Not the First Digit! Using Benford's Law to Detect Fraudulent Scientif ic Data

Andreas Diekmann
2007 Journal of Applied Statistics  
standard by checking digit frequencies in published statistical estimates. Second, we conducted experiments in which subjects were asked to fabricate statistical estimates (regression coefficients). These experimental data were scrutinized for possible deviations from the Benford distribution. There were two main findings. First, the digits of the published regression coefficients were approximately Benford distributed. Second, the experimental results yielded new insights into the strengths
more » ... to the strengths and weaknesses of Benford tests. At least in the case of regression coefficients, there were indications that checks for digitpreference anomalies should focus less on the first and more on the second and higher-digits.
doi:10.1080/02664760601004940 fatcat:tuw3otazezgcbe4fgs762kqrym