ABCDE
1
ItemValuePriority for Further Investigation?SourceNotes
2
Time Spent (hours)1.4Does not include time spent fulfilling additional small bounty to write up my calculations in words (+~15 minutes).
3
PromptFB$30 to do a small research project into the micromorts from driving.
Wikipedia has a 1 micromort/230 miles which seems high, but maybe not.
I know that most fatal accidents occur on adverse conditions, e.g. people being drunk.
Can someone do a data-backed estimate saying that if you're not an idiot about driving (i.e. not drunk, not playing with your phone), what is you actual micromorts/mile driven?

I don't think it should be large, just looking up driving fatality statistics and combining them accident-situation stuff (if that hasn't been done already).

Caveat: I'll want to share the results online (LessWrong post).
4
Response to PromptFBI’d do up to an hour on this for $30, and spend up to 1.5 hours if I get particularly interested.
5
Micromort definitionA micromort is a unit of risk defined as one-in-a-million chance of death
6
7
Global annual car crash deaths1,350,000ASIRT
8
US annual car crash deaths38,000ASIRT
9
US annual fatalities due to work-related crashes (2018)1276CDC
10
11
US annual car crash deaths (2019)36096NHTSA
12
Share of deaths on motorcycles (2019)14%NHTSA
13
Share of deaths to non-occupants (2019)20%NHTSA
14
US passenger vehicle deaths (2019)22215NHTSAIncludes passenger cars and light trucks (SUVs, Pickups, Vans)
15
16
Front-seat belt use for adults in passenger vehicles (2019)90.70%NHTSA
17
Deaths amongst those using seat-belts in passenger vehicles (2019)10815NHTSA
18
Deaths amongst those not using seat-belts in passenger vehicles (2019)9466NHTSA
19
Deaths with unknown seat-belt use in passenger vehicles (2019)1934NHTSA
20
Estimated deaths if all wore seat-belts in passenger vehicles (2019)11924CalculatedFor this, I assign the fatality rate in all cases to the seatbelt-wearing rate. This may not be a very valid action for this estimation, for example, perhaps those who do not wear seatbelts drive more/less safetly than those who do, then we'd expect an associated lower/higher death likelihood in crashes if seatbelts were worn.

I perform this calculation by multiplying the number of deaths amongst those using seat-belts in passenger vehicles by 1 / the percent of adults that wear seat-belts in the front-seats of passenger vehicles.
21
22
US daytime deaths (2019)19133NHTSA
23
US nighttime deaths (2019)18597NHTSA
24
25
US passenger vehicle occupant deaths in single-vehicle crashes9973IIHS
26
US passenger vehicle occupant deaths in multiple-vehicle crashes12443IIHS
27
Estimated deaths in multiple-vehicle crashes if all wore seat-belts in passenger vehicles (2019)6619CalculatedFor this, I:
1. Assign the fatality rate in all cases to the seatbelt-wearing rate. This may not be a very valid action for this estimation, for example, perhaps those who do not wear seatbelts drive more/less safetly than those who do, then we'd expect an associated lower/higher death likelihood in crashes if seatbelts were worn.
2. Assume the share of multiple-vehicle crashes vs. single-vehicle crashes is the same regardless of seatbelt-wearing status. This may be invalid for similar reasons as #1.

To calculate this, I take the previously calculated figure of the "Estimated deaths if all wore seat-belts in passenger vehicles" and multiple it by the percent of deaths that are attributable to multiple-vehicle crashes (vs. single-vehicle crashes).
28
29
US fatalities in which at least one driver was distracted (distraction-associated) (2019)3142NHTSA
30
US fatalities in which at least one driver was drowsy (drowsy-associated) (2019)697NHTSA
31
US alcohol-impaired-driving fatalities (2019)10142NHTSA"An alcohol-impaired-driving fatality is defined as a fatality in a crash involving a driver or motorcycle rider (operator) with a BAC of .08 grams per deciliter (g/dL) or greater."
32
Estimate for US distraction-associated fatalities for passenger vehicle occupants in multiple-car crashes only1073.4CalculatedAssumes distraction-associated fatalities are proportionally distributed by vehicle type, type of fatality, and number of vehicles in crash.

