Archive for bogus statistics

a perfect fit

Posted in Books, Statistics with tags , , , , on January 25, 2017 by xi'an

perfitHere is the table that kept most French commentators on the primaries busy on Monday, since the fit of the results at 10am to the results at the previous midnight is perfect, indicating the 10am results are not results… There was even a line in Le Monde about achieving this perfect fit being of order 10⁻³, “un miracle statistique de l’ordre d’une chance sur mille”, although no details of this arcane calculation are provided! I wonder why a journalist writing about nonsensical data feels obliged to add more nonsensical data to “support” one’s argument. [Thanks to Arthur Charpentier for sending me the data.]