Someone on Scienceblogs said they'd be interested in learning about training compression. Industrial geeks will notice that what I am posting here is extremely similar to cGLP/cGMP methods, with a little bit of Mayo Clinic-style training and Dale Carnegie thrown into the mix. I've used these methods to train technicians, scientists and engineers to run the old-fashioned stainless steel chemical reactors as mammalian cell culture bioreactors, to teach undergrads the fine art of protein purification and various biochemical techniques, and to teach chemical engineers basic biology and tissue culture. It used to take my employer about two years to train people to run the bioreactor hall, and now it takes about three to four months. However, if you choose to cherry-pick what I'm posting, realize that you may only get incremental results depending on what technique you select. I'll explain why in a bit.
First: Lots of people who teach in academia have never, not once, had any training or seminars on how to be a good teacher. Some people are naturally good teachers, but those are few and far between. When you’re teaching an auditorium full of bored undergrads, it’s a qualitatively different experience from training someone whose failure could mean your livelihood and health on the line. Trainers cannot rely on their awsum innate skillz to be good at teaching. Don’t rely on student feedback forms to tell you if you’re a good teacher or trainer, either; if you handed out free As for everyone regardless of merit, you’d get great reviews too. If I had a dozen students sleeping through lecture in an auditorium full of 400 other kids, it was no biggie—I handed them an F, told them they could take it up with the dean if they wanted, and sent them on their merry. When the engineer I worked with didn’t thoroughly check the valve configuration and filled the entire 2000 square meter bioreactor hall with hot caustic NaOH to a depth of 2 cm, that was a Problem. In order to be a good trainer, you have to want to be a good trainer, is what I’m getting at. The old, “But I’ve been teaching for a zillion years! That should count for something!” is really not sufficient. My bum has been a bum for…well, a very long time…and I don’t listen to what it says, either. The true test of whether or not you are a good trainer is how quickly your trainees get up to speed and the trainee’s skill level.
Now, about the cherry-picking: The first lesson you learn in education is that people learn in different ways. Some people learn by visual demonstrations. Some people learn by reading. Some people have to have something verbally explained to them. Other people learn by doing something hands-on and figuring out the spatial relationships themselves. Most people have some combination of techniques that helps them learn best. In order to make a whole group of people learn something, you need to have a way of teaching to all of their preferred methods. So, if you decide you’ll only have a ring binder of protocols in the lab for anyone to peruse at their leisure, you’re leaving out the folks who need something explained verbally, the folks who need a demonstration, and so on and so forth, not to mention leaving yourself at the tender mercies of the poor grad student who gets the thankless task of writing up all the protocols. By having super-redundant training methods, you not only reach everyone in your audience but you also make sure that all those crucial details have been communicated to several people for proofreading and troubleshooting.
OK, enough caveats. The actual method. You start by writing every single protocol your lab has ever run in a shared electronic document where everyone can access it. The protocols need to be extremely detailed. For the sake of future publications, I strongly recommend a “Summary” section of each protocol at the beginning that can be neatly copy/pasted into manuscripts. Your Methods sections might be boring and formulaic, but they’ll be consistent and readable. After all the protocols have been written as drafts, send them round to other lab members with “Track Changes” turned on, so they can add their comments in other colors. Or send around hard copies and colored pencils. Make sure everyone gets a crack at it and that all variations on the protocol are added.
Example: Incubate the blot at room temperature for 3-18 hours in a 1:1000 solution of primary antibody in Tris buffer. TBST 0.1% Tween-20 works better—M. Smith I used 3% milk once and it was fine—J. Johnson
Then have the author execute it right in front of you while you are standing beside them. If they missed anything out of habit (“oh yeah, I forgot, you kinda have to flip that switch”), now’s the time to add it. Print hard copies. Make a shortcut on a computer that’s available to everyone. Index it, make it searchable. And be 100% certain that everyone who sets foot in the lab knows where it is and what they should expect to find in it. Follow up and make sure the book is being used and updated regularly. If it’s not, find out why not and hold the authors accountable. If someone in your lab hands you a manuscript with a method that differs significantly from your lab book, demand a resolution.
This maybe should be its own separate step, but it’s really just a part of method review: Are your protocols any good? “Why of course they are good! They were good enough for great-grandad and they are good enough for me!” Have a long hard look at your technology. Then have a long hard look at the latest offerings from Invitrogen, Perkin-Elmer, VWR, and Qiagen. Then look some more. If a protocol takes three years to do properly, and there’s no new alternative method, maybe you need to hike across campus and see if the Engineering department can come up with something better for you. Yeah, I know, back in your day you had to walk to work in the snow, uphill both ways! So did I. It only makes me all the more grateful that now I have a bunch of sleek shiny boxes and an on-site reagent shop and I don’t have to waste years cooking up endonuclease from bacteria. There’s a paper or three in developing and validating a new technology, too. Heck, there’s a whole multimillion-dollar startup, early retirement and a really cool car in that sort of thing. I’ve heard the worry that trainees won’t truly learn how to troubleshoot with a machine, and I tell you it is not so. When the machine breaks down and the field service guy can’t come out till next week, you’ll figure it out right quick. I’ve seen million-dollar automation systems temporarily repaired with rubber bands, plumbing Teflon tape and plastic zip ties by relatively inexperienced chemists.
