Expression for misrepresenting statistics [duplicate]

Sometimes when a lot of data is gathered, an argument against the majority can rely on selecting the rarer case.

For example, a medical procedure which usually does more harm than good can be defended by statements from the few it has benefited.

Is there a word or expression for this? It could be "cherry-picking", but that doesn't signify the deception taking place.

Solution 1:

Cherry-picking is the exact expression used when discussing this problem, including when deception is present. It falls under the general header of QRPs (questionable research practices) in psychology and other sciences---often pertaining specifically to statistical analysis---and came to the fore during the so-called replication crisis which has made great waves in the past 5 years. See the Rationale section at https://en.wikipedia.org/wiki/Preregistration_(science)#Rationale.