Test Analysis
Normalized Learning Gain Calculator
Express pre-to-post improvement as a share of the gain still possible below the maximum score.
Enter the two averages and maximum
Ceiling position affects interpretation
A high pretest leaves little possible gain and can make the statistic sensitive to small changes.
Report the pretest level and raw change beside normalized gain.
A fuller learning report can report raw and baseline-relative change or inspect which objectives contributed to the result.
Calculation method
The denominator is the improvement available above the baseline, so the baseline must be below the maximum.
Show raw change with normalized gain
- Verify a common maximum.
- Check baseline is below it.
- Calculate raw and possible gain.
- State the aggregation method.
Gain relative to the remaining scale
Moving from 45 to 70 on a 100-point assessment gains 25 of the 55 points available, a normalized gain of 0.455.
Aggregation method matters
Gain from group averages can differ from the average of individual normalized gains.
Choose the method before analysis and identify it in the report.
Avoid detaching gain from its scores
Publish the pretest, post-test, maximum, raw change, and normalized result together.
If the scale has a meaningful minimum other than zero, confirm that the chosen normalized-gain convention remains suitable.
Large negative or positive values deserve record-level review for mismatches and data-entry errors.
Do not translate descriptive gain into a causal effect without a comparison design and assumptions that support that claim.
Averages near the maximum compress the available-gain denominator, so modest measurement error can create a large change in normalized gain.
Report confidence or variability when supported by the data rather than treating three decimal places as certainty.
If individual gains are averaged, decide how to handle learners starting at the maximum before examining results; excluding them after the fact can bias the summary.
Normalized gain is most interpretable when the assessment maximum represents meaningful attainable performance and pre/post scores lie on the same interval-like scale. If the test has strong floor or ceiling effects, report raw change and score distributions rather than relying on the normalized value alone.
Where decision makers are unfamiliar with normalized gain, lead with the original scores and raw difference. Explain the denominator in plain language so the result is not confused with percent correct, percentage-point growth, or the share of learners who improved.
Questions about the result
Can normalized gain be negative?
Yes, when the post-test score is lower.
What if the pretest equals the maximum?
No gain remains available, so this formula is undefined.
Is 0.5 the same as 50 percentage points?
No. It means half of the available gain was achieved.