⅔ of peer reviewed published scientific papers fail replication. That means whenever you read a "new study finds" headline, the claimed effect is twice as likely to be false as it is to be true. The effect is smaller in 'hard' sciences and most pronounced in psychology and public health fields.