Flawed IR Two-Part Analysis Questions

I have worked through thousands of official GMAC math and verbal questions, from GMATPrep and from the Official Guides going all the way back to the Third Edition from 1986. In all these questions, I have not found a single incorrect answer. Although there has been a rare verbal question with a debatable answer, I have never felt that any answer was truly wrong. This is why it troubles me that GMAT Genius has found two different half-baked Two-Part Analysis Integrated Reasoning questions, both in the online Integrated Reasoning component of the Official Guide (13th Edition). After multiple reviews, I remain convinced that both of these questions are flawed.

I am surprised that these two Two-Part Analysis IR questions do not live up to the standards of rigor and excellence that the GMAC is known for. Because the questions are part of the Official Guide, which was released before the Integrated Reasoning section launched, these two questions never appeared as real GMAT questions. As a result, these questions may not have been thoroughly tested with actual test takers. But that does not excuse the GMAC for allowing sloppy questions into its study guide. We can only hope that live Integrated Reasoning questions on the real GMAT will have received a much higher level of rigor and scrutiny, so that student IR scores are not unjustly impacted due to similarly negligent questions.

You may wonder why GMAT Genius has come across this only now. Honestly, we haven’t paid too much attention so far to Integrated Reasoning. Since this is a relatively new section, we believe that business schools won’t place much weight on IR in the near term. As we approach Integrated Reasoning’s one year birthday on June 5, however, we felt that IR deserved a closer look. We therefore decided to carefully evaluate all 105 official IR questions that the GMAC has made available for practice. It is worth noting that the other 103 IR questions that we evaluated are perfectly fine. Although a flaw rate of less than 2% would actually be good for some companies, this is a total shocker by the GMAC’s standards. My next post will detail some observations that we made from our analysis of the 105 IR questions.

Be forewarned that I will now be discussing in extensive detail the two flawed Official Guide IR questions. The details that follow may not make sense unless you have read these questions. You may wish to try these two questions on your own first, using your OG access to IR questions, before viewing the screenshots and discussion. Clicking the thumbnail of a screenshot below will bring up a full-size image of the question along with the official explanation.

IR Question #37

This question pertains to various causes for the decline of coral reefs. One definite factor is a type of sea star that eats coral. Human fishing has reduced the sea star’s predators. This clearly implies that there are now more sea stars, and therefore more coral consumption by sea stars. A second possible cause is that a runoff has resulted in greater phytoplankton blooms, and we are told that the sea star happily eats phytoplankton. Our task is to pick a cause-and-effect sequence that probably results in coral decline.

The correct answer to this IR question is that an increase in phytoplankton (cause) resulted in an increase of sea stars (effect). How would a non-biologist know that eating phytoplankton causes an increase of sea stars? Just because an organism gladly eats something doesn’t necessarily mean that the organism reproduces more rapidly. This is akin to saying that when people eat pizza, the world population increases. This analogy shows the absurdity in the purportedly correct cause-and-effect sequence.

GMAT questions expect us to make reasonable inferences from the information provided in the stimulus. But we are not expected to possess any specialized outside knowledge (such as the definition of phytoplankton) in order to correctly answer a question. So what could be a reasonable inference here? In my interpretation, I guessed that phytoplankton blooms on or near coral, and since the sea star enjoys eating phytoplankton, coral reef decline inevitably occurs.

Part of the problem lies in the wording of the stimulus. The first sentence tells us that various causes are at hand. So this is not a debate between two distinct causes (i.e. one OR the other). Furthermore, we already know that a lack of predators has led to an increase in sea stars, so it is not necessary to justify an increase in sea stars. Admittedly this IR question does not contain any other satisfactory answer other than the credited response. But the correct answer would be more understandable if the question stated that only one of the two mentioned causes can be correct. In that case, it would be necessary to explain an increase in sea stars. As worded, however, the leap-of-faith required to pick the correct answer goes significantly beyond a reasonable inference.

IR Question #50

This question is about a mattress company with one store in City X and one store in City Z. Pretty much everything about the two stores and cities is the same, except that the mattress store in City Z conducts twice as much radio advertising as the City X store. A consultant claims that the radio advertising has not benefitted sales, and we must decide upon what changes we could make in Cities X and Z to test this hypothesis.

Unfortunately the correct answer, eliminate radio advertising in both cities, fails to test the hypothesis. Why? As we see in several GMAT Critical Reasoning questions, we need a control group to accurately carry out a test. We should have two groups that are identical except for in the one variable that we are testing. Without a control group, we cannot be confident about the true cause of an effect that we observe. In this specific question, let’s suppose that we eliminate radio advertising in both cities and sales in both cities subsequently decline. How do we know whether the sales decline is due to the lack of radio advertising or due to another factor, such as seasonality or an economic slump? Without a control group, we cannot isolate the true cause.

On the other hand, if we eliminate radio advertising in only one city but keep it in the other city, then we could compare the impact to sales. If sales declines in both cities, then clearly another factor (such as an economic slump) came into play. If sales declines in only the city with no radio advertising, then clearly radio advertising does have an impact on sales. Finally, if sales do not decline in either city, then the consultant’s hypothesis is correct. Only with a control group can we truly evaluate cause-and-effect. The GMAC’s answer to this IR question is undoubtedly wrong.

Do you agree or disagree? Feel free to let us know by submitting a comment.