In our last article, we explored the foundational distinction between descriptive and inferential statistics, two broad approaches to making sense of data. Building on that foundation, we now turn to another equally important concept: levels of measurement. When working with data, not all variables are created equal. Before diving into any statistical analysis, it’s essential to understand these levels, nominal, ordinal, interval, and ratio, as they describe the nature of the information stored within a variable, and have real consequences for how you collect, visualize, and analyze your data. Choose the wrong statistical test for your level of measurement, and your conclusions could be meaningless. This article breaks down each level, what makes them distinct, and why getting this right is the first step toward sound statistical thinking.
All variables we work with in Statistics have different levels of measurement or scales of measurement. There are mainly 4 levels of measurement in Statistics.
- Nominal
- Ordinal
- Interval
- Ratio
So, why do we even care about levels of measurement? Well the simple answer is this, the levels of measurement determine the kind of analysis and interpretations we can draw from the data, in some cases the level of measurement also communicate to us how we can go about collecting this data, hence, which specifies the possible statistical tests and analysis that can be done on the data.
NOTE
In some Statistical books or papers, you’ll notice that they provide 3 types of measurements instead of the 4 we listed above. The reasons for this is because, Interval and Ratio measure are often used for the same statistical analysis, hence some group them under the same umbrella and call them “Metrics Level Of Measurements”
Importance Of Level Of Measurement
Help us decide on what statistical tests we can conduct.
Different levels of measurements of variables communicate to us what possible statistical metrics we can use. Example when working with Nominal data, using metrics like mean and variance makes no sense, but it makes sense to use it when we have Metric Level of measurement eg: Average height of people in a study, average score in a given test etc.
Data Visualization
When it comes to data visualization, different charts make interpretation more easier than other charts depending on the level of measurement of the underlying variable. Histogram charts make sense if the variable level of measurement is metric. Pie charts and bar charts are suitable for Nominal data, Box plots for Ordinal data, Scatter plots for metric data.
Level Of Measurement In Experiment Design
When designing an experiment, it is crucial to collect the right type of data(level of measurement) that will allow you to conduct statistical tests that answer the underlying research question at the end of the research process.
Nominal Level of Measurement
The nominal level of measurement classifies data into distinct categories that have no inherent order or ranking. The categories are simply labels used to identify or name groups. This is the lowest level of measurement.
Characteristics
- Categories are mutually exclusive
- No order among categories
- Numbers (if used) are labels only
- No meaningful arithmetic operations
Examples
- Gender (male, female, non-binary)
- Blood group (A, B, AB, O)
- Nationality
- Eye color
- Type of operating system (Windows, Linux, macOS)
- Yes / No responses
What you can do with nominal data
- Count frequencies
- Calculate mode
- Use bar charts or pie charts
Ordinal Level of Measurement
The ordinal level of measurement classifies data into categories that have a meaningful order, but the distance between categories is not measurable or equal. In other words, differences between ranks do not have a statistical meaning, not quantifiable.
Characteristics
- Categories are ranked
- Order matters
- Differences between ranks are not equal or known
- Arithmetic operations are not meaningful
Examples
- Education level (high school, bachelor’s, master’s, PhD)
- Satisfaction rating (poor, fair, good, very good, excellent)
- Class rank (1st, 2nd, 3rd)
- Likert scale responses (strongly disagree → strongly agree)
- Socioeconomic status (low, middle, high)
What you can do with ordinal data
- Compare rankings
- Calculate median and percentiles
- Use bar charts
Metric Level Of Measurement
Measurements that use the metric level of measurement are those measured on interval or ratio scales. In short, the values are numeric, ordered, and the distances between values are meaningful. This type of level of measurement is very similary to ordinal variables, the only difference being that the interval between values are equal, allowing the calculation of the differences and sums.
1. Interval scale (metric)
- Equal intervals, no true zero
- Differences make sense, ratios don’t
Examples:
- Temperature in °C or °F
- IQ scores
- Calendar years (e.g., 1990, 2000)
- Time of day (e.g., 14:00, 16:00)
2. Ratio scale (metric)
- Equal intervals and a true zero
- Both differences and ratios make sense
Examples:
- Height (cm, m)
- Weight (kg)
- Age (years)
- Income (€)
- Reaction time (ms)
- Distance (km)
- Number of children
- Duration (seconds, minutes)
Conclusion
Congratulations for making it to the end! Understanding levels of measurement is not just a theoretical exercise, it is a practical skill that shapes every decision you make in a statistical analysis. From choosing the right chart to selecting the appropriate statistical test, knowing whether your variable is nominal, ordinal, interval, or ratio keeps your analysis grounded and your conclusions valid. As we move forward in this series, you’ll notice that this concept keeps showing up, whether we’re discussing hypothesis testing, regression, or data visualization. Think of levels of measurement as the grammar of statistics, get it wrong, and the rest of the sentence falls apart. Now that you have a solid grasp of this concept, you’re better equipped to approach data with clarity and confidence. In our next article, we’ll continue building on these foundations as we dive deeper into the world of statistical analysis.
Other platforms where you can reach out to me:
Happy coding! And see you next time, the world keeps spinning.