Pretty small study for what they're looking at, and they don't report (because they didn't measure) whether left or right leaning people are more likely to misjudge people. They also got very even groups (about 60 each of r-men, l-men, r-women, and l-women) but didn't report on differences between genders.
I didn't look at the raw data but I wonder if these questions can be answered with it?
I'm not even sure that I find the methodology convincing. Based on answers to one set of political questions, they pair you to an "out-group" member and then ask you to predict an unrelated opinion. How did they choose the second question? Because a person's views on... gay marriage and trans rights would likely be similar, but people can have wildly different views on gun control and immigration (and exclusively, if they're a mythical single-issue voter.)
Seems messy, and I think the conclusion (we just can't know how other people think, so we should be more open and tolerant) is exactly the kind of wishy-washy "can't we all get along" crap that allows right-wing governments to mass kill people. Not of fan of manufacturing excuses for that.