Avis and I will both tell you that we're not impressed with the traditional model for workplace education:

• Employees are flown to a training center, taking them away from their jobs.
• An engaging and knowledgeable person is placed at the front of the room — often with a PowerPoint presentation.
• Trainees sit in the room and listen to the speaker spin yarns and generate hot air.
• At the end, attendees vote on how well they enjoyed the donuts, and whether they like the speaker.

In an age when industry leaders are focused on their shoe-tops, busily calculating the ROI in each passing moment, how much credence can they give to a training model that measures whether the learners are happy at the end?!

There are other choices:

• A certification program associated with the training can measure the learner's abilities.
• The curriculum designer or the facilitator may take steps to measure effective learning, (I add a "live action review" conversation to my demanding courses to make learners aware of the progress they'd made) but in general, this is little more than window dressing.

What really matters: After the training is complete, are the learners better able to do their jobs?

One way to discover the answer is to test them on just that matter.

The problem with many certification programs is that they rely heavily upon selected-response testing (multiple choice). This mostly measures the learner's ability to take tests. It causes facilitators to prep learners for the test. Call it a hunch, but I think that managers who pay for training from operating budgets would prefer the facilitator preps the learners for their jobs.

This is where performance-based testing (PBT) shines. We talked to a professional who has long championed PBT, Mr. Tom Berry. He is director of training at Meru Networks in Sunnyvale, Calif., and previously ran a company that brings PBT to enterprise training programs.

Why is performance testing necessary?

If you look, you'll notice that performance based testing is routinely used when public safety or critical operations are at stake. You wouldn't want a surgeon who is in the operating room on the strength of a successfully completed multiple choice exam. You wouldn't take a flight if you knew the pilot was "qualified" on the basis of having completed an after-class survey.

We find that when a company identifies mission critical skills required of its workers, performance tests are often indicated. It's especially easy to justify when the required skills are mechanistic and where successful performance can be objectively observed

Today when managers want to assess the value of money spent on training, performance testing increasingly looks like a great choice.

Can performance testing be used to measure ability in subjective skill areas?

At first blush, it seems like that would be more difficult. But if you look more closely, you can find ample precedents.

Consider the "skills" of a ballerina. To provide a reliable assessment of her skills, you could take this approach. You take several examples of someone performing the "tasks" and rate them in a continuum that accounts for important factors. For instance maybe you'd rate "grace," "interpretation," and "technical merit."  We train the assessors by using these model. Then a tested performance can be rated in terms of the continuum the examples provide. We might believe that such factors can only be rated subjectively, but the findings from a group of assessors will produce a reliable result.

In a way, it's much like the judging methodology used in the Olympics for skaters or divers.

Is it more expensive to implement, and does it scale?

The cost to develop performance testing is consistent with that of selected response. Both approaches  require some careful thought, good design, and a lot of discussion. The cost to administer the performance test can be made comparable with some creativity, dependent upon scale. If the tested population is of sufficient size, it is possible to spread the cost over the entire population efficiently.

Imagine that we need to test all the massage therapists in a particular state. We could devise a means to do that economically. We might have all of them report to a single testing location. We have each one perform the tasks of their job under observation. We could submit the recorded results to the assessors soon after the "test" is over. If we can get enough applicants to do this in a short period of time, we can keep the costs in check.

You cannot simply compare the cost of deploying a performance test regime with that of deploying a selected response solution. You must consider the cost of having the job performed poorly or wrong, and the cost of hiring or certifying unqualified workers over the course of time. If you have something at stake with the worker, the most important consideration is demonstrated competence.

We have to agree with Mr. Berry about this. The value of a reliable, effective workforce is measured with much larger numbers than the cost of test deployment here. In the case of public safety or danger to the customer, there is no question whatsoever. We like performance-based testing.