In one Australian study, doctors who were asked how often they washed their hands, reported that their hand washing rate was 73%, whereas when these same doctors were observed independently, their actual rate was a mere 9%. The point being that when we self-report, the quality and truthfulness of our answers are often suspect. Maybe those doctors lied because they were concerned about their reputation and what others would think about them?
Which brings me to the question of surveys and ratings. Doctors doing satisfaction surveys on their patients, car dealerships asking customers about their car service, On-line reviews and ratings etc. None of these surveys or ratings are scientific. They sample a non-representative group (those who like or have the time to answer surveys) and they measure self-reported data. So basically, junk in, junk out. Sadly, once you convert the junk data into a pie chart, a percentage or even a star rating, it takes on a mystical significance. Now I would be the first to admit that a five star restaurant rating on Yelp might have some value and 92% rotten tomatoes for a movie, might have some street cred but that is about as far as it goes. It’s all about the numbers and who is creating them. If a physician has a 100 good reviews or just one bad one – either way it is a 100% good or bad review. But the former has more value, just so long as they were not all written by the same person. Incidentally we give every patient a survey to answer and our ratings are excellent.