8.3 Integrating Tables, Graphs and Text

Extract claims, then check the visuals

Many QL questions do not live in a single box. The National Benchmark Tests Project lists, among the abilities QL assesses, reading and interpreting tables, graphs, charts and text; integrating information from different sources; identifying trends and patterns; and reasoning logically. In practice that means a short paragraph, a table, and a chart that only make sense together. Independent OpenExamPrep practice for this skill uses a repeatable method: extract claims from the prose, check those claims against the numbers in the visual, then resolve any conflict before you choose an option.

Work the method in order.

Step 1 — Extract claims. For each number in the paragraph, write four labels: the figure, its unit, the group it describes, and the time window. Ignore adjectives until the numbers are pinned. Words such as "strong," "collapsed," and "every week" are claims too; they still need a numeric test.

Step 2 — Check the visual. Find the cell, bar, or point that should match each claim. Recalculate rates yourself. Confirm that the chart's title, axis units, and footnotes use the same group and the same year as the sentence.

Step 3 — Name conflicts. Typical mismatches are count versus rate, enrolled versus sat-the-exam, this year versus last year, rounded prose versus exact table, and a truncated vertical axis that makes a small change look dramatic.

Step 4 — Answer from the reconciled picture. If the question asks what the dean claimed, report the claim. If it asks what the table shows, trust the table. If it asks whether the paragraph is supported, you must compare.

Here is a worked campus example from fictional Kei River University of Technology (KRUT) in the Eastern Cape.

Paragraph (faculty newsletter): "The Faculty of Engineering reports a strong start to 2026. Average tutorial attendance rose from 62% in 2025 to 70% this year, and fewer than one in ten first-years failed the statics module."

Table: first-year statics outcomes

Chart: average tutorial headcount (students physically present per session) — 149 in 2025 and 126 in 2026.

Extract. Claim A: attendance rate 62% → 70%. Claim B: "fewer than one in ten" failed in 2026. Claim C (tone): "strong start," which is not a number. Claim D, hiding in the chart rather than the prose: headcount 149 → 126.

Check claim A. 149/240 = 0.6208 ≈ 62%. 126/180 = 0.70 = 70%. The percentages in the newsletter match the table's enrolment as denominator. The chart, however, shows fewer bodies in the room. That is not a printing error. The 2026 class is smaller (180 enrolled instead of 240), so a higher share can sit in a smaller headcount. Counts and rates moved in opposite directions. A candidate who reads only the bars will conclude that attendance fell. A candidate who reads only the adjectives will miss that the room is quieter.

Check claim B. Failed among those who sat in 2026: 16/176 ≈ 9.1%, which is fewer than one in ten. Failed among enrolled: 16/180 ≈ 8.9%, also under 10%. Failed among enrolled in 2025: 41/240 ≈ 17.1%; among those who sat: 41/228 ≈ 18.0%. The newsletter is defensible for 2026. It is not a statement about 2025, and it is not a statement that 16 is a small number in every faculty — 16 failures in a class of 50 would be a disaster. The phrase "first-years failed" still leaves a small ambiguity: enrolled or sat? Here both readings stay under 10%, so the conflict is mild. On other items the two bases disagree, and the question will hinge on that difference.

Check the means and medians. The mean rose from 52 to 58; the median from 54 to 57. Both shifted up, and they stayed close, so a single high-scoring outlier is unlikely to be the whole 2026 story. If the mean had jumped to 68 while the median stayed 57, you would look for a handful of very high marks. Integrating sources includes noticing when two summaries of the same list tell different tales.

Trends need several points, not a two-year slogan. KRUT's after-hours library gate counted late-night entries in the six weeks after orientation: 42, 55, 61, 58, 70, 73. A paragraph that says "entries increased every week after orientation" is false: week 4 dipped from 61 to 58. A paragraph that says "entries rose overall across the six weeks" is true: the last point is well above the first, and the pattern is upward aside from one dip. "Every," "steady," "constant," and "unchanged" are logical words. One counterexample week is enough to break them.

Reasoning across a paragraph plus a table plus a chart often means refusing a false compromise. Suppose the newsletter says failure "halved" because 41 fell to 16 (16 is not half of 41). The count fell by 25, so 16/41 ≈ 0.39 of the previous failure count, while the failure rate among those who sat went from about 18% to about 9% — that rate did roughly halve. If you mix the count story with the rate story you can "prove" anything. Keep the measure fixed.

A last family of conflicts is visual rhetoric. A vertical axis that starts at 140 instead of 0 makes the drop from 149 to 126 look like a collapse. The table's 62% to 70% would look almost flat on a 0–100% axis and dramatic on a 60–72% axis. QL will not always warn you. Compare the axis minimum with zero, and compare any drawn percentage with a recalculated fraction.

Exam routine: underline every number in the paragraph and draw an arrow to its cell or bar. If an arrow has nowhere to land, the claim is unsupported. If two arrows land on different quantities (headcount versus rate), you have found the trap. Then answer the question that was asked, not the louder source. KRUT's dean can be right about rates and still misleading about how full the tutorial room feels. Your job is to say which statement the combined evidence supports.