Bailey does have evidence about the existence of bisexuality in men. It's very preliminary,and hardly tells a definitive story, but at least it's some evidence. It's a shame that in his interview, he didn't stick to it, and that the reporter wasn't responsible enough to do so either. The evidence is reported in a paper titled "Sexual arousal patterns of bisexual men," by Rieger, Chivers, and Bailey (Rieger and Chivers are two of his graduate students) that is currently in press at Psychological Science. Here are the basics of the experiment. Bailey and his students recruited 30 gay, 38 straight, and 33 bisexual men through "gay-oriented" and alternative magazines in the Chicago area. Sexual orientation was measured entirely through self-report, using the Kinsey scale, with 1 indicating homosexuality, 5 indicating heterosexuality, and 2-4 indicating bisexuality. They had these men view six 2-minute videos of sexual material, along with videos of sexually-neutral material. Two of the sexual videos showed male-male sex (oral and anal), two showed female-female sex (oral and vaginal penetration using a strap-on dildo), and two showed male-female sex (oral and vaginal). They measured physical arousal using a technique called plethysmography that measures changes in the circumference of the penis. This is a fairly widely used technique, and short of neuroimaging (which comes with its own set of problems), it's the best way to measure physical arousal. In addition, they had the participants rate their subjective arousal continuously, using a lever that could be moved in a 180 degree arc, with 0 meaning no arousal, and 180 meaning orgasm-level arousal.
After weeding out several participants due to insufficient levels of arousal to any of the sexual videos (plethysmography is notoriously bad at detecting low levels of arousal), they compared the physical and subjective arousal of the 22 remaining self-reported bisexual to the 21 straight and 25 gay men who made the statistical cut. The question was, do self-reported bisexual men show and report significantly more arousal to male-male sex videos than straight men, and to female-female videos than gay men*? This is what we would predict if bisexual men are attracted to both sexes. This is in fact what they found for the subjective arousal measure. Bisexual men showed high levels of subjective arousal to both male-male and female-female videos. However, the measures of physical arousal were not consistent with bisexual attraction to both genders. Instead, the bisexual men showed high levels of arousal to either the male-male or female-female videos, but not both. Most of them showed high levels of arousal only to the male-male videos. Bisexual males were the only ones in the study whose subjective and physical arousal levels did not show high positive correlations.
What does this mean? Here is what the authors of the study concluded:
[O]ur results suggest that male bisexuality is not simply the sum of, or the intermediate between, heterosexual and homosexual orientation. Indeed, with respect to sexual arousal and attraction, it remains to be shown that male bisexuality exists. Thus, future research should also explore nonsexual reasons why some men might prefer a bisexual identity to a homosexual or heterosexual identity.Now this is a carefully worded paragraph at the end of a discussion section in a scientific paper (the emphasis is all mine). Their results don't say anything conclusively, they merely "suggest," and while the authors clearly believe that bisexuality doesn't exist, they only say that its existence "remains to be shown." In addition, they don't suggest that further research on the existence of bisexuality should not be conducted, but that researchers should also study alternative explanations for self-reported bisexuality.
Now there are problems with this conclusion. First, they offer no justification anywhere in the article for equating sexual attraction with sexual arousal as measured by changes in the circumference of the penis. In fact, they do have evidence that bisexual "attraction," if not arousal, exists. The evidence is in their measures of subjective arousal. If subjective arousal doesn't measure attraction, then I have no idea what it measures. What they did find, it appears, is a dissociation between attraction and arousal for bisexual men. What this means, however, is a question that will have to be answered by future research. Bailey and one of his undergrads are actually conducting more research, this time using neuroimaging to measure arousal, a technique that they suggest in this paper. The results of that study are not currently available, though.
In addition to the problems with the conclusions they draw from the study, there are problems with the study itself. The biggest problem is that all of the men were recruited through gay and alternative magazines (though they do not report how many were recruited from each). It's quite reasonable to assume that bisexual men who regularly read gay magazines are more aroused by men than women. This is another instance of the recruitment problem that plagues much of the research on sexual orientation. With a relatively small and likely unrepresentative sample, it's not clear how generalizable the results are. Another problem, which they do mention, is the crudeness of their measure of arousal. While plethysmography is the most reliable method currently available, roughly one third of the men in this study (and others that use the method) had to be excluded due to failure to detect significant differences in arousal between the sexual and non-sexual videos. Better measures might show more subtle differences in arousal levels between gay, straight, and bisexual men. Imaging techniques, if they show converging results, could help, though imaging data would be difficult to interpret.
The point, then, which is completely lost in the NYT's article, is that the question, "Does male bisexuality exist?" is not answered by this study. The results are interesting, and certainly make the case for further research, but that's about all they do. It would be a serious mistake to dismiss self-reported sexual attraction and orientation based on one preliminary study. Instead, we should do what scientists should always do, and what the authors suggest: explore all of the possible explanations for the data, while continuing to attempt to gather better data.
UPDATE: One more comment on Bailey's quote in the NYT article, in which he said, "For men arousal is orientation." This seems to be a hypothesis, derived from the evidence showing that for both gay and straight men, self-reported sexual orientation, subjective arousal, and physical arousal are all highly and positively correlated (the correlations are up around .95). However, this theory makes a prediction about the correlation of self-report, subjective arousal, and physical arousal for bisexual men, and this prediction was not supported by the data. Instead of admitting this, Bailey instead sticks to his hypothesis, and offers an explanation for the bisexual data that is not justified by that data -- that male bisexuality does not exist. A more scientif approach would be to note that the prediction is not supported by the data, and then call for further data to attempt to either bridge the gap between the hypothesis and the data (as the research article actually does) , or suggest alternative hypotheses. Given the data we do have, from Bailey's own experiment, the hypothesized identity between sexual orientation and arousal in men has no more, or depending on your interpretation, less support than the existence of male bisexuality (which is supported by self-reported sexual orientation and the subjective arousal data).
* The male-female videos were only included to ensure that straight men were not simply averse to homosexual sex. Since all males (gay, straight, or bisexual) showed physical arousal to all types of video, and since straight men showed high arousal to the female-female videos, the male-female condition is really unnecessary. The female-female video is a better for comparison to the male-male video, because it does not involve any member of the unpreferred sex for straight men, or the preferred sex for gay men.