DataTerminology.comNEW
A measurement scale that orders categories by rank without assuming equal intervals between them.
DESCRIPTION
Ordinal data encode order or rank among categories but not the magnitude of differences between ranks. Common examples include survey responses such as ‘poor’ to ‘excellent’ and military ranks. Appropriate summaries include medians, percentiles, and nonparametric tests rather than means. Ordinal scales are widely used in social sciences, customer feedback, and risk ratings. Careful treatment avoids inappropriate arithmetic on rank codes, which can mislead analysis.
ALTERNATIVE DEFINITIONS
Encyclopaedia Britannica
Ordinal scale, in statistics, a scale of measurement that permits the ranking or ordering of items according to some criterion but does not indicate the magnitude of difference between them.
Highlights ranking capability without information on magnitude of differences.
NIST/SEMATECH e-Handbook of Statistical Methods
Ordinal data have a natural order, but the intervals between successive values are not equal or known.
Emphasizes unknown or unequal intervals between ordered categories.
Penn State STAT 501
Ordinal level of measurement classifies data into categories that can be ranked; however, precise differences between the ranks do not exist.
Educational framing that differences between ranks are not meaningful.
RELATED TERMS
Likert Scale Nonparametric Statistics Median Spearman Rank Correlation Measurement Scale
COMPETING TERMS
Nominal Scale Interval (NOIR) Ratio (NOIR)
KNOWLEDGE AREAS
Mathematics Data Science Data Modeling Data Quality
TARGET AUDIENCE
Data Scientist Data Analyst Data Steward Data Modeler BI Specialist
EXTERNAL SOURCES
- Wikipedia - Ordinal data
- NIST - NIST guidance on ordinal data
- Penn State - University lesson on measurement levels
TOP QUESTIONS
When should I use the Ordinal (NOIR) scale?
Use ordinal scales for ranked categories such as satisfaction levels or severity ratings. The key is order matters but differences are not measurable. This avoids implying false precision.
Category: Requirements
What statistics are appropriate for Ordinal (NOIR) data?
Use medians, percentiles, rank correlations, and nonparametric tests like Mann–Whitney. Means and standard deviations are generally inappropriate. This preserves ordinal meaning.
Category: Best practices
What is a common mistake with Ordinal (NOIR) data?
Treating ordinal codes as interval data and averaging them is a common error. It can introduce misleading conclusions. Preserve rank-only semantics in analysis.
Category: Mistakes
How does Ordinal (NOIR) differ from Interval Scale?
Ordinal orders categories without equal spacing, while interval has equal intervals but no true zero. Only interval supports meaningful differences. Ordinal suits ranks.
Category: Comparison
Is Likert data truly Ordinal (NOIR)?
Most Likert response items are considered ordinal because distance between choices is not guaranteed. Some aggregate scores approximate interval. Debate depends on design and validation.
Category: Opinion
Can I apply regression to Ordinal (NOIR) outcomes?
Yes, use ordinal regression models such as proportional odds or ordered logit. These respect ordering without assuming interval distances. Avoid linear regression on raw codes.
Category: Application
How do I visualize Ordinal (NOIR) data?
Use ordered bar charts, cumulative distributions, and boxen-like rank plots. Maintain the natural ordering on axes. Avoid implying uniform spacing visually.
Category: Visualization
What quality checks suit Ordinal (NOIR) variables?
Check monotonic relationships, response balance, and out-of-order encodings. Validate category labels and ordering. Monitor shifts in category proportions over time.
Category: Data Quality
Why is Ordinal (NOIR) still relevant in analytics?
Many human judgments are ordinal—satisfaction, risk, preferences. Proper handling enables reliable insights without overstating precision. It’s foundational in survey and behavioral data.
Category: Relevancy