The foundation under every DI question
Before the five formats, you need fluency with the raw materials: kinds of data, the displays that carry them, and the patterns worth spotting. None of this is advanced statistics — it's careful reading.
Kinds of data
- Quantitative values measure amount (revenue, weight, time) and support arithmetic. Categorical values label a group (region, product type) and don't.
- A quantitative variable can be discrete (counts — number of employees) or continuous (measured on a scale — temperature). The distinction decides what's sensible: "the average household has 2.3 children" is fine as a statistic, absurd as a literal count.
- Watch units and scale: thousands vs millions, percent vs percentage points, per-unit vs total. Most data misreads happen here, not in the math.
Reading the common displays
Read a chart in this order: title → axis labels → units/scale → legend → only then the data points. The "furniture" tells you what the numbers mean; skipping to the bars is how careless errors start.
| Display | Best for | The trap to watch |
|---|---|---|
| Bar chart | Comparing categories | A vertical axis that doesn't start at 0 exaggerates differences |
| Line graph | Trends over time | Uneven spacing on the time axis distorts the slope |
| Pie chart | Parts of one whole | Percentages are of that whole only — not comparable across two pies of different size |
| Scatter plot | Relationship between two variables | Correlation is not causation; an outlier can dominate the impression |
| Table | Exact values, multiple variables | The answer often appears only after you sort |
Patterns the GMAT rewards you for seeing
- Center and spread. Mean vs median (skew pulls the mean toward the tail; the median resists it), and how tightly values cluster. A single big outlier moves the mean and the range far more than the median.
- Correlation. Two variables can move together (positive), oppositely (negative), or not at all. Strength is about how tightly the points hug a line — and a strong correlation still says nothing about cause.
- Rates of change. On a line graph, the steepness between two points is the rate; "growth is slowing" means the line is still rising but flattening, not falling.
- Proportions across groups. A percentage always has a base — "40% of the engineers" and "40% of all staff" are different counts whenever the groups differ in size.
Percent vs percentage points. If a market share rises from 20% to 25%, that's a 5 percentage-point gain but a 25% relative increase (). DI answer choices routinely offer both — read which one the question asks for.
A figure need not be to scale. Bars and slices that look equal may not be; trust the printed numbers and labels, not the picture. (This matters even more in Data Sufficiency, where assuming a figure is to scale invents information you weren't given.)
Make the table do the work. When a question asks "which row meets a condition," sorting by the column named in that condition usually turns a scan into a single glance. The skill being tested is locating the relevant data fast, not memorizing it.
Checklist
- Identify whether each variable is quantitative or categorical
- Read title, axes, units, legend before any value
- Note whether an axis starts at zero
- Separate percent from percentage points, and each percent's base
- Sort tables by the column the question cares about
Sample Questions
Continue Your Prep