... though you should take anything I say as general information rather than advice specific to your issue.
The DMCA safe harbor provision is intended to cover user-generated content posted to your website by a third party. The idea is that you, as a website operator, should not responsible for someone else's conduct so long as you take it down.
There is also case law (Lenz v. Universal) out of a California District Court holding that a copyright owner has an obligation to consider whether or not a particular use of its work is fair use *prior to sending a DMCA takedown notice,* and that failing to do so could constitute bad faith abuse of the DMCA.
Sending a counter-notice, as you appear to have done, is definitely a good way to go especially if you believe that the reproduction of approximately 13 percent of the MMPI constitutes fair use. However, you seem to be in an odd situation. The normal case is for the website operator (you) to take down the post and notify the original author (the user). Then the user decides whether to file a counter-notice.
That being said, whether or not copying the questions constitutes fair use depends largely on a few things, which are embodied in the four-factor test that someone has surely mentioned in a post by now.
As a practical matter, it's all about context. Did you do anything other than just posting the questions? Showing that you used the questions as part of a larger work where you contributed your own thoughts and expression would make your fair use argument stronger (it's not copyright infringement - it's citation!). If all you did was copy the questions and post them without much of what courts call "transformative" use (i.e. creative input on your part), then that weakens your argument.
Looking at the Google Cache of the original post, it seems to consist of a brief introduction:
"I have done quite a bit of research on the MMPI 2 used by psychological testing. Here's the first 75 out of 567 questions. I could give out rest of them & how some of them are interpeted by psychologists for advice on admitting past history on the psych & medical tests. See my post on "Help with Admissions for Psych & Medial""
And then the rest is the questions. So the issue is whether or not the user-contributed content is enough.
Computers are unreliable, but humans are even more unreliable. Any system which depends on human reliability is unreliable. -- Gilb