This is calculated by taking the total number of distraction-associated driving fatalities, multiplying it by the percentage of all fatalities that are attributable to passenger vehicles, and multiplying that result by the percentage of passenger vehicles occupant deaths that are attributable to mutliple-car crashes.
33
Estimate for US drowsy-associated fatalities for passenger vehicle occupants in multiple-car crashes only238.1CalculatedAssumes drowsy-associated fatalities are proportionally distributed by vehicle type, type of fatality, and number of vehicles in crash.

This is calculated by taking the total number of drowsy-associated driving fatalities, multiplying it by the percentage of all fatalities that are attributable to passenger vehicles, and multiplying that result by the percentage of passenger vehicles occupant deaths that are attributable to mutliple-car crashes.
34
Checking if Table 8 and Table 9 use same criteriaFALSENHTSA
35
Estimate for US alcohol-impaired fatalities for passenger vehicle occupants in multiple-car crashes only3464.8CalculatedAssumes alcohol-impaired fatalities are proportionally distributed by vehicle type, type of fatality, and number of vehicles in crash.

This is calculated by taking the total number of alcohol-impaired driving fatalities, multiplying it by the percentage of all fatalities that are attributable to passenger vehicles, and multiplying that result by the percentage of passenger vehicles occupant deaths that are attributable to mutliple-car crashes.
36
Estimate for distration-associated, drowsy-associated, and alcohol-impaired fatalities for passenger vehicle occupants in multiple-car crashes only, if all wore seat-belts (2019)2563.7CalculatedAssumes seatbelt-wearing and non seatbelt-wearing drivers have similar baseline risk of fatality in crashes other than the presence of seatbelt-wearing, and assumes that distraction, drowsiness, and alcohol-impairment associated fatalites are proportionally distributed by vehicle type and number of vehicles in crash.

