6.4 Interpreting Information: Data-Based Decision Making Items

6.4 Interpreting Information: Data-Based Decision Making Items

Of the six formats that recur in Decision Making, Interpreting Information is the one candidates most often meet unprepared, because it looks like a Quantitative Reasoning question and is marked like a syllogism. A small data set appears — a table, a bar chart, a pie chart, an infographic, or a short block of text with embedded figures — followed by five independent conclusions, each requiring a Yes or No response by drag-and-drop.

---

The Marking Model Determines the Strategy

That partial-credit rule is doing a lot of work. Getting three statements right earns nothing at all, while getting four earns half the item. The practical consequence: there is no such thing as a safe partial attempt. If you have confidently resolved three statements and are unsure of two, guessing both is strictly better than leaving them, because you need at least four to score anything — and there is no negative marking.

It also means the five statements deserve unequal time. Resolve the three or four that are cleanly decidable first, then spend whatever remains of your ~63 seconds on the ambiguous ones.

---

The Core Rule: The Data Is the Whole World

The standard for a Yes is not "this sounds right" or "this is true of UK hospitals". It is: does this conclusion follow from the information provided? Anything that requires a fact from outside the stimulus — however well established — is a No.

   ┌────────────────────────────────────────────────────────────────┐
   │  Answer YES only if:                                           │
   │    the conclusion is stated in the data, OR                    │
   │    it follows by arithmetic or logic from the data alone       │
   ├────────────────────────────────────────────────────────────────┤
   │  Answer NO if:                                                 │
   │    the data contradicts it, OR                                 │
   │    the data is silent on it, OR                                │
   │    it needs an outside fact, assumption or plausible guess     │
   └────────────────────────────────────────────────────────────────┘

Note the asymmetry with Verbal Reasoning: VR gives you a third option, Can't Tell. Interpreting Information does not. "The data does not tell me" collapses into No, because the question is whether the conclusion follows, and an underdetermined conclusion does not follow.

---

A Four-Step Routine

Step 1 — Read the axes, units and totals before the statements. Ten seconds spent on the header row, the units (£ thousands? per 100,000? percentages of which base?) and the row/column totals prevents most errors that follow. Note especially whether a chart is a stacked chart (parts of a whole) or a clustered chart (independent series), and whether a percentage column is a share of the row total or of the grand total.

Step 2 — Classify each statement before evaluating it. Almost every conclusion is one of four kinds:

Step 3 — Test the statement against the data, not against memory. Put a finger (or the cursor) on the exact figure you are using. If you cannot point at it, you are inferring rather than reading.

Step 4 — Reset before the next statement. The five conclusions are independent. A common and expensive failure is momentum: having decided that statement 2 was a sneaky "No", candidates become suspicious and mark 3, 4 and 5 as "No" as well. Each statement is a fresh question against the same data.

---

The Recurring Traps

1. The unsupported comparison

The data reports Ward A's readmission rate and Ward B's staffing level. The conclusion compares Ward A's staffing with Ward B's. The figures exist for different variables in each ward, so the comparison cannot be made. No.

2. Percentage versus absolute

A table shows that 40% of Clinic X's 250 patients and 25% of Clinic Y's 480 patients were referred onward.

• Clinic X: 0.40 × 250 = 100 referrals
• Clinic Y: 0.25 × 480 = 120 referrals

The conclusion "Clinic X referred more patients than Clinic Y" is No, despite Clinic X having the far higher rate. A larger share of a smaller base is frequently a smaller count. This is the single most productive trap in the format, and the calculator settles it in about eight seconds.

3. Rates with different denominators

"Infections per 1,000 bed-days" and "infections per 100 admissions" are not comparable numbers. A conclusion that ranks two units measured on different denominators is No unless the data also supplies the conversion.

4. Correlation dressed as cause

Two series rise together across five years. The conclusion says one caused the other. No. The data establishes co-movement; causation requires an assumption the data does not supply.

5. Quantifier drift

The data supports "some" and the conclusion asserts "most", "all" or "every". Check the quantifier word explicitly — it is the most common single-word difference between a Yes and a No.

6. Silent projection

The data covers 2022 to 2025 and the conclusion describes 2026. Extrapolation is an assumption. No, unless the stimulus explicitly states a trend will continue.

---

Worked Example

Stimulus. A hospital trust reports emergency attendances for four sites in one quarter:

Site	Attendances	Admitted (%)	Mean wait (minutes)
Northgate	12,400	22%	168
Eastcliff	8,900	31%	141
Southfield	15,600	18%	195
Westbrook	6,200	27%	122

Four of those five needed the calculator or a quantifier check rather than a glance at the table. That ratio is typical.