In this case I'm thinking of two very different situations. First, international development program often want to target funds to those most in need so they need some way to indentify who should receive aid. A program could just ask people their income, but then there would an incentive to lie in order to meet the criteria to receive funds. Instead programs like conditional cash transfers often use a proxy means test. In a proxy means test program staff visit a house and see if it has indoor plumbing, a cement floor, a TV, a car ect. to see if the family actually needs the money. It doesn't work perfectly to figure out who needs the money perhaps a family was once relatively well off and had plumbing and a cement floor, but was now starving. If done right proxy means testing is pretty close and harder to fake how well off you are.
So totally unrelated, except for the fact that this is also a discussion of related things. But OK Cupid (a dating website), uses their data to see what answers to questions correlate well with others. When asking a potential mate if they like beer, they find "No matter their gender or orientation, beer-lovers are 60% more likely to be okay with sleeping with someone they've just met"
This could be a valuable question, because no one would mind being asked if they liked beer, but asking about sleeping with someone on the first date probably sends a bad signal and you might not get a truthful answer. On the other hand if everyone discovers the signal, then people may lie about their preferences in order to avoid sending the wrong signal.
Read more from OK Cupid here.