I obviously agree that it's unlikely to have been a test.But how on earth could they factor in guest that don't complain but still didn't like it? Or at least ones that complained to the CM and not formally? Does the CM report "I got yelled at four times and three different parents approached me" to a superior? Unless the results of a 'test' yield results or can be measured, then there is no reason to test anything. In fact, without someone there to actually record the results, I doubt it was a test at all. Just a CM on a power trip.
But statistical measures are always an approximation, never perfect. They just do what they can to filter out distortions. If they know from past experience that on a customer satisfaction survey, 10 out of 100 people disliked something strongly, but in reality they only got 1 complaint directly to Guest Services from every 100 people, then they know that every 1 complaint to Guest Services means 10 dissatisfied people. And with the vast number of people that Disney serves, they probably get a really good handle on things from the Guest Services complains without measuring ever single informal complaint to a CM (because with a large set of samples, statistics can be really reliable).