After the 2016 presidential election, a massive surge of liberal young women began identifying as bisexual. But won't have sex with other women.
My belief is that what people choose to self-identify with is subject to a variety of psychological and social factors, the virtue signaling potential, as well as ideological inclinations. To this end, it's not much surprising that a larger percentage of the increase in women identifying as LGBT are liberal women (my theory is that it's ideologically more consistent and meaningful for them to do so compared to conservative). On the other hand, a behaviour such as sexual intercourse would be expected to be pretty much resistant to those same factors mentioned above. Very few people would choose to have sex with someone they're not attracted to just for ideological loyalty or virtue signaling. Hence, it's not surprising that this measure has stayed relatively the same.
Taking these two theses together, they shed some light on the last statistical fact reported in your article: that the number of bisexual-identifying women with only male partners has increased four folds. I believe this is exactly what would be expected in a scenario where women of certain ideological slant, who are motivated in droves to identify with a socio-politically chic sexual orientation, are unable to translate this artificial (extrinsic) self-identification into the real domain of sexual decision and action (which answers more consistently to intrinsic forces).
The four-fold quadrupling would thus appear to be a consequence of a similar fold in false identification. It is reasonable in this age and era to use as a yardstick to measure sexual orientation (as against sexual preference) the sex one chooses to have intercourse with rather than relying exclusively on (or at least in addition to) one's word. It's a well known limitation in studies relying exclusively on self-report that what people claim about themselves often doesn't exactly match what they actually do in real life situations. The margins of these discrepancy are often significant enough to skew statistical perceptions.
In short, we immediately complicate the meaning of all these reported data the moment we start with the ridiculous assumption that we can absolutely trust what people reported about themselves much more than how they actually behave (when under no duress).
The question is, if bisexuality is genetic, would there be any difference in male vs. female rates? Considering that identical twins don't have the same sexuality most of the time, sexuality is likely partly genetic and partly cultural. And we have an ideology in America that affects liberal women the most. But even 20%+ of conservative female undergrads consider themselves LGBT! That's a crazy stat.
But this issue should correct itself. Once claims of bisexuality reach a certain point, no one will care anymore. Then the fakers will go back to identifying as straight, leaving only the real bisexuals.