Critical Reasoning
Verbal
7 min readInference questions ask what must be true based on the passage. The correct answer follows from the evidence without adding assumptions. Unlike conclusion questions, the answer does not have to be the author's main point. Unlike assumption questions, the answer is a fact that can be derived, not an unstated premise the argument needs.
An inference is a statement that must be true if the passage is true. It can be a modest consequence of the evidence rather than the author's main claim. If the passage says that all members of a club pay dues and that Maria is a member, an inference is that Maria pays dues. The passage does not need to state that conclusion directly.
The key word is "must." An inference is not a statement that could be true, or a statement that is likely to be true. It is a statement that cannot be false if the passage is accurate.
Inference answers stay close to the evidence. If an answer requires an extra assumption, a prediction, or a broad generalization, it is not an inference.
An assumption is unstated and necessary for the argument. An inference is unstated and supported by the evidence. The direction differs: an assumption supports the argument, while an inference follows from the argument.
For example, consider the argument "The store raised prices and sales fell. Therefore, the price increase caused the decline." A necessary assumption is that no other factor caused the sales drop. An inference is that the store's revenue may have fallen, although that depends on the size of the price increase and the sales drop. The assumption is needed for the argument. The inference is a consequence of the evidence.
The GMAT often includes choices that are plausible but not required. "Could be true" is not enough. If the passage says a company hired 20 employees, a choice saying the company hired more than 10 employees must be true. A choice saying the company hired 25 employees could be true, but it is not supported.
Watch for choices that are stronger than the evidence. If the passage says "some customers prefer the new design," a choice saying "most customers prefer the new design" is too strong. If the passage says "the drug reduced symptoms in the study," a choice saying "the drug cures the disease" goes beyond the evidence.
Quantifiers decide many inference questions. "All," "most," "some," and "none" are not interchangeable. From "some A are B," you can infer that some B are A. You cannot infer that most A are B, or that some A are not B.
From "all A are B," you can infer that every A is B, but not that every B is A. From "no A are B," you can infer that no B are A. From "most A are B," you can infer that some A are B, but not that most B are A.
Example. Passage: "All of the company's managers have attended the leadership seminar. Some of the managers have also attended the finance seminar."
Valid inference: some people who attended the leadership seminar also attended the finance seminar.
Invalid inference: all attendees of the finance seminar are managers. The passage does not say who else attended the finance seminar.
An inference cannot turn a correlation into a cause unless the passage states causation. If the passage says ice cream sales and drowning rates both rise in summer, an inference is that both are more common in summer. It is not an inference that ice cream causes drowning. A choice that claims causation goes beyond the evidence.
Similarly, a passage that says a policy was followed by a result does not prove that the policy caused the result. An inference can say the result occurred after the policy, but not that the policy produced it.
Inference questions with numbers often test the difference between absolute and relative amounts. A company can have higher revenue and lower profit at the same time. A school can have a higher graduation rate and fewer graduates if the student body shrank. Do not assume that one measure moves with another unless the passage links them.
Percentages and totals are a common trap. If the passage says a state has the highest percentage of a certain group, it does not mean the state has the largest number of that group. A small state can have a high percentage and a small total.
Comparative statements support limited inferences. If a passage says that Company A grew faster than Company B, you can infer that Company A's growth rate was higher. You cannot infer that Company A is larger, more profitable, or better managed. Growth rate and size are different measures.
If a passage says that a treatment is more effective than a placebo, you can infer that the treatment performed better in the study. You cannot infer that it is the best treatment available or that it works for every patient. The comparison is limited to the groups and conditions in the passage.
Percentage statements often support weaker inferences than they appear to. If a passage says that 60% of a group supported a policy, you can infer that the majority supported it. You cannot infer that 60% of the whole population supported it unless the group is representative. You also cannot infer the number of supporters without the size of the group.
Example. A survey finds that 70% of respondents at a conference prefer online sessions. What can be inferred?
You can infer that most respondents at the conference prefer online sessions. You cannot infer that most professionals prefer online sessions, because the conference attendees may not represent all professionals.
Some inference questions ask what the author implies about a person or group. The answer may be a negative implication, such as that a critic overlooked a factor. Stay close to the passage's wording. If the author says a scholar "fails to consider" a factor, you can infer that the scholar's argument may be incomplete. You cannot infer that the scholar is dishonest or that the argument is entirely wrong.
The correct inference is often modest. If an answer makes a strong claim that the passage does not require, it is probably wrong. Choose the answer that must be true, not the one that would be interesting if true.
| Statement in passage | Valid inference |
|---|---|
| All A are B | Every A is B |
| Some A are B | Some B are A |
| No A are B | No B are A |
| Most A are B | Some A are B |
| A happened after B | A happened after B, not necessarily because of B |
| Revenue rose | Profit may or may not have risen |
The safest inference is often the least exciting one. If an answer sounds like a dramatic conclusion, check whether the passage actually requires it. If the passage can be true while the answer is false, the answer is wrong.
Free Critical Reasoning questions on this topic — try a few now.