Regardless, it's debatable whether "12% more likely" even qualifies as statistically significant.
"12% more likely" is bad wording.
Later in the article, it explains that 11% of heteros were criminals, versus 23% of homosexuals. So its more than 100% more likely.