In any event, spanking and child abuse is like nudity and obscenity, we know it when we see it, so to speak. Statistical analysis tends to be too blunt a tool to analyze all the shades and subtleties of human interaction, heck, you could massively traumatize a kid by not spanking them and telling them every time that they were too weak and frail to sustain the kind of punishments 'we had back in my day' etc.

They should be interviewing adults, not kids. Most adults are less likely to lie about their spanking level, whereas an adult talking about their own children or a child talking about their own disicpline is very inclined to be false, intentionally or otherwise. Since your average adult dates to a time when spaking was entirely acceptable, they are unlikely to feel queasy about answering yes on something they also intellectually know is anonymous anyway, whereas most kids are going to have at least some sort of paranoia they might be being personally analyzed, so you could easily end up with one group false reporting for fear of their parents getting into trouble and another group massively rounding up for sympathy and attention, or general pranskter tendencies.