mknewman writes "Between 2006 and 2008 about four dozen physicists buried 19 Germanium-based detectors and 11 silicon-based detectors deep in a mine in Minnesota. They believed the Germanium detectors might be just right to capture the rare, but theoretically possible collision between a WIMP and an atomic nucleus. The silicon detectors were just there to confirm the result — i.e. if a Germanium detector recorded such a collision and a silicon detector did not, that would be good evidence for a WIMP.
After taking their data for three years the scientists got a ho-hum result — the Germanium detectors recorded two events, when on average they would have expected to see 0.9 events during the time period. This was not statistically significant, and moreover, they later concluded these events were attributable to the leakage of electrons.
Since the primary detectors showed no significant results, data collected by the silicon detectors, which could only detect WIMPs up to a mass of about 15 GeV were not analyzed.
Then, after some considerations, the physicists came to believe that maybe the WIMPs weren’t really, really big. So they went back and studied the silicon detector data and found three events, when they would have expected just 0.7 events during the time period of data taking. This is statistically significant.
So they published their results on Monday (see paper). Based upon their statistical analysis, they are 99.8 percent sure they have observed some WIMPs at a mass of about 8 GeV. But in particle physics, certainty doesn’t come until they are 99.9999 percent sure."
Link to Original Source