Hi superbutros!
That's a great question, and it's good to see that you're giving such a keen eye to analyzing your strengths and weaknesses on the test.
Ultimately, I can only take a stab at suggesting what might be the cause of scoring worse on those types of questions. This PowerScore article, "How and Why Difficulty Varies" (see:
https://www.powerscore.com/lsat/help/ls ... sition.cfm) discusses question difficulties for a handful of tests (Feb. 1996, 1997, 1999, 2000) that LSAC released with question difficulty levels and their explanations.
As you can see from the table in that article, it seems like the average question is around level 3. And that seems to make some sense to me--for example, the first two levels are less likely to help differentiate test takers (i.e., for purposes of plotting them along the 120 to 180 scale) because they're easier, and a level 4 question similarly might not help differentiate test takers if most people get them wrong. By contrast, level 3 seems to be a middle ground, a challenging type of question that enough people are getting right, and enough people getting wrong, that it makes for a good type of question.
So at least from the figures suggested in that post, perhaps one reason that you might be getting the most wrong in that category is because its the most common difficulty type. If that's right, then there could be a variety of explanations. For instance, you note that level 4 questions have a small sample size, so maybe the 100% isn't indicative--but if you're noticing a pattern that you're still getting proportionally more level 4 questions right versus level 3, maybe that's because when you spot hard questions, you focus your energy on diagraming/thinking them through to ensure you get them right? Whereas perhaps on a level 3 question you might spend less time on it? Again, in the end, I can only speculate, but my best guess would be that they make up the bulk of questions, they might not initially appear difficult, but it's that fact that might cause the LSAT test taker to spend less time on such questions.
As for the best way to try to break that type of pattern, one thing you can do is feel free to tackle the questions out of order. Some students find parallel-the-reasoning questions to be the most challenging, for example, and therefore choose to skip over them, only returning to them if they have time. I'm not sure how useful that will be in the case of level-3 questions, as I don't know the average level of parallel the reasoning questions, and it also seems to be a question type that it's probably hard to detect. Another thing you can try is to use the various drills in PowerScore's lessons, focusing on questions they've generally labelled as difficult, and also focusing on drills that are specifically difficult for you. They won't have these labeled as level-3 or -4, but it's at least a rough proxy for drilling yourself on those types of questions.
Finally, in working on those questions, keep in mind the importance of (1) diagramming out conditional reasoning, and (2) sorting answers into contenders and losers. If you have your work diagrammed, you have something solid to compare potential answer choices against--this helps reduce the risk that the test writers will trip you up with an answer choice that makes a slight change to the conditional reasoning. And sorting the answers into contenders and losers is an important tool to use for the sake of being time efficient, which seems especially important for level-3 questions, while also being able to make a reasoned choice in selecting an answer. If you can narrow down the answer choices to just two, hopefully that'll help you be able to spot a problem in one of them.
Hope that helps!
