In a hard-disc drive, a defect region on the hard disc is detected by generating two statistical measures (e.g., β1(k) and β2(k)) based on signal values (e.g., x[n] or y[n]) and soft-decision values (e.g., L[n]) corresponding to the signal values. The measures are compared to detect the location of the defect region of the hard drive. Using the soft-decision values reduces fluctuations in a ratio of the statistical measures compared to a ratio formed from statistical measures that are not based on soft-decision values, resulting in a more-reliable test for detecting defect regions.