If an evacuation order has been issued, your original plan is pretty good. It may be that a potential hazard could quickly turn in a major disaster, and the authorities want everyone out of the danger zone ASAP.

Otherwise, without knowing the location of the spill, what has been spilled, and a clear indication of wind direction, hopping in the car is potentially a death trap. A toxic, asphixiating, or explosive cloud will knock out the car and its occupants. I guess if you can move fast and know the wind direction, it's a personal choice.

For sheltering in place, anything that can be done beforehand to reduce air movement between the interior and exterior of a building would be time well spent. Caulk is cheap (haha). A plan to kill all HVAC and ventilation fans, shut of natural/propane gas (if any) and retreat to a safe room can be executed very quickly. For the safe room, pre-cut plastic sheets and duct tape applied to the inside of doors and windows can seal things up very quickly.

I now wonder how airtight a house has to be in order to be safe in case of a chemical spill. We all know that houses aren't constructed to be airtight, and no amount of caulking will ever get us there, especially if the house was built 100 years ago or earlier. Most of us will have to make a bubble (or a box) out of plastic sheets in order to be close to airtight.

The reasons I'm asking is because people seem to be a great deal of trust in the construction of houses. What am I missing? Is there some sort of standard for "safe for spills" such that some gas leaking into your house is OK? Is there some standard for testing how airtight a house is?

Its probably not a black and white safe vs not safe, its more like the more airtight the house the more safe it is. There isn't a clear dividing line.

Wonder what it would take to get filtered air in to create a slight positive pressure

I think Eugene has it right. In a crisis, a building that is as closed off as possible provides a blob of static air for the inhabitants while a blob of toxic gas passes overhead.

Edit: There are widely used test methods for airtightness. They are associated with energy audits. Basically they replace your front door with a giant blower to find/measure air leakage. An IR image of a home could also be instructive.

I'm looking for a more precise, actionable answer, but it's possible no such answer exists. After all, if you think about it, many beautiful old houses leak like a sieve. FEMA recommends plastic sheeting and duct tape because that's what people can do, rather than what would be best for your survival. What are you going to tell the owner of the old house that has been demonstrated to be leaky? Sell the house and move out because there is a one in a million chance they might be in a train spill?

Doug and our neighbors to the north may be lucky in that their houses are probably more tightly constructed and insulated than the houses, say, in the American South, where energy bills are high and unintended ventilation is almost a feature and not a bug...

I started thinking about respirators when I saw recommendations from someone working in some sort of hazmat-related industry for the type of respirators that would be suitable for that kind of work. I looked them up, and man are they expensive -- like four digits expensive! When I looked into what prep sites recommend, I was not filled with confidence. I think this sort of stuff -- what counts as a good respirator and what home prep for spills -- may need more study.

Or maybe even leaky old houses are good enough in an emergency?