What does it mean to be the BEST at something? Today's TBF looks at a common "best" claim that you have probably seen: the best selling author.
"More than 80% Of Dentists recommend Colgate"
"Colgate, used and recommended by most dentists"
If this TBF intrigues you, you can read more about it here.
True?: Yes, Colgate ran a survey of dentists and 80% of those surveyed identified Colgate as a brand they'd recommend.
Intention? ...to convince you that dentists prefer Colgate over other brands.
False?: You most likely think that the survey question asked dentists which one brand they prefer, but the actual survey allowed the dentists to select multiple brands. In other words, Colgate was just one of potentially several that each dentist identified. This means that 80% of dentists selected Colgate, but 80% might have also selected Crest! In fact, this is what happened. One of Colgate's competitors was selected just as much as Colgate was in their own survey. The data science moral today is that you need to be transparent about the context in which your data was collected (in this case, a multiple-selection survey).
There are other questions of reliability of the data, such as Colgate designing the survey themselves, but this TBF was so blatantly misleading in 2007, that the UK courts ordered Colgate to stop saying it.