Read the question before exploring tabs
Identify the decision and required data; open only the sources needed to evaluate that decision.
← GMAT Course
DATA INSIGHTS · GMAT-DI-MSR-01 · DRAFT
Multi-Source Reasoning: Evidence Navigation
Compare records, qualifications and numerical constraints across independent sources
Learn: Evidence and decisions
Separate facts, forecasts and assumptions
A target, an estimate and an actual outcome are not interchangeable. Track the source and time period.
Reconcile conflicting evidence
Compare units, dates, definitions and scope before treating two statements as contradictory.
Build an evidence chain
Start with the claim. Identify the source defining its metric, then locate the relevant counts and apply any exceptions. A conclusion may need all three tabs even when only one contains numbers.
Use the correct combined denominator
Add favorable counts and total counts separately before calculating an overall rate. Averaging percentages gives each group equal weight, which is inappropriate when their sizes differ.
Test support, not plausibility
For a statement asking whether a conclusion follows, select No when the evidence leaves it unresolved. This differs from claiming that the conclusion has been disproved. Look for a missing comparison, an unobserved group or an unsupported causal link.
Resolve scope before arithmetic
Identify who or what belongs in the reported population. An exclusion may change the denominator without changing the favorable count. Two sources can report different percentages without disagreeing about the underlying events.
Use bounds when evidence is incomplete
Treat unknown outcomes as unknown. Work out the smallest and largest possible totals, then compare the target with that range. A result can be possible without being guaranteed.
Avoid double counting across sources
When two sources describe overlapping groups, identify the intersection before combining counts. A person included in two categories is still one person when eligibility is defined per person.
Build a short source map
Give each source a role: definitions, recorded results, proposal, or forecast. For each claim, find the needed facts and link them using the stated rule. A proposal is not evidence that an event happened; missing information is not proof of the opposite.
Learning Objectives
- Navigate evidence across source tabs.
- Combine numerical records with definitions and target statements.
- Distinguish actual outcomes from plans and projections.
Prerequisite Knowledge
- Percentages and percentage change.
- Profit as revenue minus costs.
Worked Examples
Reconcile reported profit with a target
Finance reports Q2 revenue of $500k and operating expense of $380k. A separate board note sets a profit target of $110k.
- Calculate actual profit from the financial source.
- Compare the actual with the target from the second source.
Conclusion: Actual Q2 profit is $120k, so it exceeds the $110k target by $10k. A target is not a substitute for actual results.
Compare rates with different denominators
A weekly report shows 84 late deliveries out of 1,200 orders. A later memo says 96 late deliveries out of 1,600 orders. Did the late-delivery rate rise?
- Calculate the rate in each source using that source’s own order total.
- Compare rates rather than raw counts.
Conclusion: The number of late deliveries increased, but the rate fell by 1 percentage point. Mixing the raw counts with the rates would lead to the wrong conclusion.
Worked example: combine unequal groups
One source reports 72 satisfied respondents out of 80 in Branch A. Another reports 42 out of 60 in Branch B. A methods note defines the combined rate using all completed responses.
- Identify the denominator in the methods note: completed responses, not branches.
- Add the favorable outcomes before dividing.
- Check the tempting alternative: an unweighted average treats unequal groups as equal.
Conclusion: Combine counts when group sizes differ. The simple average of group percentages need not equal the overall rate.
A disagreement caused by different populations
A register lists 125 bookings, including five canceled before an event. An attendance log records 108 attendees. The reporting rule excludes pre-event cancellations and sets a target of at least 90%. A memo divides attendance by all bookings and says the target was missed.
| Register | 125 bookings; 5 pre-event cancellations |
| Attendance log | 108 attendees |
| Reporting rule | Exclude cancellations; target at least 90% |
- The reporting rule determines the denominator; the memo does not override it.
- Apply the definition and compare with the inclusive threshold.
- Explain the disagreement rather than choosing a source by intuition.
Conclusion: The event meets the defined target. Both divisions are arithmetically possible, but only one uses the required population.
Unknown outcomes: find what is guaranteed
A survey invites 250 people. Of 200 respondents, 150 support a proposal. The other 50 opinions are unknown. Approval requires at least 70% of everyone invited.
- Translate the target into a required number of supporters.
- Subtract known supporters to identify the missing requirement.
- Bound the overall support rate without assuming nonrespondents resemble respondents.
Conclusion: Approval is possible but not guaranteed. It occurs exactly when at least twenty-five nonrespondents support the proposal. The respondent percentage alone cannot settle it.
A recommendation is not a recorded outcome
Source 1: A training pilot enrolled 80 people; 64 completed it. Source 2: Expansion requires at least 75% completion and a cost no higher than $150 per enrollee. Source 3: Actual pilot cost was $11,600; a manager recommends expansion, but no expansion decision is reported.
- Map the sources: attendance, eligibility rule, and actual cost plus recommendation.
- Use enrolled people for both denominators; both numerical requirements are satisfied.
- Separate eligibility from action. Meeting the requirements and receiving a recommendation do not show that expansion was approved.
Conclusion: The pilot qualifies under the stated rule. Whether it was expanded cannot be determined.
Guided Practice: Try, then check
Work through each question on paper before opening its hint or explanation. This learning activity is separate from the scored Practice and Simulation pools.
Guided case: use a rule and an exception together
A delivery audit has three sources, summarized below. Decide whether the eligible on-time rate meets a target of at least 95%.
| Dispatch log | 100 orders; 92 on time; 8 late |
| Policy | Exclude customer-requested delays from numerator and denominator |
| Support log | Four of the eight late orders had customer-requested delays |
Which orders should be removed, and from which counts?
Show hint
The excluded orders were all late. Were they ever in the on-time numerator?
Check reasoning
Remove four orders from the total; the ninety-two on-time orders remain. Do not subtract the exclusion from the numerator when it was never counted there.
Does the policy-defined rate meet the target?
Show hint
Divide eligible on-time orders by eligible orders.
Check reasoning
The eligible on-time rate exceeds the target.
Does this prove the support team improved delivery speed?
Show hint
Separate reclassification under a rule from a measured change in delivery outcomes.
Check reasoning
No. The exclusion changes the reporting population; it does not show that any order arrived sooner or that support caused an improvement.
Follow the definition, locate the exception, and avoid converting an accounting adjustment into a causal claim.
Think Like a GMAT Test-Taker
Write a short evidence ledger containing source, period, metric and unit.
Compute derived metrics from the exact numerator and denominator in the relevant source.
Look for qualification language such as preliminary, only, at least or projected.
Before calculating, name the metric and its denominator. After calculating, return to the exact claim and its threshold.
For a claim using every or all, a single valid counterexample is enough to reject it.
For unknown outcomes, translate the percentage threshold into a count and determine exactly how many additional favorable outcomes are required.
For apparently conflicting figures, compare population, time period, units and exclusions before doing more calculations.
Record what each source is for, then return only to the facts needed for the current claim.
Common Traps
Do not replace actual figures with budgeted ones.
Different sources may use different time periods or denominators; compare rates only after using each source’s matching base.
An unsupported statement is not necessarily false, but cannot be concluded from the evidence.
A revised operational promise may not change the official reporting definition.
A higher total does not necessarily imply a higher rate; unequal group sizes matter.
Counting the same person twice when combining overlapping groups.
Assuming an observed sample rate is the rate of an unobserved group.
Confusing a guaranteed conclusion with one that is merely possible.
Do not turn a forecast, recommendation or eligibility decision into an observed outcome.