That aside, sobriety check points are one of the most successful LE operations going. In the 70s and 80s more than half of all traffic crashes were caused by impaired driving. The overall count (adjusted for population) is dramatically reduced and the proportion resulting from impaired drivers even more so. Check points work.
Do you have data to support this assertion and conclusion of causation?
Seems to me that a 1-1.5% detection rate(as noted in the Michigan decision linked earlier) is hardly "working." Much less able to prove causation for something...