Diagram Questions: A Deeper Method

Build the discriminator grid

The single highest-value move in Part 2 is to find, for each of the four options, what it has that the other three do not. Two or three items each, so eight to twelve in total, and they can be held in your head.

Do this by reading across rather than down. Take one dimension, say price, and look at all four values together. Then the next dimension. Reading across finds differences in a way that reading each column top to bottom does not, because a difference only exists between two things and reading one column at a time hides it.

What falls out is a small map of the kind: A is cheapest but expires in 30 days. B is the only transferable one. C is the only one with evening access. D is the most expensive and the only annual one. Four sentences, and every question in the part is built on one of them.

The corollary matters as much as the grid. Any feature shared by three or four of the columns cannot be the reason one option wins or loses, so it can be discarded for the rest of the part. Most of the chart is that kind of feature, which is why it looks dense and is not.

Mine the email for constraints, not for meaning

The email exists to tell you what this person cannot have. A budget ceiling. A date they need it by. A vehicle that will not fit. A brother who wants to borrow it. Six weeks rather than six months.

Read it straight through with the blanks empty, and do not stop to answer anything. What you are collecting is a list of three or four constraints, and every blank in the part is a collision between one constraint and one entry in your discriminator grid.

Note especially the negative constraints, which are the ones candidates skip. She will not be here in August. He does not drive. A negative constraint eliminates rather than selects, and elimination questions are common in this part because they are easy to build and hard to rush.

The email also usually contains a recommendation, stated or implied. Hold on to which option it lands on, because questions 6 to 8 are about the reasoning behind that recommendation rather than about the chart.

The five shapes a blank can take

Uniqueness. It is the only one that... The blank must be filled by a discriminator, and anything shared by two or more columns is dead on sight. Verify by checking the other three columns for the same feature.

Elimination. That rules out... A constraint kills one or more options. Verify by naming the constraint and the specific line it collides with. If you cannot name both, you have guessed.

Comparison. Twelve dollars less than... Two specific values, one arithmetic operation. These are the fastest questions in the part and the most often lost to carelessness, because the chart usually contains the number that would be right if you had compared the other pair.

Threshold. Since she needs it for at least six weeks... One value against a limit. Watch the direction of the inequality and watch whether the boundary is inclusive, because distractors here are built from the option that sits exactly on the line.

Combination. The only one that is both... Two conditions at once, and the hardest shape in the part. Three of the four options will satisfy one of the conditions, which is the whole design. Check both conditions against the column separately before you commit.

Name the column, then look

Before opening any drop-down, say which of the four options the sentence is about. This single habit disarms most of the wrong answers in Part 2, and it is worth understanding why it works so well.

The standard Part 2 distractor is a genuine feature copied from a different column of the same chart. It is true, it is printed, and you did read it. What makes it wrong is only the pairing. So a candidate evaluating options by asking whether each one is true faces four plausible statements, while a candidate who has named the column faces one column to check and three irrelevant claims.

The email's own grammar often names the column for you before you look. Both of those refers back to two options already mentioned. Neither of the cheaper two tells you which pair is in play. The one I would actually get points at the recommendation. Read each sentence past the blank to its full stop, because these signals sit after the drop-down at least as often as before it.

When the sentence does not name a column, the constraint usually does. A blank in a sentence about lending the pass is about transferability, which your grid says belongs to exactly one option.

Questions 6 to 8 are about the argument

The last three questions in Part 2 are sentence stems about the email's reasoning rather than blanks in it, and they are a different task. They ask why the writer ruled something out, what they assumed, or what would change their recommendation.

The evidence for these is the argument itself, so the chart is secondary. A stem asking why the writer rejected option D is answered by the constraint in the email that collides with D, not by whatever is most notable about D in the chart. Candidates who go back to the chart for these questions spend thirty seconds retrieving a true fact that is not the reason.

The commonest distractor here is a real disadvantage of the rejected option that the email never mentioned. Option D is indeed the most expensive, and if the email rejected it because it expires too early, the price is a true fact and the wrong answer.

Stems about what would change the recommendation are the one place in Part 2 where you reason forward rather than retrieve. Take the constraint the recommendation rested on, remove it, and see which option the grid then favours. Usually one, and usually the option that was second on the original constraint.

Verifying in the seconds you have

The whole verification is two questions, and together they take about four seconds.

First: is this feature actually in the column I named? Not in the chart, in the column. This catches the copied-feature distractor, which is most of them.

Second: does this answer the sentence, including the part after the blank? This catches the shape errors, particularly combination blanks where only one of the two conditions was checked.

If you are running short, verification is the last thing to cut and the first thing candidates do cut. A Part 2 answered quickly without checking the column is a part answered at roughly the rate of a coin toss on the harder half, and the four seconds are cheaper than the question.

Practise it

Five questions on the diagram procedure: order of reading, discriminators, and the shapes a blank can take.