Cronbach's Alpha & Reliability Testing Explained

Cronbach's alpha is the most commonly reported reliability statistic in thesis and dissertation methods sections — and one of the most commonly misinterpreted. Here's what it actually tells you.

What it measures

Cronbach's alpha measures internal consistency — how closely related a set of items are as a group, as an indicator that they're measuring the same underlying construct. It's calculated from the average correlation among all items on a scale, and ranges from 0 to 1.

Interpreting the number

• Below 0.60 — generally considered unacceptable for research use.
• 0.60–0.69 — questionable, often needs revision.
• 0.70–0.79 — acceptable for most research purposes.
• 0.80–0.89 — good.
• 0.90 and above — excellent, but see the caveat below.

Why higher isn't always better

An alpha above roughly 0.95 often signals item redundancy — several items asking essentially the same question in slightly different words, which inflates the internal-consistency estimate without adding real measurement information. This is worth flagging in your write-up rather than simply reporting an impressively high number.

What alpha doesn't tell you

A high alpha confirms your items correlate with each other — it says nothing about whether they measure the right construct (that's validity, covered in our survey validation guide), and it can be artificially inflated just by adding more items to a scale, regardless of quality. Report alpha alongside validity evidence, not as a substitute for it.

Reporting it correctly

State the alpha for your specific sample (published alphas from the instrument's original validation study don't automatically transfer to your data), and report it per subscale if your instrument has more than one dimension — a single overall alpha can mask a weak subscale.