Skip to content
Omnikalc

Spearman-Brown calculator: how long must a test be to reach the reliability you need?

Enter the current reliability, the change in test length and a target to get the predicted reliability and the number of items required.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Value out of range

Results

Predicted reliability after the length change

0.824

Acceptable for most uses
Length factor needed to reach the target
3.86
Items needed to reach the target
78
Items to add
58
Predicted reliability if the test were cut to half
0.538

The Spearman-Brown prophecy formula links the reliability of a test to its length. The Spearman-Brown calculator takes the current reliability, a length factor and a target reliability, and gives the predicted reliability, the factor needed to reach the target and the number of items to add.

How the calculation works

  • The Spearman-Brown prophecy formula predicts the reliability of a test whose length is multiplied by a factor k: r_new = k × r ÷ (1 + (k − 1) × r), where r is the current reliability.
  • To find the factor needed for a target reliability r_t, solve the formula for k: k = r_t × (1 − r) ÷ (r × (1 − r_t)).
  • Multiply that factor by the current number of items and round up to get the number of items the longer test needs. The difference is the number of items to add.
  • The formula assumes that the added items are parallel to the existing ones, which means they measure the same thing with the same quality.

Example

A 20-item test has a reliability of 0.70. Doubling it (k = 2) gives 1.4 ÷ 1.7 = 0.824. To reach 0.90, the factor is 0.9 × 0.3 ÷ (0.7 × 0.1) = 3.857, which means 20 × 3.857 = 77.1, so 78 items, or 58 more than today. Cutting the test in half would drop reliability to 0.538.

Why it matters

Test designers often ask whether a short scale is reliable enough or how many more questions they need. This calculator answers before any new items are written, which saves the cost of piloting versions that were never going to reach the target.

Good to know

The prediction rests on the assumption that new items behave like the existing ones. Reliability is also specific to a population and a purpose, so a value measured in one group may not transfer to another. Use the result to plan, then check the real reliability once the longer version has been tried.

The Cronbach’s alpha calculator estimates the reliability to start from, and the intraclass correlation calculator measures agreement between raters.

Frequently asked questions

What does the Spearman-Brown formula do?

It predicts how the reliability of a test changes when you add or remove items of similar quality. Longer tests are more reliable because random errors on single items tend to cancel out, and the formula puts a number on that gain.

Does adding items always help?

Adding good parallel items always raises reliability, but with diminishing returns. Going from 0.70 to 0.82 takes a doubling, while going from 0.90 to 0.95 takes another doubling. At some point the gain is not worth the burden on test takers.

Can I use it for split-half reliability?

Yes. If you correlate two halves of a test, set k to 2 to estimate the reliability of the full test from the half-test correlation. This is the classic use of the formula and it corrects the underestimate that comes from using only half of the items.

What are its limits?

It assumes the new items are parallel to the old ones, which is rarely perfect. If the added items are weaker, harder to understand or about another topic, the real reliability will be lower than predicted, so treat the result as an upper estimate.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.