I feel as though the risk assessment line needs a little more nuance. It's not just the assessment of risk, but the ability/desire to assess long vs. short-term risk.
Teenagers and those in their early 20s are less able to assess long-term risk as their frontal lobes are still developing. This is particularly true if whatever is being assessed is also the subject of any kind of peer pressure.
Republican voters, and especially Trump supporters (and any other authoritarian followers), also tend to be less able to deal with long-term risk assessment. However, as noted in the article, for the elderly, the risk assessment is short term.
Republican voters are also more likely to be rural, so their interactions with other people are fewer. Given that other people are the vector for the virus, it seems reasonable to be less concerned about risk. What this fails to take into account is the mere 10 seconds of exposure that the Delta Variant seems to require, so, one doesn't need to be in Grand Central Station or packed in on a train for hours to get Covid, one just needs to be in the same bar, general store, or even passing on the street. The point being that, on any given day, the number of opportunities may be fewer and further between, but the cummulative effect of multiple short exposures is higher than the rural Republicans credit (which ties in well with the article's point about Republicans being "less realistic about Covid-19's harms in aggregate").
The Democrat-supporters may be wrong about the likelihood of suffering severe disease after contracting Covid, but equally, it's the resultant health behaviors that are reducing their chance of getting it at all. But let's not forget, the anti-vaxx movement started on the left.