Everything the NIDA says about cannabis should be taken with a rather large grain of salt.

Cannabis does cause some impairment, however studies have shown that consumers tend to overestimate this impairment, and that they compensate for it with added caution. Alcohol tends to do the opposite, consumers perceive their impairment to be less that what it actually is and often become overconfident, aggressive, and careless. [Robbe and O'Hanlon. 1993; Robbe. 1995]

To find out how cannabis use affects crash risk overall, in 2015 the U.S. government completed the largest case controlled study to date regarding DUI of cannabis and crash risk. It involved over 9,000 cases and controls spanning a 20-month period. It found that cannabis use while driving is not associated with increased crash risk once adjusted for confounding variables such as age, race, gender, and the presence of other drugs, including alcohol:

"This analysis shows that the significant increased risk of crash involvement associated with THC and illegal drugs shown in Table 3 is not found after adjusting for these demographic variables."

Further, they found that cannabis did not add to the crash risk for drivers under the influence of alcohol:

"As was described above, there was no difference in crash risk for marijuana (THC)-positive drivers who were also positive for alcohol than for marijuana (THC)-positive drivers with no alcohol, beyond the risk attributable to alcohol."

They found that alcohol significantly increased crash risk:

"at moderate alcohol levels (0.05 BrAC) risk increases to double that of sober drivers, and at a higher level (0.10 BrAC) the risk increases to five and a half times. At a BrAC of 0.15, the risk is 12 times, and by BrACs of 0.20+ the risk is over 23 times higher."

Compton and Berning. DOT HS 812 117. Drug and Alcohol Crash Risk. U.S. Department of Transportation - National Highway Traffic Safety Administration. 2015.

That said, at some point one could be high enough to significantly increase crash risk, something roughly the equivalent of 0.08% BAC, and DUI laws should reflect that, and be based on actual impairment, not "per se" limits. However it is rare for anyone that high to want to actually drive a car, whereas it is commonplace for someone very drunk to attempt to drive.


SOURCES:

--Robbe and O'Hanlon. DOT HS 808 078. Marijuana and actual driving performance. U.S. Department of Transportation - National Highway Traffic Safety Administration. 1993.
--Robbe H. Marijuana’s effects on actual driving performance. HHMRC Road Research Unit, University of Adelaide. 1995.


Colorado legalized recreational cannabis in Dec 2012 (Jan 2014 for retail sales) and did not see a statistically significant change in fatal traffic accidents:

2012: 474 (Population: 5.19 million, 0.0091%)
2013: 481 (Population: 5.27 million, 0.0091%)
2014: 479 (Population: 5.36 million, 0.0089%)
[SOURCE: Colorado DOT & "As Reported" to NHTSA by FARS]

Recreational cannabis use was legalized in Washington state in Dec of 2012. Retail sales began in July, 2014. Fatal traffic accidents have not increased:

2012 - 403 (Population: 6.90 million, 0.0058%)
2013 - 401 (Population: 6.97 million, 0.0057%)
[SOURCE: Washington State Department of Transportation - 2012, 2013 Annual Collision Summary]