This is really, really sloppy reporting. It appears that KQED just rebroadcast a press release without even looking at the paper.
1) this is not a peer-reviewed study; it is only a working paper.
2) the paper counts ONLY HOMICIDES (ignores collisions that do not result in homicide) and counts a collision as DUI-induced EVEN IF THE DRIVER IN THE WRONG IS STONE COLD SOBER. It also seems to not distinguish between drivers and passengers (they count number of involved parties, not number of involved drivers.
3} I can find no information that they used any controls, e.g. contrasted regions with similar demographics and tracking history of DUI that DO have Uber/Sidecar/etc. versus those who do NOT; . That would be the most meaningful data. DUI rates can drop for a given population by other factors, e.g. increased penalties, enforcement; concomitant decrease in homicides to due improved vehicle safety (collision strength, air bags, percentage of involved vehicles with air bags and such, etc.); significantly fewer bad-weather days in CA in the recent drought years, etc.
1) this is not a peer-reviewed study; it is only a working paper.
2) the paper counts ONLY HOMICIDES (ignores collisions that do not result in homicide) and counts a collision as DUI-induced EVEN IF THE DRIVER IN THE WRONG IS STONE COLD SOBER. It also seems to not distinguish between drivers and passengers (they count number of involved parties, not number of involved drivers.
3} I can find no information that they used any controls, e.g. contrasted regions with similar demographics and tracking history of DUI that DO have Uber/Sidecar/etc. versus those who do NOT; . That would be the most meaningful data. DUI rates can drop for a given population by other factors, e.g. increased penalties, enforcement; concomitant decrease in homicides to due improved vehicle safety (collision strength, air bags, percentage of involved vehicles with air bags and such, etc.); significantly fewer bad-weather days in CA in the recent drought years, etc.