Skip to content
phdguide
Statistics

How to Analyse Likert Scale Data in SPSS (Without Getting It Wrong)

Mean or median? Parametric or not? Item vs summated scale? The practical, defensible way to analyse Likert data in SPSS — the workflow, the tests and the classic examiner objections.

The phdguide Research Team 9 July 2026 2 min read

Likert scales power most survey research in management and social science — and generate more viva arguments than any other data type, because of one old dispute: are Likert responses ordinal or interval? Here's the honest position and a defensible SPSS workflow, so your analysis survives both your data and your examiners.

The distinction that resolves the debate

A single Likert item (one 1–5 question) is ordinal — the distance between 'agree' and 'strongly agree' isn't guaranteed equal. But a summated Likert scale (the mean of several items measuring one construct, e.g. five job-satisfaction items) behaves approximately continuously, and treating it with parametric statistics is standard, defensible practice across the literature. Most thesis analysis operates on summated scales — which is why means, t-tests and regression are everywhere and mostly fine.

The SPSS workflow, in order

  1. 1Screen the data — missing values, straight-lining respondents, out-of-range codes.
  2. 2Reverse-score negatively worded items (Transform → Recode into Different Variables) — forgetting this silently wrecks reliability.
  3. 3Test reliability: Cronbach's alpha per construct (Analyze → Scale → Reliability Analysis); 0.70+ is conventionally acceptable.
  4. 4Compute summated scales (Transform → Compute Variable, mean of items).
  5. 5Describe: means and SDs for scales; frequencies/medians for single items.
  6. 6Check distributions (skewness/kurtosis) to justify test choices.
  7. 7Run inferential tests matched to your questions (below).

Which test for which question

  • Compare two groups on a scale: independent-samples t-test (Mann-Whitney U if badly skewed).
  • Compare 3+ groups: ANOVA (Kruskal-Wallis as the non-parametric fallback).
  • Association between scales: Pearson correlation (Spearman for single items).
  • Predict an outcome from several constructs: multiple regression — or SEM in AMOS/SmartPLS when constructs are latent.
  • Single items: stick to frequencies, medians, chi-square — don't average one item and call it a construct.
The two examiner objections to pre-empt

(1) 'You used parametric tests on ordinal data' — answer: analysis was on summated multi-item scales with acceptable reliability, per standard practice. (2) 'Your scale isn't reliable' — answer with your Cronbach's alpha table. Report both proactively and the objections never land.

If you're staring at SPSS output unsure what's defensible, our data analysis help pairs you with a statistician-mentor who works through your actual dataset with you — see also SPSS mentoring and the SPSS thesis guide.

Frequently asked

Can I use a t-test on Likert data?+

On summated multi-item scales, yes — it's standard practice. On a single Likert item, prefer non-parametric alternatives like Mann-Whitney U, or report medians and frequencies.

What Cronbach's alpha is acceptable?+

0.70+ is the usual threshold, 0.80+ comfortable. Very high values (0.95+) can signal redundant items. Report alpha for every construct before using its summated scale.

Should I analyse a 5-point and 7-point Likert scale differently?+

The workflow is identical. Seven points can capture slightly more variance; what matters far more is item quality, reliability and matching tests to your questions.

About the author
The phdguide Research Team
Research mentors & senior academics

phdguide's mentors are senior academics, former supervisors, statisticians and publication specialists with 25+ years of combined experience guiding MBA, MPhil and PhD scholars from topic to viva.

Ethical, compliant guidance: We provide academic support, mentoring, analysis, editing and structuring — not authorship. Your work stays compliant with university policies.

Continue your research

Services, free tools and templates that pair well with this article.

Related services
Related tools
Related templates
Related workshops

Not sure where to start?

Book a free 15-minute consult. We'll map your next three steps — no obligation.

Ethical, compliant guidance

We provide academic support, mentoring, analysis, editing and structuring — not authorship. Your work stays compliant with university policies.