It is a false assumption that "people have no motivation to lie in a confidential study." People could lie to make themselves feel better. People lie to themselves all the time. Repeating that lie in an "official study" will definitely cement that false belief more firmly in thier head. Several studies have shown this. Some people could give false answers in a study, especially one of this nature, because they believe others will not be brave enough to come forward. So the respondant fudges the truth to, in their thinking, skew the results further toward what they think the "real" results should be. In other words, they ruin the survey by trying to "fix" it. Still other people may lie just to screw up the results, just for the hell of it. Yes, even young women.
Finally, there are many women who do not understand what sexual harrassment really is. These women tell dirty jokes, then get upset when the man laughing the loudest is not attractive to them. Women who act flirty and touchy one day, then the next day complain about that same man being flirty and touchy back. Many women believe that sexual harrassment is absolutely ANYTHING that makes them feel uncomfortable in ANY way. But this is too subjective. That is why "harassment" is defined as a continuing pattern of behavior that continues even AFTER someone has been asked to stop. Remember, this pattern must be either one person doing lots of things, or lots of things being done to one specific person or group by another specific group. A "pattern," in this context, is NOT establshed by a survey that shows lots of different disconnected people feeling uncomfortable about what happened in a lot of different, disconnected, situations. Many social researchers conflate these two definitions of "pattern" in order to gain more sympathy for their cause. This is called, "the fallacy of equivication."
This false belief in the sanctity of confidential surveys has got to stop if the "soft sciences" ever hope to gain the respect of "real scientists." There are means to control for various forms of false reporting but it requires designing your study very carefully from the beginning; A lot more than just using follow up interviews (Who is going to say, "Yeah, I lied in that study my professor required me to participate in for part of my grade"?).
This study was a great start. But all it really shows is that there is something that needs to be investigated further. Not to say that steps shouldn't be taken to aleviate the problem, if and where it exists. Just that you can't claim to have anything more than a vague grasp of the problem without a better study.