Self reports are used extensively in criminology studies. Read the article that I posted it has a nice explanation.
The real problem is not truth telling, but the the interpretation of the question. The women are not reporting being raped or assaulted. That is the interpretation of the researchers. They are counting a woman as "raped" if they say that anyone ever had any level sexual contact with them (including kissing) while they were unable to consent because they were drunk or asleep. That includes boyfriends and even husbands that made out with them while they were drunk or kissed them while they were asleep. The respondents do not need to view the act as any form of assault or in any way wrong or unwanted. In fact, other research using these questions shows that most women counted as being raped using these questions do not view the act as wrong. IT just means that technically someone did kiss or grope them while they had been "drunk" (which can mean completely conscious and aware) or sleeping as couples often do.
Another problem is the fact that 58% of the people given the survey did not bother to complete it and return it. That is a massive sampling bias, and nothing in their methods deals with the fact that non-responders are generally people who do not see the survey as relevant to them, which in this case means they were not raped even by the researchers loose definition of it. It could easily be that nearly all non-responders would not qualify as being raped by their definition, making the number closer to 1 in 11, with many or most of those being people who mutually groped or kissed their regular sexual partner while intoxicated and not "raped" or assaulted as the concept is understood by the vast majority of people.