Suppose you wanted to call attention to an issue. You'd want to define the statistics up for that particular problem. One way of doing so is to expand the definition. So take these examples.
hunger vs. food insecurity
If you are serious about hunger, then you might want to define the issue in terms of "food insecurity." That encompasses many more people, not just those are suffering from famine conditions, but malnourished people or those who might have to skip meals. Some of these food insecure people might be overweight, in fact.
By defining a population as "at risk," then you are expanding from people who are actually suffering from whatever it is, to those who are at risk of doing so.
Racism & Sexism
By defining racism in structural terms, we find that every white person becomes a racist just by benefiting from racism.
Now this might sound like a right-wing post, and that is not my intention, but the expansion of definitions has some unintended consequences. One of these is to muddy the waters by definitional elasticity (forgive the mixed metaphor.). Another is to trivialize real problem by throwing disparate phenomena in the same sack. Suppose we had a statistic that included both bank robbery and jaywalking, and said that "90% of respondents reported that they had robbed a bank or jaywalked in the past two years." That might be true, but you'd want to have mechanism for sorting out those two categories. Or if you asked: "Have you ever stolen money or a ballpoint pen from a bank?"
If we no longer distinguish between serious and less serious instances of the problem, then it becomes too difficult to treat the more serious offenses with the degree of seriousness that they deserve. So if we are really after bank robbers, then it makes sense to not have an expanded version of bank robbery, that also includes stealing the pen for the bank when you fill out your deposit slip.