In the first example, the research question is not simply interested in the daily calorific intake of American men and women, but what percentage of these American men and women exceeded their daily calorific allowance. So the dependent variable is still daily calorific intake, but the research question aims to understand a particular component of that dependent variable (., the percentage of American men and women exceeding the recommend daily calorific allowance). In the second example, the research question is not only interested in what the factors influencing career choices are, but which of these factors are the most important.
Even if you used a quantitative research design, but failed to employ a probability sampling technique, there are still many perfectly justifiable reasons why you could have made such a choice. For example, it may have been impossible (or near on impossible) to get a list of the population you were studying (., a list of all the 20,000 students at the single university you were interested in). Since probability sampling is only possible when we have such a list, the lack of such a list or inability to attain such a list is a perfectly justifiable reason for not using a probability sampling technique; even if such a technique is the ideal.