Thanks for the SciHub link, but it doesn't say what you're saying it does. It says that a particular kind of upbringing predicts a discrepancy between self-reported sexuality and a measure of "implicit sexuality." They further found a relationship between self-reported straightness and homophobia when "implicit sexuality" was measured as "more gay".
Leaving aside the fact that (in my quick read-through, at least) although there was a lot of effort given to validating that this measure measured something, there was little effort given to validating that it measured sexuality, this correlation does not allow one to conclude that "those who profess anti-gay views are likely to be gay themselves" which is the distillation of what was expressed above. Let us start from someone who professes those views. The research means that, if you know this detail of their upbringing and if you know that they explicitly identify as straight (not the same thing as public identification) then you can predict (with clear statistical significance, but still quite low correlation) that that person scores highly on this measure of "implicit homosexuality".
If you check the summary table you can actually just read off the correlation coefficient between homophobic views and the measure of implicit homosexuality and see that it's not statistically significant.
And I do think that the measure of implicit sexuality, though clearly interesting and measuring something is equally clearly not a measure of "are you gay regardless of what you say about yourself." It's reasonable to believe we can use it to estimate homosexuality, but it's like measuring distance with a ruler where all the markings have been scraped off. So even if a study like this did have a correlation with its measure, you then would have to mute the strength of that correlation by the strength of correlation between the measure and the underlying reality we're interested in.