on 'mischievous responders'

so this article talks about how some measure of mental health of queer teenagers might be skewed, because teenagers find it amusing to lie on surveys and that includes saying they're gay when they're not and saying they use drugs more than they do.

They say their current methods to detect 'mischievous responders' require 'supercomputers' to work and so they are trying to work on discouraging it from happening in the first place, but they also say they were likely to do things like report a height of 10000 feet or, if provided with a drop-down rather than a text box, just pick the highest or lowest possible value

doesn't this straightforwardly provide a method of detection? just put a bunch of questions on your survey about things with a known distribution (hopefully un- or weakly correlated ones) and use them as bait. if your survey-answerer is in the 99.9th percentile for height and has improbably declared that they had three kids at age fifteen... just throw them out. you can get a pretty good estimate of how many 'legit' answers you are throwing out if you know the distribution, and it doesn't require a supercomputer to do that.