| ||||||
| Home | Blog | Projects | Methods | Downloads | Credits | Links |
|---|---|---|---|---|---|---|
**** This file has a corrupted %%EOF marker, or garbage after the %%EOF. **** The file was produced by Adobe PDF library 5.00: **** please notify the author of this software **** that the file does not conform to Adobe's published PDF **** specification. Processing of the file will continue normallyHmm… Should I tell them they don't conform to themselves?
Update: I often complain about Melbourne's habit of serving lattes in glasses, which are of course too hot to hold. Nathan points out that since I drink them anyway (wrapped in a napkin), I have just proved the sensibility of handles. Sigh.
A counterpart to high-energy physics. Psychology does not need much energy, but it needs a lot more power. Statistical power. I have a plan.
Physicists occasionally discover that to have a chance in heck of seeing something of interest, they need a terawatt collision. Now, most physicists don't have terawatt colliders. If they were psychologists, they would run some experiments on their megawatt colliders and hope that nevertheless they would shed some light (pardon the pun). Or that if they didn't, then at least one of their colleagues in the same boat would.
Funny thing, is that they don't. They cooperate. "Hey, I need a terawatt. Dang. Don't have one, and beamtime at Los Alamos is 200 times what I have on my grant. Hey, why don't I collaborate with 200 other physicists?"
I think psychology would be a lot better off if we thought that way.
Consider: instead of 50 studies with N=20, where half say "significant" and half say "not significant", we could have one well-designed study with N=1000, and a brilliant analysis giving not only tiny p values, but gasp! tight confidence intervals around an estimate of effect size! And all in less time than it would take to publish 50 separate studies.
I've been told that the main barrier is publication: people need to publish regularly, even if it means publishing tiny experiments which are no use on their own. Here is another publication model.
Welcome to the Journal of High-Powered Psychology! Here's the deal: you contract with us to do an experiment. We only accept papers whose designs we have pre-approved. But if we pre-approve your design, and you carry it out as agreed, we'll publish it no matter how the results turn out! Just think, for a little extra care up front, you can be sure that our panel of experts approves of your design and analysis, and finds that the question has been posed to be of interest whether the results are positive or negative. No more reporting bias, and no more wasted experiments.
Of course, this makes reviewers far more valuable. They're not just guardians at the gates: they become coaches helping to make sure that all experiments are good experiments. So you have a good question, but we determine you need 3000 subjects? No problem: welcome to our experimental network. That's right, we have a pool of psychologists who have agreed to use their labs to help run your experiments. Once you've got your design, you hold off running until enough collaborators have signed up, and then you all follow the same method. (Or, preferably, you all pool your funding and collect the data in one place.)
The catch? Well, as one of our potential authors, you too must agree to timeshare your lab and subject pool. Obvious benefits: subjects from more than one area of the country, larger subject pools, etc.
But what about my publication rate? Well, it's true that any one experiment will run slower. First, it will have to be approved. Second, you may have to coordinate runs with collaborators. On the other hand, you have the chance to be co-author on many other papers. And if being one of 200 co-authors is expected for high-energy physics, being one of 20 will be just as prestigious for high-powered psychology.
Any takers?
