The good news is it doesn’t really matter. The absolute numbers might be inaccurate, but the skew is likely to stay the same over decades. So it probably is dropping.
This is a huge assumption.
Edit: There's probably dozens of variables with how the survey is given that will skew the bias. They probably don't even ask the exact same question in the exact same way on the physical survey.
Is this your guess, or is there other conflicting data?