Test Analysis
Scenario comparison
Pre/Post Test Improvement Calculator
Compare average pretest and post-test scores on a shared scale, including proportional improvement from baseline.
Enter the paired assessment averages
Change between two group averages
A group average moving from 58% to 76% improves by 18 percentage points, or 31.0% relative to the baseline average.
Match the assessments before comparing
Content coverage, scale, testing conditions, and scoring must be sufficiently aligned.
A harder post-test can understate learning, while repeated identical items can introduce practice effects.
The calendar behind the result can express change against remaining possible gain.
Average change can conceal different paths
Some learners may improve while others decline even when the mean rises.
Use paired records and a change distribution when individual progress matters.
The completed record may then compare individual practice attempts.
Working through the denominator
Boundary: The calculation compares group averages; it does not establish that the same individuals improved unless the records are paired.
Matched and unmatched summaries
For paired learners, calculate each change when distribution and individual trajectories matter. For repeated cross-sectional groups, describe the difference in averages without calling it each learner's gain.
State the sample count at both occasions and the number successfully matched.
Align the two assessment occasions
- Confirm a shared score scale.
- Reconcile the pre and post populations.
- Calculate the point change.
- Inspect individual and subgroup patterns.
Questions that affect interpretation
Is relative improvement the same as normalized gain?
No. Relative improvement divides by the baseline; normalized gain divides by the remaining possible gain.
Can unmatched groups be compared?
They can be described, but the change should not be presented as paired learner growth.
Does improvement prove the instruction caused it?
No. Causal claims require an appropriate evaluation design.
Preserve population and instrument details
Save sample sizes, matching rules, assessment versions, dates, accommodations, and unrounded means.
Report attrition between testing occasions because missing post-tests can change the composition of the group.
If scores are not percentages, enter them only after defensible conversion to a common scale.
Pair the mean change with spread or individual change summaries when the underlying data are available.
Relative improvement becomes unstable when the baseline is very close to zero, which is why the raw percentage-point change should remain the primary transparent result.
If the post-test scale was equated statistically rather than literally identical, use the program's reported scale scores and uncertainty instead of converting raw points informally.
When only aggregate averages are available, avoid describing an eighteen-point mean increase as though every learner gained eighteen points. Report it as a group-level difference. Paired data are needed to count improvers, unchanged learners, and declines or to describe the distribution of individual change.