2: A significant positive correlation between the incidence of mental illness and the incidence of divorce among middle class
people is likely to mean that

D: all of the above may be true

    Correct.

In this case, all of the previous choices are possible explanations for the finding in question 2.  Since causation can NEVER be infered from a correlation, we are forced (at least in this question) to hypothesise potential reasons for the finding.  Each of the selections is a possible factoral, so D, all of the above is correct.  The point of the question is that a correlation study cannot prove that A or B or C is correct.