Build an audit that identifies the problem, uses evidence and explains a better representation.
Critique and improve a graph, identify unsupported conclusions and justify corrections using data and context.
Begin with a specific judgement about the representation. Quote or describe the relevant feature, cite the data and explain the effect. ‘The graph is bad’ is a reaction. ‘The cropped bar axis makes a change from 96 to 100 look much larger than its approximately 4.17% increase’ connects a design choice to a mathematical consequence.
Check whether the plotted values match the data, whether the representation suits the question, and whether the headline follows from the evidence. These are related but separate checks. A correctly plotted graph may accompany an unjustified causal claim. An attractive graph may omit a group or time period needed to test its conclusion.
Choose changes that address the identified problem: a clearer graph type, a shared scale, accurate labels, consistent units or missing context. Keep the recorded values intact. If the source itself is ambiguous or inconsistent, flag the issue and request clarification; do not silently invent a corrected dataset. Keep your reconstruction separate from the original evidence.
A correction should make the intended comparison easier and fairer. State what changed, what stayed the same and what conclusion is now supported. A good audit also describes limitations. If the sample is small or unrepresentative, a better graph cannot remove that limitation from the underlying evidence.
Repair this display: replace its overstated title, restore a zero baseline and explain what the values support.
Line graphs use equally spaced positions here: use observations at equal time intervals. Do not use a line to connect unrelated categories. Circle graphs require a positive total.
A neutral title such as “Recorded sales before and after” and a zero-based bar graph show the increase without exaggerating visible lengths. Sales rose by 4 units, approximately 4.17%. The two observations do not establish the cause.
Check the numerical representation separately from the proposed cause.
The graph may faithfully show a rise while the causal claim remains unsupported. Explain what additional comparison or investigation would be needed to evaluate the logo’s effect.
The task is to audit the evidence, not erase the inconsistency.
No. Preserve both originals, describe the mismatch precisely, cite the relevant values or labels and explain what clarification is needed. Build a separate corrected representation only when the source data is established.
Feedback explains your chosen response. These activities are practice, not a school mark.
Apply this audit structure to Case Files: Misrepresentation. Work from the original evidence, draft your correction separately and use data to justify every proposed change.
Give a partner your corrected graph without your explanation. Ask them to infer the intended conclusion. Revise your labels or reasoning if the graph still invites an unsupported interpretation.