(ok. this is substantial drift...but i'm assuming that given the header, anyone who reads this is interested in the subject [and probably off his/her rocker but that's ok *smile*])
On Tue, 23 Dec 2003, Sven Venema wrote: > Okay, attribution wise I'm kinda lost.. but "SOMEONE" said... :) > > > > > Not in this case....the power of the experiment is that you're > able to > > > > ferrett out the causal element... > > > > > > > > Take a population of subjects and split them in two (or three, > four, etc.) > > > > such that if you compare the resulting groups to each other, > there is no > > > > substantive difference. There aren't more women than men, there > aren't > > > > significant income differences, etc. > > > > > > > > Then you expose one group to one set of stimuli, and expose the > other > > > > group to some sort of placebo. > > > > > > > > The difference in the response to the dependent variable HAS to > come from > > > > exposure to the stimuli. > > Not a flame, just a comment: > > The idea that the difference in response "HAS to come from exposure to > stimuli", and that you could find groups of people who have "no > substantive difference" seems a little simplistic to me. IMHO this > approach is probably not generally applicable to humans. It seems that > humans (from an outsider's point of view) have a penchant for acting on > impulse or whim. Even though someone's actions may be entirely > consistent within their own internal framework, those actions may seem > irrational to an outside observer. I'm not sure that it's possible to > define a 'standard' human being, let alone bring together a group of > these standard human beings against which responses can be measured. > > Sven In the aggregate it is very applicable to humans. The key here is replication. If what you're saying is correct...that humans (in the aggregate) are prone to responding on a whim, then my experimental results wouldn't be replicable. That is, the first time I do this experiment with group A then I'll get one result, then the next time I do it, I'll get another result, then the third time I'd do it I'll get yet another result. The results would be random and totally unpredictable. But this isn't what happens. For example, in experiment after experiment, it's been shown that white americans exposed to stories about welfare (accompanied by an image of a black woman) are far more prone to say "poverty is the individuals fault" and that "we shouldn't do anything to help the poor" than white americans exposed to stories about welfare (accompanied by an image of a white woman). If what you say is true, then we wouldn't get consistent results over time and space here. So while your general point is correct--and it is important to say that I'm talking about groups of people, not individuals--I think there is enough consistency to think about experiments as a solid research tool that can be used to explain and predict human activity. peace lks
