Drugmonkey has an excellent post about the latest <achtung!> command from LordHighCollins of the The NIH.
I absolutely agree with DM that I would be *thrilled* to do more sex-difference studies/analyses in my work. And publish the results, whether or not they are significant. And explain what my perception of those differences, or lack thereof mean.
As usual, his reasoning hits the nail on the head. It's going to cost a huge amount more. I do animal work with 'real mammals' (ie. USDA species, not rats & mice). In fact they're not rodents. In fact each one costs A Lot Of Money. Animals and their housing are two of the biggest line items on my grant. And, as was mentioned, doubling won't be enough, because it won't permit testing of sex*effect interactions. Given that I'm at the limit for power, I can't reduce sample sizes anywhere else. We do one or two animals at a time. We can't really scale up (they are big animals, ha ha ha, and therefore will hit biomechanical failure with scaling... just a little allometry joke) in terms of person-work-effort, so it will be at least a doubling of time.
I think, however, this kind of misses the argument about the value of animal models. There, I put it in bold italics to catch your attention. Animal models do not prove that whatever we are testing will cure cancer or make disabled children walk again. Animal models give us the basis for what is happening, the mechanisms, the underlying biology.
I receive lots of criticism at the clinical meetings I attend (often from older clinicians who don't understand data analysis and from younger clinicians who are already know everything) for working with animals. Those arguments are roughly: when you do animals you have no compliance issues, you have no problems that these are older/younger/infant people who don't do <whatever> the same way you get your animals to do <whatever>. Animals don't have all the confounding conditions that my patients do. Animals don't have cognition to think about what you're asking them (though I disagree with this, a bit). Animals don't have families who support or don't support the intervention (though they do have my brilliant postdoc and tech).
My answer is to these critiques is "exactly right". With an animal model you can isolate whatever it is you are testing, examining, discovering from all those other things. If I want to show this whole class of interventions works because of CNS response to peripheral something or other, I am not going to have to worry about all those other factors. I want to know about this set of nerve connections. If it is nerve pathway A, then A1 will help, but if it is pathway B then it won't, but B1 has been shown to do something. In your precious human samples, you have so many factors, that you can't possibly know what is making what else happen.
In fact, I use female animals. They are more tractable. They take less time to train. My supplier will give me a 20% discount if I buy males. I've done that from time to time. They are a pain in the ass. I've seen no real difference in the clinical conditions I'm working with. But my power is low in those tests. The effect size is negligible compared to the condition/intervention effect size. Does that mean its clinically irrelevant? I don't know. Male and female nervous systems are different. But that is not the main effect I'm testing. The people who blithely say "oh stop moaning" and "there will be resistance" "just lie back and thing of Mother England", sex is already their main effect. They frost my shorts.