This is calcualted by first adding the estimates for distraction-associated, drowsy-associated, and alcohol-impaired fatalities generaged in previous calculations for passenger vehicle occupants in multiple-car crashes. This amount is then multiplied by the estimated percentage of passenger vehicle deaths that would still occur if all wore seat-belts.
37
Estimated deaths amongst passenger vehicle occupants only, if setting single-car crashes to 0 and approximating the seatbelt-wearing only rate and if setting alcohol associated and distracted deaths to 50% of current level (as an approximation of controlling one driver's behavior in two driver crashes)5337.1CalculatedThis is calculated by taking the previously estimated deaths in multiple-vehicle crashes if all wore seatbelts in passenger vehicles, and subtracting 50% of the estimate for distraction-associated, drowsy-associated, and alcohol-impaired fatalities for passenger vehicle occupants in multiple-car crashes only, if all wore seatbelts.
38
39
Fatality rate per 100 million vehicle miles traveled (2019)1.10NHTSA
40
41
Total vehicle miles in US (2019)3,266,400,000,000BTS
42
Total motorcycle miles in US (2019)19,688,000,000BTS
43
Light duty vehicle, short wheel-base + light duty vehicle, long wheel-base vehicle-miles in US (2019)2,924,053,000,000BTS"The new category Light duty vehicle, short wheel base replaces the old category Passenger car and includes passenger cars, light trucks, vans and sport utility vehicles with a wheelbase (WB) equal to or less than 121 inches. The new category Light duty vehicle, long wheel base replaces Other 2-axle, 4-tire vehicle and includes large passenger cars, vans, pickup trucks, and sport/utility vehicles with wheelbases (WB) larger than 121 inches"
44
45
Total vehicles miles in US (2019)3,281,454,545,455CalculatedTo calculate this, I divide 2019 vehicle deaths by the fatality rate per 100 million vehicle miles, and multiply that by 100 million vehicle miles. This was later validated by being approximately equal to the figure explicitly provided by BTS.
46
Total micromorts experienced due to driving in US (2019)36,096,000,000CalculatedThis calculates the number of micromorts attribted to vehicles in 2019 in the US, by multiplying 1 million * the number of vehicle deaths in the US in 2019.
47
48
Miles per micromort, no adjustments, in US (2019)91CalculatedSee cells referenced in calculations for notes and sources. This calculation takes the total number of vehicle miles driven in 2019 in the US, and divides it by the number of micromorts attribted to vehicles in 2019 in the US, which is calculated as 1 million * 2019 vehicle deaths.
49
-- If excluding motorcycles105CalculatedThis does the same calculation as the previous row, except it subtracts motorcycle miles and motorcycles deaths from the figures.
50
-- If excluding motorcycles and pedestrians, pedalcyclists, and other nonoccupants137CalculatedThis is a continution of the same approach, but (in addition to the previous motorcycle adjustment) it also removes the 20% of vehicle-caused deaths that are incurred by pedestrians, pedalcyclists, and other nonoccupants of vehicles. This uses a simplifying assumption that motorcycles are not responsible for any pedestrian, pedalcyclist, or non occupant deaths.
51
-- Amongst passenger vehicle occupants only132CalculatedThis calculation takes the total number of passenger vehicle miles driven in 2019 in the US, and divides it by the number of micromorts attributed to passenger vehicles in 2019 in the US, which is calculated as 1 million * 2019 passenger vehicle deaths.
52
-- Amongst passenger vehicle occupants only, if setting single-car crashes to 0235CalculatedThis calculation takes the total number of passenger vehicle miles driven in 2019 in the US, and divides it by the number of micromorts attributed to passenger vehicles in 2019 in the US in multiple-vehicle crashes, which is calculated as 1 million * 2019 multiple-vehicle passenger vehicle deaths.
53
-- Amongst passenger vehicle occupants only, if approximating the seatbelt-wearing only rate 245CalculatedThis calculation takes the total number of passenger vehicle miles driven in 2019 in the US, and divides it by the number of micromorts attributed to passenger vehicles in 2019 in the US if all wore seatbelts (previously estimated), which is calculated as 1 million * 2019 estimated passenger vehicle deaths if all wore seatbelts.
54
-- Amongst passenger vehicle occupants only, if setting single-car crashes to 0 and approximating the seatbelt-wearing only rate442CalculatedThis calculation takes the total number of passenger vehicle miles driven in 2019 in the US, and divides it by the estimated number of micromorts attributed to passenger vehicles in multiple-vehicle crashes in 2019 in the US if all wore seatbelts, which is calculated as 1 million * 2019 estimated passenger vehicle deaths in multiple-vehicle crashes if all wore seatbelts.
55
-- Amongst passenger vehicle occupants only, if setting single-car crashes to 0 and approximating the seatbelt-wearing only rate and if setting alcohol-impaired, drowsiness-associated, and distraction-associated deaths to 50% of current level (as an approximation of controlling one driver's behavior in two driver crashes)548CalculatedSetting alcohol, distration, and drowsy deaths to 50% is only a useful approximation if multiple-vehicle crashes average 2.0 involved vehicles.

This calculation takes the total number of passenger vehicle miles driven in 2019 in the US, and divides it by the estimated number of micromorts attributed to passenger vehicles in multiple-vehicle crashes in 2019 in the US if all wore seatbelts and when setting three preventable classes of impairment at 50% of their rate, which is calculated as 1 million * Estimated deaths amongst passenger vehicle occupants only, if setting single-car crashes to 0 and approximating the seatbelt-wearing only rate and if setting alcohol associated and distracted deaths to 50% of current level (as an approximation of controlling one driver's behavior in two driver crashes).