Graziano & Raulin
Research Methods (9th edition)
Non-Equivalent Control Group Question #3
This may or may not be a situation in which the control group is well matched with the experimental group. Both cities are large and so are well matched on that variable. However, one might expect that one's emotional response may be a function of many variables, such as religious background, type of work, level of family support, and so on.
It would be best to identify as many of these potential confounding variables and routinely get measures of them in addition to the measures of emotional distress. By computing correlations between these potential confounding variables and the measure of emotional distress, the researcher can identify what variables may be most important to control. Since these potential confounding variables have been measured in each group, it is possible to see if the groups differ on the variable. If they do differ, it is possible to select groups that are matched on critical variables for comparison.