LSAT 90 - Section 2 - Question 18

This is a Flaw/Descriptive Weakening question.

This is a causal argument with phenomenon-premises and a hypothesis-conclusion. The first premise reveals that for physical therapy patients that received less than six weeks of treatment, around a third showed major improvement. This result was the same regardless of whether the treatment was by a generalist or specialist. The second premise reveals a similar phenomenon for longer treatment. This time, around half showed major improvement. Again, the result was the same regardless of whether the treatment was by a generalist or specialist.

So those are all the facts we have. That’s collectively the “phenomenon.” What causal conclusion can be drawn? Well, not much. Certainly not the conclusion actually drawn: the choice between a generalist and specialist makes no difference.

Let’s say you wanted to test this hypothesis. You think that the choice between a generalist and specialist makes no difference. How would you go about designing your (ideal) experiment? I put “ideal” in parentheses because it really doesn’t have to be ideal. It just has to be better than the data in the stimulus and that’s a low bar. So, you’d collect a ton of people prescribed less than six weeks of physical therapy. Then you’d randomly assign them to specialists and generalists. Then you wait out the six weeks and you measure to see what percentage in each group experienced major improvement. If it turns out that about a third experienced major improvement in either group, then that’s good evidence that the choice between generalist and specialist has no causal power. (And then you do it all over again for the long-term treatment.) But notice how different that is from what actually happened in the stimulus. The stimulus presented observational data, not experimental data. That means there were no controls. Specifically, there were no controls for self-selection. The assignment to generalist or specialist was not random. People chose whether to see a generalist or a specialist and presumably with good reason! What reason? Maybe because generalists and specialists are better suited to treat different kinds of injuries.

This is exactly what Correct Answer Choice (E) points out. This argument is vulnerable because it overlooked the possibility that generalists and specialists each excel at treating a different type of injury. (E) is subtle in exposing the flawed logic of the argument. I’m pretty blunt so let me shine a spotlight on this error with the following argument.

Records reveal that of patients who received heart surgery, about 75% regained full cardiovascular functions one month post operation regardless of whether they received a double bypass or a quadruple bypass surgery. Therefore the choice of a double or quadruple bypass will not affect one’s chances of regaining full cardiovascular function.

Is the flaw glaring now? Imagine telling the patient who’s scheduled for a quadruple bypass that actually, you know what, you don’t need four unclogged arteries, let’s just clear up two of them and you’ll be fine.

Answer Choice (D) is very attractive. It claims that the reasoning is vulnerable because the argument failed to indicate whether the number of patients surveyed who saw a generalist was equivalent to the number who saw a specialist. While (D) is descriptively accurate, the argument’s failure to indicate isn’t where the argument is weak. To see this, imagine if we supplied an additional premise that fixed this “problem” by indicating what (D) wants. So the same argument, but now an additional premise indicates that the numbers were equal. Voila, the conclusion follows! Just kidding. The argument is still garbage for all the reasons discussed above.

If (D) was attractive to you, perhaps you thought that experiments require equal group sizes? First, note that (D) isn’t talking about an experiment (because the stimulus isn’t an experiment). Second, it’s not true that equal sizes are a requirement for experiments. It’s nice to have (for statistical reasons) but it’s not necessary. I think you can intuitively understand this just by imagining an experiment where one group was slightly larger than the other, say 60/40. Your results will still be fine as long as you avoid the other experimental pitfalls (random assignment, blinding, etc.).

Answer Choice (A) is also attractive though for a different reason. (A) is attractive mainly because it’s gibberish but fancy-sounding gibberish. It claims that the argument is vulnerable because it assumes (without warrant) that if the effectiveness of different practitioners in bringing about any (which includes minor) improvement does not differ, then their effectiveness in bringing about major improvement cannot differ. What? No, the argument makes no such assumption. The argument does not assume there is no difference in how effective generalists and specialists are with respect to bringing about minor improvements.

Answer Choice (B) claims that the argument provides no information about the kinds of injuries that require short-term versus long-term treatment. Okay, this is descriptively accurate but nobody cares because this isn’t why the argument is weak. I think (B) could have been right had the argument been edited just a little. Like this: Records reveal that regardless of whether patients receive short-term (six weeks) or long-term treatment, about 50% experience major improvements within six weeks. Therefore, the choice between short- or long-term will not affect one’s chances of major improvement. This argument is weak because it assumed that the assignment was random whereas it almost certainly was not. It almost certainly was the case that there was self-selection determining who got short-term and who got long-term. And that’s because doctors diagnose and prescribe proportional treatment. One way to describe this weakness is to say that the argument should have provided information about the kinds of injuries that require short-term as opposed to long-term treatment. That information would have most likely revealed that it’s the more severe injuries that require long-term treatment.

Answer Choice (C) claims that the argument overlooks the possibility that patients are more strongly biased to report favorably on one of the two types of medical professionals than on the other. (C) thinks the weakness in the argument has to do with whether the “one third” and “one half” major improvement results are believable. This kind of objection, the “wait, but are you sure your data maps onto reality?” is more powerful in arguments that rely on survey results. It’s not clear this is such an argument. The premises say, “records reveal.” What kind of records? How was the data collected? Did we ask the patients if they showed major improvement? Or did we simply measure their physical abilities to determine major improvement? If the latter, then (C) completely misses the mark for it assumes the data came from subjective reports. If we patch this issue up, then (C) at best identifies a minor issue in the argument. While it’s true that the argument does assume that these records reflect reality, the argument also failed to account for self-selection. That second issue is the major one because if that were fixed, the argument would improve dramatically whereas if we fixed the “records reflect reality” issue, the argument would still be flawed.