Yes, they used "_do you consider yourself_ underrepresented". And there's also the implicit "and _do you feel like this matters_ for this survey". Some people may feel underrepresented but don't feel like this should matter on this survey, so they'd be likely to answer "No".
So a decrease in the "underrepresented" statistics doesn't necessarily mean it's dealing with the actual situation.. It wouldn't necessarily imply anything "bad" or "undesired".
And this is why every survey I've ever taken has pissed me off and now I don't take them at all. I've never felt that the information gleaned from a survey was honestly useful for making decisions.
This is typical of political surveys and other surveys where there is a "preferrable" outcome, but I don't think in this case they are trying to get a specific outcome. Maybe feedback can make subsequent surveys better and reduce the ambiguity of results.
It's typical of all surveys. You either need to go into extensive detail to remove the ambiguity of the questions, or you need to allow people to state what they think in plain language. Taking surveys to gather information is like using record sales to measure what kind of music is best; there are a billion confounding factors and none of them are covered by the abjectly simple metric used.
So a decrease in the "underrepresented" statistics doesn't necessarily mean it's dealing with the actual situation.. It wouldn't necessarily imply anything "bad" or "undesired".