Archives for February 2009

Previous Series Post |Next Series Post

This is one in a series of thoughts on some broken areas of ID that I‘m posting for Mondays.   I intend to provide insight into many ways much of instructional design fails, and some pointers to avoid the problems. The point is not to say ‘bad designer‘, but instead to point out how to do good design.

I see several reliable problems with examples, and they aren‘t even the deepest problems. They tend to be mixed in with the concept, instead of separate, if they exist at all.   Then, when they do exist, too often they‘re cookie-cutter examples, that don‘t delve into the necessary elements that make examples successful, let alone are intrinsically interesting, yet we know what these elements are!

Conceptually, examples are applications of the concept in a context.   That is, we have a problem in a particular setting, and we want to use the model as a guide to solving the problem. Note that the choice of examples is important. The broader the transfer space, that is, the more general the skills, the more you want examples that differ in many respects.   Learners generalize the concept from the examples, and the extent to which they‘ll generalize to all appropriate situations depends on the breadth of contexts they‘ve seen (across both examples and practice).   You need to ensure that the contexts the learner sees are as broadly disparate as possible.

Note that we should also be choosing problems and contexts that are of interest to the audience.   Going beyond just the cognitive role, we should be trying to tap into the motivational and engagement factors.   Factor that into the example design as well!

Now, we know that examples have to show the steps that were taken.   They have to have specific steps from beginning to end.   And, I add, those steps have to refer back to the concept that guides the presentation.   You can‘t just say “first you do this, then you do this”, etc, you have to say “first, using the model, you do this, and then the model says to do that”.   You need to show the steps, and the intermediate work products.   Annotating them is really important.

And that annotation is not just the steps, but also the underlying thought processes.   The problem is, experts don‘t even have access to their thought processes anymore!   Yet, their thinking really works along lines like “well, I could‘ve done A, but because of X, and thought B was a better approach, and then I could do C, but because of Y I tried D”, etc.   The point being, there‘s a lot of contextual clues that they evaluate that aren‘t even conscious, yet these clues are really important for learners. (BTW, this is one of the many reasons I recommend comics in elearning, thought bubbles are great for cognitive annotation.)

Another valuable component is showing mistakes and backtracking. This is a hard one to get your mind around, and yet it‘s powerful both cognitively and emotionally.   First, experts model the behavior perfectly, and when learners try, they make mistakes, and may turn off emotionally (“I‘m having trouble, and it looks so easy, I must not be good at this”).   In reality, experts make mistakes all the time, and learners need to know that. It keeps you from losing them altogether!

Cognitively it‘s valuable, too.   When experts show backtracking and repair, they‘re modeling the meta-skills that are part of the expertise.   Unpacking that self-monitoring helps learners internalize the ‘check your answer‘ component that‘s part of expert performance.   This takes more work on the part of the designer, like we had with the concept, but if the content is important (otherwise, why are you building a course), it‘s worth doing right.

Finally, I believe it‘s important to convey the example as a story.   Our brains are wired to comprehend stories, and a good narrative has better uptake.   Having a protagonist documenting the context and problem, and then solving it with the model to achieve meaningful outcomes, is more interesting, and consequently more memorable.   We can use a variety of media to tell stories, from prose, through audio (think mobile and podcasts) and narrated slideshow, animation, or video.   Comics are another channel.   Stories also are useful for conveying the underlying thought processes, via thought bubbles or reflective narration (“What was I thinking?…”).

So, please do good examples.   Be exemplary!

While I agree vehemently with most of a post by Lars Hyland, he said one thing I slightly disagree with, and I want to elaborate on it.   He was disagreeing with   “buying rapid development tools to bash out ill formed ‘e-learning’ to an audience that will not only be unimpressed but also none the wiser – or more productive”, a point I want to nuance.   I agree with not using rapid elearning to create courses for novices, but there is a role for bashing out courses for another audience, the practitioner.   And there’s something deeper here to tease out.

I want to bring up John Carroll’s minimalist instruction, and highly recommend it to you. He focused on a) meaningful tasks, b) active learning quickly, c) including error recogition & recovery, and d) making learning activities self-contained (a lot like games, actually).   In The Nurnberg Funnel, he documented how this design led to 25 cards, 1 per learning goal, that beat a 94 page traditionally designed manual hands-down in outcomes.

Another way to think about it is something Jim Spohrer mentioned to me once. Now, Jim’s been an Apple Fellow, and is leading research at IBM’s Almaden Research Center.   He really cares and likes to help people, but he’s very busy.   So he adopted a ‘least assistance’ principle, where he would ask himself what’s the least he can do to get this person going, because there was more to do and more people to help than he was able to keep up with.   And I think it is a useful way to think about supporting learning.

