What is Ordinal Scale?
An ordinal scale records responses that have a clear order (for example: low → high) but where the gaps between points aren’t assumed equal. It’s commonly used in surveys—most notably in Likert-style questions—to capture ranked opinions or attitudes.
An ordinal scale arranges responses into a meaningful sequence: respondents can be placed higher or lower on the scale, but the exact distance between adjacent options is unknown. Examples include ratings like “never / sometimes / often,” satisfaction scales from “very dissatisfied” to “very satisfied,” and preference rankings. Ordinal differs from nominal data (which has categories without order) and from interval/ratio data (where numeric differences are meaningful and consistent). In analysis you can report counts, percentages, medians and percentiles and use non‑parametric tests; you should be cautious about calculating or interpreting arithmetic means unless you have a strong reason to treat the scale as interval.
Usage example
Survey question: “How satisfied are you with the service?” Options: Very dissatisfied, Dissatisfied, Neutral, Satisfied, Very satisfied. These responses form an ordinal scale: you can tell which responses are higher or lower but not assume the distance between each label is equal. You might report the percentage of respondents who were “satisfied” or use the median satisfaction level to summarise results.
Practical application
Why it matters: ordinal scales are simple for participants to understand and quick to analyse, making them ideal for inclusive, multilingual surveys where you want consistent, comparable answers across communities. Correctly recognising a question as ordinal guides how you summarise and test results (use medians and non‑parametric tests rather than assuming equal intervals). In multilingual contexts, preserving the intended order and meaning of each labelled option is crucial—differences in translation or cultural interpretation can change how respondents perceive the scale. Best practice: label every point, keep the number of points reasonable (commonly 3–7), balance positive and negative options, and use Hearo’s translation review and participant feedback tools to ensure labels remain ordered and meaningful across languages.
FAQ
Can I calculate the average (mean) of ordinal data?
Generally you should avoid relying on the mean for ordinal data because the distances between points aren’t guaranteed equal. Use medians, modes, percentage distributions or non‑parametric tests (e.g. Mann‑Whitney, Kruskal‑Wallis). In some applied settings researchers treat aggregated multi‑item scales as interval, but do so with caution and clear justification.
How many points should I use on an ordinal scale?
Most surveys use between 3 and 7 points; 5 and 7 are common because they balance discrimination with simplicity. Fewer points are quicker for respondents and reduce translation ambiguity; more points can give finer detail but may be harder to interpret consistently across languages and cultures.
Is a Likert scale the same as an ordinal scale?
A Likert item is a specific type of ordinal scale: each item uses ordered response labels (e.g. strongly disagree → strongly agree). When you combine several Likert items into a composite score, some teams treat the result as interval for analysis, but each single Likert item remains ordinal and should be analysed accordingly unless you justify otherwise.
How should I handle ordinal scales when translating surveys?
Ensure each label’s meaning and order are preserved in translation—literal word choices can shift perceived intensity or introduce bias. Label every response point rather than relying on numeric markers alone, test translations with native speakers, and use participant feedback to flag problematic wording so you can refine translations over time.