When to use it
- •You have a multi-item scale intended to measure one construct.
- •You need reliability evidence before combining items into a composite score.
- •A reviewer expects a reliability figure alongside your measurement section.
Assumptions to check first
- •The items are unidimensional — alpha on a multidimensional set is close to meaningless.
- •Items are roughly equally related to the underlying construct (tau-equivalence); when they are not, alpha understates reliability and omega is the better estimate.
- •At least three items — alpha on two items is unstable.
What to report
- •Alpha itself, per scale, to two decimals.
- •The number of items in the scale.
- •Corrected item-total correlations if you dropped anything, and why.
APA 7 example
The six-item engagement scale showed good internal consistency, α = .88. All corrected item-total correlations exceeded .45, and no item's removal would have raised alpha appreciably.
Numbers are illustrative — the Advisor generates this sentence with your own results.
The mistake reviewers catch
Alpha rises with the number of items regardless of quality, so a long scale can post a comfortable α while measuring two different things. It is a necessary check, not evidence of validity — and the familiar .70 threshold is a convention, not a law.
Run it on your own data
Paste or upload your dataset — nothing leaves your browser. The Analysis Advisor checks the assumptions above, runs reliability, and drafts the APA Methods and Results text with your numbers in it.
Open the Analysis Advisor →Related analyses
Frequently asked questions
When should I use reliability?
Do the items in my scale hang together? You have a multi-item scale intended to measure one construct.
What do I need to report for reliability in APA 7?
Alpha itself, per scale, to two decimals. The number of items in the scale. Corrected item-total correlations if you dropped anything, and why.
What is the most common mistake with reliability?
Alpha rises with the number of items regardless of quality, so a long scale can post a comfortable α while measuring two different things. It is a necessary check, not evidence of validity — and the familiar .70 threshold is a convention, not a law.