On Thu, 18 Mar 2004, Wojtek Sokolowski wrote:
> Its is an accepted standard in empirical science that the burden of
> proof lies with those who make the claim - if they cannot substantiate
> it, then the null hypothesis is accepted (no pun intended).
This is a textbook caricature of science, not typical scientific practice. If a scientist wants to make a strong claim about the relationship between two variables--even the claim that the variables are not related!--she needs data to back up her claim. To claim that two variables are not related without any data to back up your claim is not proper scientific practice.
Example: when social psychologists conduct research on altruism, they often hypothesize that situational factors will strongly influence helping behavior, swamping the effects of any personality traits like selflessness. Thus social psychologists predict that personality traits will not be strongly related to helping behavior in typical social situations. In contrast, personality psychologists argue that some people have "altruistic" personalities, so of course personality and helping behavior will be related.
Using Woj's logic, we should just accept the social psychologists' view here, because the "burden" is on the personality psychologists to show there is an effect. --Of course, you'd get laughed out of any research psychologist's office if you made that claim. Whether or not the social psychologists or the personality psychologists are correct depends on gathering data; it doesn't matter which theory makes the prediction that the independent and dependent variables are related.
--And just so with the efficacy of medical treatments.
> SSRI's are used in many affective disorders - and claiming that they are
> less effective than talk therapy is plainly false. For example, talk
> therapy has very little effect in bipolar disorders (although it is
> sometimes used in conjunction with medication) - SSRIs have a much
> better track record there.
It's important to distinguish mood disorders here. Depressive disorder and bipolar disorder are quite different: much higher heritability ratio for bipolar, lower cure rates for bipolar, different prevalence rates (depression about 10x more likely in the U.S.), different effective therapies for each.
I've said it before, I'll say it again: in clinical trials comparing the effectiveness of SSRIs alone to psychotherapy like cognitive-behavioral therapy for the treatment of depression, the data are clear: long term effects are just as good, and in some studies better, for the psychotherapy treatment. I can provide the references if you like, or you can look it up yourself, but that's really the story.
For bipolar, cognitive therapy is not this effective. Even so, check out Nathan & Gorman's (2002) A guide to treatments that work. There's a chapter reviewing psychosocial treatment for bipolar, and a number of studies have demonstrated better outcomes for combined psychotherapy-med treatment than meds alone (better med adherence, fewer hospitalizations, improved social functioning).
Also, anti-depressant drugs aren't that useful for bipolar disorder; they tend to encourage more rapid cycling between manic and depressive moods (they exacerbate the symptoms!). Lithium is the typical treatment, but even this is not as effective as Woj seems to assume (about 1/3 of the people with bipolar don't get any benefits from lithium treatment).
Miles