The format: a figure plus a claim
Quantitative Command of Evidence lives in the Information and Ideas domain, and like every Reading and Writing item it is one question per mini-passage. You get a short setup (a researcher measured something), a graph or a table, and a sentence with a blank or a claim to complete or support. Your task: pick the choice that uses the data correctly to do that job. There is no penalty for guessing, and only a small share of the section is quantitative, so this is high-yield: a handful of points that reduce to one repeatable skill, not to being fast at numbers.
One thing to accept first: the graph is not the question. The claim is the question, and the graph is just where the answer has to be provable.
Claim first, data second
The instinct is to study the figure the moment it appears, tracing bars and comparing heights before you even know what you are looking for. Reverse it. Read the claim first, then read the data with that claim in hand.
Say the claim back to yourself in plain words: what quantity, for which entity, over what span? "Reading time rose the most, from 2019 to 2022, in one of these programs." Now you know exactly what to hunt for (a rise, largest, in that window) before a single answer choice gets to suggest anything to you. This is Predict-Then-Peek pointed at a graph: commit to what the data must show before you look, so the choices cannot anchor you to a number just because it is printed on the page.
The data-misread trap
Here is the trap that quietly eats these questions. The wrong choice reports a value the figure does not show for the point in question. It is a misread axis, the wrong row or column, the wrong time period, or two numbers transposed. Your reasoning about the claim was fine. Your extraction of the number was wrong.
This distractor survives a glance for one reason: the number is real. It is on the chart. It came from that same figure, just from a different cell, an adjacent interval, or a mis-scaled reading of the axis. So it looks harvested from the right place, and your eye accepts it. This is not a knowledge gap. It is a behavioral leak: an extraction slip, the same species of error as reading a clock hand wrong. The cure is not more studying. It is a mechanical check on the number itself, which is the Receipt Rule applied to data: you may pick a choice only if you can point to the exact cell or bar that proves it.
Units, axes, and scope
Two more ways the point walks out the door, both about reading the frame instead of the datum.
Units and axes. Miss the units and every comparison you make is off. Thousands versus millions, a percentage versus a raw count, a scale that starts at 40 instead of 0 so a tiny gap looks enormous. Read the axis label and the unit before you read any value, out loud in your head if you have to.
Scope. Some choices report a completely correct number, but for the wrong entity or the wrong claim. The claim asked about the rise; the choice reports the highest total. Both are true readings of the figure, but only one answers the question. This is the true-but-irrelevant trap in numeric clothing: right chart, right reading, wrong target. Others sprawl the other way, generalizing past what one figure can show. Match the exact entity, quantity, and span the claim names. Nothing wider, nothing narrower.
Worked original example
A researcher recorded the average hours per week students spent reading for pleasure in four after-school programs, in 2019 and 2022.
| Program | 2019 (hrs/wk) | 2022 (hrs/wk) |
|---|---|---|
| A | 3.1 | 4.4 |
| B | 5.2 | 4.9 |
| C | 2.8 | 3.0 |
| D | 6.0 | 5.1 |
The claim to complete: "Reading time rose the most from 2019 to 2022 in ___."
Claim first. You need a rise (2022 higher than 2019), and the largest one. Now the traps line up:
- "Program D, which reached the highest level at 6.0 hours." A perfectly correct number, and a losing one. 6.0 is the biggest value on the table, so it tempts the eye, but it is the 2019 figure, and D actually fell to 5.1. This is scope: the claim asked about the rise, not the level. Right number, wrong claim.
- "Program B, which rose from 4.9 to 5.2 hours." This is the data-misread trap in the open. B did not rise. It fell, from 5.2 to 4.9. The choice transposed the two years to turn a decline into an increase. The numbers are both real and both from row B, which is exactly why it reads as legitimate.
- "Program A, which rose from 3.1 to 4.4 hours." The answer. A gained 1.3, the largest increase on the table. C also rose, but only 0.2, right direction and far too small to be "the most." Trace it to the cells and it holds.
Notice you never calculated anything harder than a subtraction you could do in your head. The whole question turned on reading the correct two cells for the correct entity.
The method: the Label-First read
One named routine covers every one of these:
- Read the claim, not the chart. Restate in your own words the exact quantity and exact entity it asks for. Do this before the figure gets a vote.
- Label the figure. Ten seconds on the title, both axes or all column headers, the units, and the legend. Say the units in your head.
- Locate the exact data point. Find the cell or bar the claim names and read its value. Not the neighbor, not the biggest number, not the one that happens to match a choice.
- Verify each choice like a receipt. For every option, point to the specific data point that proves it. A choice you cannot trace to one exact cell or bar is out, no matter how plausible the sentence sounds.
Run this self-test today. Pull your last practice module and find every chart or table question. For each miss, ask one question: did my reasoning fail, or did my extraction fail? If you understood the claim but read the wrong row, wrong axis, or wrong year, that is the data-misread leak, and it will not respond to more content review. It responds only to the label step. If instead you read the number right but matched it to the wrong claim, that is scope, and the fix is restating the claim first. Sorting your own misses into those two buckets tells you which half of the method to drill.
Practice it now: the free simplesat command-of-evidence drill runs these graph and table items with per-trap feedback, so you see whether a miss was a misread or a scope slip. No signup, works as a guest.
For the non-quantitative side of this question type (the paragraphs that ask which quote or finding supports a claim), the companion Command of Evidence guide uses the same claim-first move on text instead of figures.