Serious preventable accidents happen at 60+ MPH. Generally a slow 20-45 mph (35 mph average) drive around the neighborhood is much safer as there are almost no "serious" accidents.
Source: an organization with a fleet installed some driver tracking gps's and looked at the data over a year or two.
Most car seats are improperly installed. If you have one, get its installation checked by a professional.
In the SF Bay Area, you can have CHP check it for free. The full-time job of the officer that checked ours is investigating accidents where kids in car seats died. Almost all the child fatalities around here are due to improper car seat installation.
We had it dangerously wrong. Before the appointment, I read the manual twice and spent something like 60 minutes installing it. The seat used the latch system with a top anchor, it was a new car and we had a top-ranked car seat. All the straps were connected up correctly. During installation, I sat on it with all of my weight, pushed against the ceiling of the car with my back, pulled the tightening strap until it felt like it'd break, and then pulled harder. The straps were still too loose. I'm not small.
Our last car seat had a failure mode where one of the latch straps would just self-release every 1000 miles or so. I'm not sure if it was user error or not. It happened twice.
After that experience, I think part of the certification process for car seats and cars should involve proving that the majority of people that have never installed a car seat (i.e., first time parents) can get it right on the first try without reading the manual. The test group should include people with physical impairments, and should include at least a dozen popular car models going back a decade. (Car certification should be the same, but with multiple car seat models.)
100% this should be mandatory reading for all new parents.
Since you mention only SF, for anyone else who happens to read this, just go to your local Fire Department. They will 100% either know how to secure it for you or guide you to the correct place.
Apparently the NHTSA stats aren't different for babies (likely because babies are usually heavily restrained, and probably brought out in a car less e.g. don't need school runs). Below are some US NHTSA publications with detailed answers, but here's a loose paraphrase a) it's dangerous being a nonoccupant in a driveway: nontraffic backover/backing crashes caused ~460 deaths in 2008 (~300 nontraffic + 160 traffic), see [3].
b) reported fatalities per year age of child seem to be roughly constant, i.e. babies do not exhibit a higher rate of traffic deaths (probably because they're typically restrained in carseats). See [1] p5: Table 1. Passenger Vehicle Occupants Involved in Fatal Traffic Crashes, by Survival Status and Age Group, and Restraint Use, 2021. In fact, the peak is for children passengers 13+ or 15+, not babies.
c) (These do not report crash speed, or separate freeway accidents from residential roads (posters here asking about "a drive around the neighborhood"), and you'd expect people would drive babies slightly slower.) NHTSA only separately reports nontraffic accidents (e.g. parking lots). Or non-crash deaths such as hyperthermia.
Key Findings
• Of the 42,939 traffic fatalities in 2021 in the United States, 1,184 (3%) were children 14 and younger.
• An estimated 162,298 children were injured in traffic crashes in 2021, a 17-percent increase from 139,058 in 2020.
• Of the 26,325 passenger vehicle occupants killed in 2021 in traffic crashes, 863 (3%) were children. Of these 863 child passenger vehicle occupants killed in traffic crashes, restraint use was known for 769, of whom 308 (40%) were unrestrained.
• Of the 1,184 children killed in traffic crashes, an estimated 294 (25%) were killed in alcohol-impaired-driving crashes in 2021.
[2] From NHTSA: NCSA (National Center for Statistics and Analysis) Motor Vehicle Traffic Crash Data Resource Page https://crashstats.nhtsa.dot.gov/#/
A backover is a crash which occurs when a driver reverses into and injures or kills a nonoccupant such as a pedestrian or a bicyclist. Backovers can occur either on a public roadway or not on a public roadway, i.e., in a driveway or in a parking lot. The former are called traffic backovers and the latter nontraffic backovers. There are also “other backing crashes” that are not backovers, i.e., they do not involve a pedestrian or other nonoccupant, that occur when, for example, a driver backs into a tree or pole or when a driver backs out of a driveway or parking space and is struck by another vehicle. Together, backover crashes and other backing crashes are referred to as backing crashes."
Look at the table on page iv.: Fatalities and Injuries, by Type of Crash, Type of Vehicle
Source: an organization with a fleet installed some driver tracking gps's and looked at the data over a year or two.