The "since 1982" statistic, unless there's something I'm missing, is literally confusing correlation for causation.

Your other quote on education has a strange emphasis on "short term" changes, especially given that the part regarding bans is talking on the order of decades. Presumably that is a long term effect, yes?

That paper talks a lot about changing social norms and increasing public support for laws. So if laws pass with broad public support, then presumably that broad public support is indicative of a change in social norms which confounds the data. In the end the drink-driving issue is a bad example for this kind of discussion of bans because it's not banning things that the public broadly would otherwise want to do.

Also, the logic that the "high-risk-but-hard-to-reach" group won't be reached by education also supports the notion that they won't be reached by laws either. It makes this point:

Various studies, mostly of male populations, have noted the interrelationship among certain personality traits (rebelliousness, risktaking, independence, defiance of authority ), deviant driving practices (speeding, drinking and driving), and crashes and violations. Deviant driving and crash involvement have also been found to be related to a syndrome of problem behavior including marijuana use, heavy alcohol use, smoking, trouble with the law, and various other delinquent behaviors.

The obvious thing that would reach people like this is social pressure, which again is something that requires broad social support, which confounds any notion that bans have any real effect.

Sorry, but you have a bunch of sources but they don't seem to say what you want them to say.