Solutions Magazine.)

Six mastery problems

Why do folks who build online learning too often concentrate on near(er) transfer and procedural objectives (e.g., simple content interactions like drag and drop exercises and simple quizzes) when far(ther) transfer and declarative objectives (e.g., in-depth tutorials, simulations, feedback and coaching over time, discussions with experts, realistic activities, field work) are what’s needed for learners to gain the desired level of mastery? There are many reasons for this, but here are six really problematic ones, in my opinion:

• Not knowing what critical outcomes are needed from the instruction
• Not influencing stakeholders towards interventions needed for the critical outcomes to occur
• Not realizing that far(ther) transfer is needed for certain critical outcomes
• Deficient skills at designing for transfer
• Mistaken assumptions that online learning has to be all online
• Selecting one tool and building only what that tool will allow

Recognize some common issues here? Yeah, I’d agree that realities and practicalities often make it hard to do what might be totally optimal to ensure transfer. In my mind, however, that doesn’t give us permission to develop the simplistic learning environments I see so often. Many instructional developers think they can choose only between the most expensive forms of instruction (high-fidelity simulations, for example) and the worst forms of instruction (simplistic page turners). That’s simply wrong-headed thinking, in my opinion. There’s a continuum between better and worse choices for instructional strategies. Selecting those that are on the better end doesn’t necessarily have to mean they’re more expensive or time consuming to design and build.

Clients sometimes tell me they just want to build awareness-level training and transfer isn’t a major consideration. I gently (well, maybe not so gently) push back. Is “awareness” training really what’s needed or do people need to be able to do something as a result of the instruction? Sometimes information is enough and formal training with practice and feedback is overkill. But when critical instructional outcomes are needed, designing for adequate transfer isn’t just a nice-to-do activity. We can have a conversation about designing for box-checking another time.

A final question: Did transfer occur?

If we design for transfer, it often makes sense to gauge whether transfer occurred. The classic summative evaluation model was developed by Donald Kirkpatrick in 1975. (Summative evaluation helps summarize the results of instruction, and is conducted after the instructional materials have been used by actual learners and results measured. This is distinguished from “formative evaluation,” which is part of the instructional design process and helps formulate instruction before the materials are delivered to a majority of the targeted learners.) See Table 4 for an outline of the Kirkpatrick model. It purports to evaluate instruction at these four levels. (I say “purports” because it depends on how the evaluation is done.)

Transfer is Level 3 (I could easily argue that Level 4 is about transfer as well) and asks “Can learners apply the knowledge and skills to the job?” If you have put in the effort to design for transfer, it is important to establish the critical outcomes needed from instruction at the front end and figure out what indicators (of progress, mastery, attitudes, or effect) can be used to show that transfer occurred. The key here is to plan, in the beginning stages, when and how to collect and analyze this data. Planning should, of course, involve discussions with stakeholders about what they desire as evidence because their opinions may be different than our own.

The bottom line

We build instruction to impact what people can do in the real world, not what they can do during instruction. For our efforts to be successful and worth the resources put into them, it’s not enough for learners to show “learning” in the classroom, they need to show appropriate application (transfer) in the real world. But research shows that the desired level of transfer happens fairly infrequently. This is a huge problem, but there are effective solutions, even if many instructional experiences don’t take advantage of them.

Instructional strategies that promote transfer include making the context of learning like the context of application, fostering connections between previous and new knowledge, providing extensive and varied practice, providing activities that help learners mine differing situations for underlying principles, and helping learners monitor their learning. Different types of learning objectives have different goals, necessitating different activities and practice opportunities. Procedural, how-to-do, objectives generally involve near(er) transfer, requiring activities and practice that help learners gain skill with utilizing a fairly consistent set of rules. Declarative objectives or why-to-do objectives generally involve far(ther) transfer, requiring activities and practice to help learners gain skill with determining the best course of action.

What transfer concepts tell us is how to be more effective at what we do. Folks who build online learning too often concentrate on near(er) transfer even when far(ther) transfer is what’s needed for learners to gain the desired level of mastery. This is a shame because it often reduces impact and makes it look like it’s online learning, rather than inadequate design, to blame for inadequate outcomes.

If you’re looking for additional ideas on how to design for transfer, check out my earlier columns on designing online interactions (Learning Solutions Magazine, May 3, 2004 and June 29, 2004) and the resources listed below.

References and resources

Clark, R.E. & Voogel, A. (1985). Transfer of Training Principles for Instructional Design. Educational Communications and Technology Journal, 33(3), 113-123.

Greeno, J.G., Smith, D.R., & Moore , J.L. (1993). Transfer of situated learning. In Douglas K. Detterman & Robert J. Sternberg (Eds.) Transfer on trial: intelligence, cognition, and instruction. Norwood , N.J. : Ablex Publishing.

Gruber, H., Law, L-C, Mandl, H., & Renkl, A. (1995). Situated Learning and Transfer. In P. Reimann & H. Spada (Eds.), Learning in humans and machines. New York : Elsevier Science.

Perkins, D.N. & Salomon, G. (1994). Transfer of Learning. In Torsten Husén & T. Neville Postlethwaite (Eds.), International Encyclopedia of Education, Second Edition. Oxford , England : Pergamon Press. Available: http://learnweb.harvard.edu/alps/thinking/docs/traencyn.htm

Perkins, D. N. & Salomon, G. (1987). Transfer and teaching thinking. In: Perkins D N, Lochhead J, Bishop J (eds.) 1987 Thinking: The second international conference (pp. 285-303). Erlbaum, Hillsdale , New Jersey

Perkins, D.N. & Salomon, G. The Science and Art of Transfer. Available: http://learnweb.harvard.edu/alps/thinking/docs/trancost.htm

Van Merriënboer, J. J. G., Clark, R. E., & De Croock, M. B. M. (2002). Blueprints for complex learning: The 4C/ID-model. Educational Technology, Research and Development, 50(2), 39-64. Available: http://www.ou.nl/otecresearch/publications/Jeroen%20van%20Merrienboer/Jeroen%20vanMerrienboer%20etrd.pdf

September 7, 2004

Join or login to save this to your library

Contributor

Patti Shank

President, Learning Peaks

Share

Category

Articles

Topics

Elearning Design Strategies, Data, Measurement, & Evaluation, Instructional Design, Management & Strategy