This sounds a lot like performance support, and that’s definitely a mind-set we need to adopt. When Harold Jarche and Jay Cross talk about the death of the training department, they’re talking about not focusing on courses, and instead taking a broader, performance perspective.   Obviously, we want to talk about portals of resources, but we also need to recognize that there are formal learning situations that don’t require the full formality.

We develop full courses to incorporate motivation, practice, all the things non-self-directed learners need.   But there are times when we need to provide new information and skills to self-directed learners.   When we’re talking to practitioners who are good at their job, know what they’re doing and why, and know that they need to know this information and how they’ll apply it, we can strip away a lot of the window dressing. We can just provide support to a SME so that their talk presents the relevant bits   in a streamlined and effective way, and let them loose.     That, to me, is the role of rapid elearning.

It’s not for novices, but it’s effective, and more efficient.   In this economic climate, we don’t have the luxury of full development of courses for every need.   Moreover, in any climate, we shouldn’t give people what they don’t need, instead we need to focus on what the ‘least assistance’ we can give them is.

In many cases, the least assistance we can give is self-help, which is why I believe social learning tools are one of the best investments that can be made.   The answer may well be ‘out there’, and rather than for learning designers to try to track it down and capture it, the learner can send out the need   and there’s a good chance an answer will come back!   There’s a lot to making such an environment work; it’s not the case that ‘if you build it, they will learn’, but it’s still going to fill a sweet spot in the performance ecosystem that may not be being hit as of now.

Don’t look for everything you can do in one situation, unless you’re flush with too much time and resources (in which case, watch out!), instead look for the least you can do that will get the job done so you can do more for everybody. It’s likely that’s more to their taste, anyway. And that’s enough from me on that!

Previous Series Post | Next Series Post

This is one in a series of thoughts on some broken areas of ID that I‘m posting for Mondays.   The intention is to provide insight into many ways much of instructional design fails, and some pointers to avoid the problems. The point is not to say ‘bad designer‘, but instead to point out how to do better design.

At some point (typically, after the introduction) we need to present the concept.   The concept is the key to the learning, really.   While we‘ve derived our ultimate alignment from the performance objective, the concept provides the underlying framework to guide one‘s performance.   We use the framework to provide feedback to help the learner understand why their behavior was wrong, both in the learning experience and ideally past the learning experience the learner uses the model to continue to develop their performance.   Except that, too often, we don‘t provide the concept in a useful way.

What we too often see is a presentation of a rote procedure, without the underlying justification.   In business, we‘ll teach a process.   In software, we‘ll see feature/function presentations (literally going item by item through the menus!).   We‘ll see tutorials to achieve a particular goal without presenting an underlying model.   And that‘s broken.

We need models! The reason why is that people create mental models to explain the world.   People aren‘t very good at remembering rote things (our brains are really good at pattern matching, but not rote memorization).   We can fake it, but it‘s just crazy to have people memorize rote things unless it‘s something we have to absolutely know cold (medical terminology is an example, as are emergency checklists for flights).   By and large, very little of what we need to know needs to be memorized.

Instead, what people need are models.   Models are powerful, because they have explanatory and predictive power.   If you forget a step in a procedure, but know the model driving the performance, you can regenerate the missing step.   With software, for instance, if you present the model, and several examples where the way to do something is derived from the model, and then you have the learner use inferences from the model to do a couple of tasks, you might be saved from having to present the whole system.

People will build models, so if you don‘t give them one, it‘s quite likely that the one they do build will be wrong.   And bad models are very hard to extinguish, because we patch them rather than replace them.   It requires more responsibility on the designer to get the model, as, for reasons mentioned before, our SMEs may not be able to help us, but get them we must.   Realize that every procedure, software, or behavior has a model that drives the reason why it should be done in a particular way, and find it. Then we need to communicate it.

Multiple models help! To communicate a model most effectively, we should communicate it in several ways.   Models are more memorable than rote material, but we need to facilitate internalization.   Prose is certainly one tool we can and should use (carefully, it‘s way too easy to overwrite), but we should look at other ways to communicate it as well.

Multiple representations help in several ways.   First, they increase the likelihood that a learner will comprehend the model, and then have a path to comprehend the other representations.   Second, the multiple representations increase the number of paths to activate a model in a relevant context.   Finally, multiple representations increase the likelihood that one can map closely to the problem and facilitate a solution.

Multiple representations are, unfortunately, sometimes difficult to generate (more so than finding the original model).   However, we should always be able to at least generate a diagram.   This is because the model should have conceptual relationships, and these can be mapped to spatial relationships.   There‘s some creativity involved, but that‘s the fun part anyways!

Yes, doing good instructional design does take more work, but anything worth doing is worth doing well.   On a related, but important, note, unfortunately the difference between broken ID and good ID is subtle.     You may have to explain it (I have literally had to), but if you know what you‘re doing and why, you should be able to.   And having developed a powerful representation increases the power, and success of the learning, and consequently the performance.   Which is, of course, our goal. So, go forth and conceptualize!