I'm confused. That proves his point. Every state at a secondary enforcement seat belt law before 1998. Every state aside from Maine and South Dakota (both of which had relatively high seat belt usage before the law) saw an increase in seat belt usage post-law.

I think you might be confused by the 1998 number. That year came after the after law percentage. Meaning that enforcement immediately after the law also worked.

No, the law only brought the 17% of change. The law change is binary and sudden, but the change in culture brought all the rest, gradually, year after year, before law and after.

You are making assumptions based on zero evidence. The change is binary, enforcement is over time. You are assuming the change from 48% to 65% (which is also a 17% increase) can be explained sans law, but there's zero evidence that's the case.

Your assumptions are without standing.

The change from 31% to 48% compliance represents a 54.84% increase, not 17%.

In the interest of not citing 12+ year old studies relying on 17+ year old data, 2013 usage was 87% and since 1995 there have only been two years of decreasing average usage[0], '05-'06 (82-81%) and '10-'11 (85-84%).

[0] http://www-nrd.nhtsa.dot.gov/Pubs/811875.pdf

To truly see any trend, we would need to look at not just before the law, but many years before the law.

This is one of those sneaky tricks advocates for government intervention use (or fall victim to). To prove the efficacy of OSHA, for example, proponents will show a chart of workplace fatalities or accidents where there are a few or only one data points before the law and then show some spectacular trend that follows after the law was passed. In reality when you go back many more years it is clear the law had little to do with the trend as workplace safety had been increasing for years prior to the law.