If they're testing them on GMO knowledge, and then extrapolating from that that they don't know science in general, then that's a leap in logic.
There's nowhere that it says " they're testing them on GMO knowledge". The article actually uses the phrase "science in general" for the content of the test, funnily enough. Now, you're asking whether such a test can be extrapolated to "science in general" which is hard to answer. Is it a leap in logic that a test of "science in general" can determine whether you know "science in general"? Only if we get really philosophical about it, I think.
Second, "know the least" is pretty clear. if people of a specific group scored the least correct answers on a "science in general" test then it's not much of a stretch to claim that people of this group generally "know the least" about "science in general".
You're saying "know the least" is problematic because it could apply to anything. But that's just deliberately being a jackass by ignoring the obvious context. For example if an article title was "people who don't listen to much music tend to
know the least" then it's just being a jackass to argue "know the least about
what exactly?" About
music, duh. That much is goddamn obvious. If the topic is science, "know the least" clearly refers to science, the
noun in the same sentence.
and this can be more pronounced when those fifteen questions are divided up into multiple fields, as the article seems to imply
But wasn't your first point about how the test was
too narrow? Now it's
too broad apparently.
the higher the chance that they might end up falling into a "gap" in an otherwise-educated person's knowledge
That's why they had 2000 subjects. While with 15 questions, it's possible for what you said to happen ... for it to happen to 2000 people in a row in a way which
systematically biases the results against "otherwise educated" anti-GMO people is astronomically unlikely.