Flaws and evaluation: naming what is wrong and what would settle it

Why flaws matter even when the question does not say "flaw"

Bocconi's description of verbal critical reasoning gives, as an example, a question asking which condition would best support the thesis. Whatever form a question takes, the quickest route is the same: see what is wrong with the reasoning. A flaw is the gap between premises and conclusion described as a mistake. Once you can name it, you can spot the assumption that would cover it, the fact that would weaken the argument by exploiting it, and the information that would settle it.

The flaws that come up again and again

Flaw	What the argument does	Abstract wording you may see
Correlation as cause	Treats two things that happen together, or one after another, as cause and effect	"overlooks the possibility that some other factor accounts for both"
Unrepresentative sample	Draws a conclusion about a group from a part that may not be typical	"relies on a sample that may not be representative"
Necessary as sufficient	Treats a condition that must be met as one that guarantees the result	"takes a requirement for an outcome to be enough to bring it about"
Sufficient as necessary	Treats one way of reaching a result as the only way	"assumes that because one condition produces a result, the result cannot occur without it"
Share as rate	Uses a percentage of cases without asking how large each group is	"ignores the relative sizes of the groups being compared"
Shifting term	Uses a word in two senses, or slides from one measure to another	"uses a key term in two different senses"
Part and whole	Assumes what holds for each part holds for the whole, or the reverse	"assumes that a property of each member is a property of the group"
False choice	Presents two options as the only ones	"overlooks alternatives other than the two considered"
Attacking the source	Rejects a claim because of who makes it	"responds to the person rather than to the argument"

Conditions: necessary and sufficient

"Only students with an average above 27 are eligible" sets a necessary condition: no student without it qualifies. It does not say every student with it does. Reading it as "above 27, so eligible" is the flaw of treating a necessary condition as sufficient. The reverse slip: "Anyone who scores 100 on the final passes the course; Ada did not score 100, so she failed." Scoring 100 is sufficient for passing, but there may be other ways to pass.

A quick check: rewrite the rule as "if A then B". From A you may infer B; from not B you may infer not A. From B you may not infer A, and from not A you may not infer B.

Shares and rates

"Sixty per cent of burglaries happened in homes without an alarm" is a share of burglaries. Whether unalarmed homes are more at risk depends on the rate: burglaries per home in each group. If 80% of homes have no alarm, unalarmed homes, with 60% of the burglaries, are actually burgled less often per home than alarmed ones. Whenever an argument compares groups using percentages of an outcome, ask how big each group is.

Matching a flaw to an option

Flaw options are written in general terms, so translate the specific argument into its pattern before reading them. Then check every word of the option you like: it must describe what this argument actually does. Many wrong options name a real flaw that this argument does not commit, or describe a step the argument never takes. An option can also describe the argument accurately without naming anything wrong with it.

Evaluation: what would we need to know?

Some questions ask which information would be most useful in evaluating an argument. The answer is a question whose answer bears on the gap. Use the two-answer test: answer the question both ways, yes and no, or high and low.

• If one answer strengthens the argument and the other weakens it, the question is useful.
• If either answer leaves the argument where it was, the question is not.

Evaluation is therefore strengthen and weaken folded together. The best question is usually the one that tests the most obvious rival explanation, or the representativeness of the sample, or whether the plan's key step works.

A method for both

1. Find the conclusion and the evidence, and name the gap.
2. Fit the gap to a pattern from the table.
3. For a flaw, find the option that describes that pattern in this argument's terms. For evaluation, find the option whose two answers pull the conclusion in opposite directions.
4. Check that the chosen option is about this argument's actual conclusion, not a nearby claim.