THE TECHNOLOGICAL OBSOLESCENCE OF SCIENTIFIC FRAUD

In a world where researchers prefer their experiments to have a partic ular outcome, scientific fraud and research bias are alternative metho ds to implement such preferences. A critical constraint distinguishes these two methods: biased research is 'narrowly replicable' while frau d is non-replicable. Recent proposals to severely punish fraud have ty pically excluded biased research from their purview. An unintended con sequence of such asymmetric sanctions is the substitution of creative uses of technology to bias research outcomes in favor of one's prefere nces. Importantly, such bias may be substantially more difficult to de tect than outright fraud, given current scientific conventions. This a rticle demonstrates, using the simplest of statistical examples, how b iased outcomes can derive from seemingly unbiased procedures.