Rebecca Goldin on the Misuse of Statistics in Science

[MrBulltrout]

I love how she starts with the observation that there is a lot of politics in science. Rebecca Goldin cannot be trusted when it comes to reporting on matters of concern to the public. She is director of the Statistical Assessment Service and is a colleague of journalist/ industry hack Trevor Butterworth. Both of them frequently commit many of the errors she is warning us against. They operate under a cloak of objectivity and scientific legitimacy, but they are biased and untrustworthy.

[tardifnk]

Unfortunately, there's not much science in politics.

[ProfessorWraith]

Dr. Goldin is correct that the mainstream media are prone to report the advocacy of marginal groups as "scientific" when, in fact, science is a process of building consensus through repeated experiments, peer review, and long-term consideration of results obtained.

However, the problem goes far deeper. Too many academics publish results derived from questionable, if not outright bad, statistical analysis, and this has become worse as stats programs (like SPSS) have democratized data analysis.