Interesting methodology
While the surveys did not provide data for the same individuals at different stages, they represented snapshots of the changing attitudes of respondents in different age cohorts over time. The researchers corrected for the fact that the age groups at different survey times are made of up new members with unique baseline opinions.
I'd love to see how they pulled off that "correction" to make up for the fact that they didn't actually record any changes of opinion in one person.
As far as I can see, the only thing this might actually prove is that more recent generations of elderly are more liberal than older ones, which, given our policies, makes perfect sense.