Is my questionnaire reliable enough to report?
Compute internal consistency from an average inter-item correlation, or from item and total-score variances — and get a straight answer about what the number proves to an examiner, and what it does not.
Free guide: choosing between qualitative, quantitative and mixed methods
Reliability statistics only apply once you have committed to a measurement approach. This guide helps you make that call deliberately. We’ll email it to you.
What alpha demonstrates in a thesis, and what it doesn’t
Cronbach’s alpha measures internal consistency: the extent to which the items on a scale move together, as they should if they tap one construct. There are two equivalent routes to it, and this calculator takes either.
- Standardised alpha — α = k·r̄ / (1 + (k−1)·r̄), from the number of items and the average inter-item correlation.
- Raw alpha — α = (k / (k−1)) · (1 − Σσ²ᵢ / σ²total), from item variances and the variance of the summed score.
Two cautions worth internalising before your viva. First, alpha rises with the number of items regardless of their quality — a long mediocre scale can post a comfortable figure, so a high alpha on a forty-item instrument is much weaker evidence than the same figure on eight items. Second, and more importantly, reliability is not validity. A set of items can agree beautifully with one another while measuring something other than the construct you named. Examiners ask about the second far more often than the first.
For item-level diagnostics — item-total correlations, alpha-if-item-deleted — and for modern alternatives such as McDonald’s omega, run your raw data through jamovi, JASP, R (the psych package) or SPSS.
Frequently asked questions
What alpha do I need for my thesis?
The conventional bands are 0.9 and above excellent, 0.8 to 0.9 good, 0.7 to 0.8 acceptable, 0.6 to 0.7 questionable, and below 0.6 poor. Most committees expect at least 0.7 for an established scale. Above about 0.95 is worth a comment rather than a celebration — it often means items are near-duplicates. Standards genuinely vary by field, so check what recent theses in your department report.
My alpha is 0.62. Can I still use the scale?
Sometimes, if you handle it openly. Look first at whether a small number of items are dragging it down — alpha-if-item-deleted will tell you, and dropping a badly worded item is defensible if you report that you did it and why. If the scale is exploratory or newly developed, a lower alpha is more tolerable. What is not defensible is reporting 0.62 without comment and hoping nobody asks.
My alpha came out negative. Is that a bug?
No, and it is informative. A negative alpha almost always means some items are reverse-keyed and have not been recoded — that is the first thing to check. If recoding does not fix it, the items are not measuring one coherent construct, and the scale needs rethinking rather than rescuing.
Does a high alpha mean my scale is valid?
No, and this is the question most likely to come up in your viva. Alpha shows the items are consistent with each other, not that they measure what you claim. Validity needs separate evidence: content validity from expert review, construct validity from factor analysis, criterion validity from correlation with an external standard. Report alpha as one part of a psychometric argument, never as the whole of it.
Should I report McDonald's omega instead?
Increasingly, yes, or both. Alpha assumes all items contribute equally to the construct, which is rarely true; omega relaxes that and is generally the better estimate. It needs raw data and a factor model, so you cannot compute it here, but jamovi and the R psych package both produce it easily. Reporting both signals methodological currency.
Is anything I enter stored or uploaded?
No. The calculation runs entirely in your browser and nothing is saved.