Next step: Figure out the order that the protocols are best learned in. This order is not necessarily going to depend on the project, nor will it necessarily be with the goal of making the trainee the most useful as quickly as possible. For example, in order to make technicians extremely useful very quickly, we used to train them on the simplest, most routine, everyday protocols. Then we delegated those tasks to them and had more senior techs actually run the reactors. We figured that the techs, as they observed the larger, more complex tasks, would kind of grow into being able to do bigger and better things. This turned out to be a really lousy way to train people, because they had no way of seeing the rationale of why they were being told to do this or that. They waited to be given orders, which meant that I had to think about their schedule, my schedule, everyone else’s schedule, all the bloody time, and follow up to make sure they did it right. Since we had high turnover, that meant I was always mothering over someone new and barely had time for my own projects. When they made a mistake, it was a nightmare to go back and fix it.
What worked better was gradually introducing them to the rationale of why things happened the way they did. They learned the calculation process from the very start, by working with small cultures by hand. Sure, that meant that for a few weeks they spent a lot of time warming a chair, reading articles and watching online training presentations. By using small pumps to move cultures and media around and maintain culture splits, by working with very simple disposable reactors, they learned the basic requirements, the math and engineering that guided the system, and figured out the hows and whys. Then, when they had proven themselves on a small scale, we taught them the same thing again on the big reactors. I never, never once had to correct a tech’s math when they were taught small-scale cultures maintained by hand first. When I had full-blown engineers and scientists who were trained in the “let’s make them useful first” method, I always had to correct their math and explain something fairly elementary. Over and over and over. It had nothing to do with their innate ability, everything to do with the training method. There’s an order you need to teach, and in my experience it’s best to start with the big picture first.
Third step: Train the trainers. I think the best teachers I ever had were from a small liberal arts college. Those folks were heavily invested in teaching as their primary job, and by and large they were pretty good at it. Lots of things that I have since been told are “impossible” to teach in less than two years, they happily taught in a few months as a regular lab course: various types of liquid chromatography, protein purification, electron microscopy, tissue culture, veterinary surgery, all kinds of things. The catch is that they made a big honkin’ point of getting professors who really cared about teaching. The general method, which I have seen many other teachers and corporate trainers use with great success, is this:
1. Teach a lecture explaining the theory. Give a short quiz on theory. You need to have good lecturers do this, don’t assign it to the guy who barely speaks a word of the local language. I can just about order a beer and a sandwich in German without making a total fool of myself, but I’m sure not ready to go lecturing at Max Planck Institute.
2. Once the students are in the lab, demonstrate the technique while narrating what you are doing. This is a skill that the Dale Carnegie public speaking courses teach pretty well, and I highly recommend them.
3. Walk the students through doing the protocol themselves, nitpicking their technique until they get it exactly right. Don’t give up and decide that it’s good enough when it’s not. Really nitpick, in a nice and respectful way of course.
4. Give them a protocol to run by themselves that you know will run just fine, given the material they already have. Do not expect them to troubleshoot at this stage.
5. Demonstrate yourself, and walk the students through several scenarios, where something went horribly wrong, and explain why it was wrong and how to ensure that the wrong thing doesn’t happen.
Schedule exactly 2.5 times the length of time it normally takes you to run the protocol when everything goes smoothly, just to run each protocol one time. I can split a 2500L reactor culture in about two hours, but if I’m training someone I allocate five hours for the process. This allows time for the narrative, errors, double-checking, questions on the fly, etc. That means for a two hour task, you’re going to spend a minimum of 20 hours training someone on how to do it. And yes, that’s even if they do have prior experience and education on the protocol. They might have learned a very different way.
The key here is that you don’t want students getting frustrated with something that doesn’t work all that well to begin with. You want them to build the confidence that they can carry out this protocol, by themselves, and they can make some good guesses as to why things go wrong and how to fix it. You want them to have this self-confidence before you set them loose in the lab. Otherwise, it’s a recipe for lots of wasted materials, wasted time, wasted energy.
The bonus you get for yourself from this exercise is robustness of results. Let’s imagine you have a postdoc in your employ who seems to have magic hands that get all the right results that support your pet hypotheses du jour. When you test-drive his/her protocols with trainees, the results the trainees get are only half-baked, for some mysterious reason. Now that you realize that your results aren’t really all that reproducible, you know that you need to take a long, hard look at your postdoc’s notebooks. Hey, if the notebooks check out, are high-quality legible data written in ink, signed and dated, great. If not…
Fourth step: Make sure you hired people who can do the job. “Thanks, Captain Obvious!” Yeah, you’re welcome. But when my grad school advisor made up his mind to hire a student who had no experience, no motivation to learn, and only wanted to put in a year or two before buggering off to medical school, he didn’t see for the life of him why that was a Bad Idea. Things You Can’t Teach include motivation, interest, a love for the field, and passion. Don’t hire someone who only wants the job because they don’t know what to do with themselves, because their mommy told them they needed a Master’s, or because this was Plan Y and Plans A-X have already failed them. If a trainee sees no reason to commit a procedure to memory, they’re not going to learn it. On the contrary, you can teach someone to carry out a grad school-level procedure even if they completely lack formal education, so long as they are interested in the subject and motivated to learn.
Final step: Keep on top of it. Track who is trained on what and create a training program for all new lab members. Have trainers decide who is adequately trained, don’t leave this up to trainees. Reward people who get trained on new techniques somehow, and reward people who invent or bring new techniques from other labs. Make it clear that you value training in your lab, and that innovation will be honored and respected—a Nature Methods paper should be respected just like any other Nature paper.
Bear in mind, most of the examples presented here are from my own experience organizing training for a bioreactor suite. The important thing to take away is that this is a method for organizing thought processes of a large group and delineating the degree of training involved for both trainee and trainer. It can be applied to most any training process.
No comments:
Post a Comment