Laura You know, that it's not. Probability is much higher than 50%
As far as I understood it right the probability for a wrong non-virginity result is much higher.
Kellog et al. studied a cohort in the United States with definitive evidence of previous vaginal penetration. In the study of 26 pregnant adolescents who reported sexual abuse, 22 participants (64%) had normal or nonspecific genital examination findings, eight (22%) had inconclusive findings, four (8%) had suggestive findings, and two (6%) had findings of definite evidence of vaginal penetration [23].
64% clearly wrong and only 8% rite that she is no virgin.