← GMAT Course

DATA INSIGHTS · GMAT-DI-MSRC · DRAFT

Multi-Source Reasoning: Conflicting Evidence

Build the concept, interpret evidence, and apply it in a separate assessment pool.

Learn: Evidence and decisions

Resolve the meaning before the arithmetic

Two reports can be arithmetically correct and still measure different things. Write down each numerator, denominator, time period and eligibility rule. Use the definition specified in the question.

Separate a threshold from a forecast

A forecast predicts a quantity; a policy determines whether a result passes. Exceeding a forecast count need not mean failing a percentage-based policy. Check both claims separately.

Bound what is unknown

Do not assign the observed rate to an unobserved group without permission. Compute how many additional successes are needed, then ask whether the sources guarantee that count.

Normalize before comparing

When sources report counts and rates, calculate each rate with its matching denominator. A larger count can coexist with a smaller rate.

Reconcile definitions and adjustments

Separate reported values from adjusted values, and use the definition named in the question. Do not combine numbers that refer to different periods or scopes.

Learning Objectives

  • Reconcile conflicting reports by checking definitions, units and denominators.
  • Distinguish percentage-point differences from relative changes.
  • Identify when incomplete evidence permits multiple outcomes.

Prerequisite Knowledge

  • Percentages, ratios and weighted averages.
  • Inclusive and strict numerical thresholds.

Worked Examples

Compare an actual rate with a threshold

A quarter has 1,350 dispatches and 1,242 of those orders delivered on time. The service target is at least 92%.

  1. Use the definition from the policy source.
  2. Compare the computed rate with the inclusive threshold.

Conclusion: The actual rate meets an “at least 92%” target exactly. “At least” includes equality.

Resolve two correct but different percentages

A workshop registered 240 people; 40 canceled before the event. There were 180 attendees. A dashboard divides by registrations, while the attendance rule excludes cancellations.

  1. The dashboard uses all registrations.
  2. The rule uses the eligible registrations after cancellation.
  3. Compare the reported rates using percentage points.

Conclusion: Both calculations are arithmetically correct. Only 90% applies to the stated attendance rule; the difference is a change in denominator.

Replace an unsupported estimate with a bound

Of 300 applicants, 180 have been assessed and 144 qualified. A program needs at least 70% of all applicants to qualify; the other 120 results are unknown.

  1. Find the required total qualified count.
  2. Subtract confirmed qualifications.
  3. Check whether the unknown group guarantees the required count.

Conclusion: Eligibility is possible but not guaranteed. The observed 80% rate does not establish the unknown applicants' outcomes.

Compare rates, then combine the groups

Unit A completes 72 of 90 deliveries on time. Unit B completes 108 of 120. Find each rate and the combined rate.

  1. Use each unit’s own denominator.
  2. Add counts before dividing to find the combined rate.

Conclusion: Unit B has the higher rate; the combined rate is about 85.7%. Averaging 80% and 90% directly would be incorrect because the group sizes differ.

Distinguish reported and adjusted outcomes

A report lists $48,000 profit, including an $8,000 one-time credit. A policy note defines recurring profit as reported profit less one-time credits and sets a $42,000 target.

  1. Apply the stated adjustment.
  2. Compare the adjusted figure with the target.

Conclusion: Recurring profit is $40,000, so it is $2,000 below target. The unadjusted figure answers a different question.

Think Like a GMAT Test-Taker

Read the question first and note the needed source, metric, and period.

Build a short evidence chain: source value → defined calculation → target or claim.

For percent change, divide the change by the original value, not the new value.

Use “Cannot be determined” only when a needed fact is absent, not when a calculation is inconvenient.

Common Traps

Treating a forecast as an actual result.

Using a target as if it were observed performance.

Confusing exactly 10% with at least 10%.

Ignoring whether a threshold includes equality.

Home