Split Half Reliability Calculator

Measure internal consistency from two matched test halves. Review corrected reliability, errors, and score patterns. Export results quickly for reports, audits, and classroom decisions.

Calculator Inputs

Use one score per respondent for the first half.
Match the order used in Half A.

Example Data Table

Respondent Half A Half B Total
1121123
2141327
3151429
49817
5111021
6131225
7161531
810919
9141327
10121123

Formula Used

Half-test correlation:
r = cov(X, Y) / (sdX × sdY)
Spearman-Brown corrected reliability:
rSB = 2r / (1 + r)
Guttman split-half coefficient:
rG = 2 × [1 - (Var(Half A) + Var(Half B)) / Var(Total)]
Standard error of measurement:
SEM = SD(Total) × √(1 - rSB)

The corrected coefficient estimates whole-test reliability after correlating two shorter halves. Guttman provides a useful comparison when the halves differ slightly in spread.

How to Use This Calculator

  1. Split the assessment into two comparable halves, such as odd-even items.
  2. Enter one score per respondent for Half A and Half B.
  3. Keep the respondent order identical in both score lists.
  4. Press Calculate Reliability to show the result block above the form.
  5. Review the half correlation, corrected reliability, Guttman value, and SEM.
  6. Use CSV or PDF export when sharing results with colleagues or attaching evidence to reports.

Interpretation Notes

A higher Spearman-Brown value suggests the full assessment produces more consistent scores across equivalent halves. Values below 0.70 often signal weak internal consistency, unstable content balance, or scoring differences between the two sections.

If the corrected reliability is high but Guttman is lower, inspect whether one half is easier or more variable. Large SEM values indicate wider uncertainty around observed total scores.

Frequently Asked Questions

1. What does split-half reliability measure?

It measures internal consistency by comparing scores from two equivalent halves of the same test. Strong agreement suggests the full test is reliably measuring one construct.

2. Why is Spearman-Brown correction needed?

The two halves are shorter than the full assessment. Spearman-Brown adjusts the half-test correlation upward to estimate reliability for the complete instrument.

3. When should I use an odd-even split?

Use odd-even splits when items are ordered similarly in difficulty and content. It usually creates balanced halves without manually redesigning the assessment.

4. Can unequal half lengths be used?

This page assumes paired half scores from comparable sections. If halves differ greatly in length, reliability estimates can become harder to interpret correctly.

5. What is a good reliability value?

Many applied settings prefer 0.70 or higher. High-stakes testing often targets 0.80 to 0.90, depending on consequences and score precision needs.

6. Why might reliability become negative?

Negative reliability usually means the halves do not move together. Check reverse-coded items, scoring mistakes, mismatched respondent order, or poor split design.

7. What does SEM add beyond reliability?

SEM converts reliability into score uncertainty. It helps you estimate how much an observed total score may vary because of measurement error.