What is a Scale Level? Definition and Example

In empirical research, at the beginning of quantitative data collection, such as in an online survey with a questionnaire, various systematic measurement standards or scales are used to make the examined subject measurable at all.

General Definition of a Scale Level

Scale levels (= simplified also called scales) are measurement instruments with which properties of examined characteristics (e.g. gender, age, income, grades) and their characteristic manifestations (e.g. female/male) can be determined numerically. Which scale is specifically used depends on which characteristics are to be captured by individual questions of the survey. Not all desired information can be equally well represented with the same scale. While a person's income can be very easily translated into numbers, this is far from the case for a characteristic such as gender or sexual orientation.

The more complex the desired information content, the more complex or higher the corresponding scale level must be. Therefore, there are different scale levels that are more or less suitable for different types of desired data.

Different Types of Scale Levels

In science, four measurement or scale levels are common, each allowing different substantive statements – from simple to very complex – about the desired characteristic manifestations:

Definition of "metric" and "non-metric":

Scale levels can be represented by metric or non-metric values. In metric scales, the answer options have a directly usable numerical value. These include interval and ratio scales. In non-metric scales, the answer options are not bound to a direct numerical value. These include nominal and ordinal scales.

 

1. Nominal Scale (for categorization - "same / different")

The nominal scale is a categorical and non-metric scale and offers the lowest statistical information content. With this scale, simple statements about equality or difference can be made, whereby it can be calculated how often a particular characteristic (e.g. how many women participated) occurs in a study.

Examples: gender, profession, car brand, marital status, party affiliation
 

 

2. The Ordinal Scale (for ranking with ordinal numbers, without defined intervals - "smaller / larger")

The ordinal scale is also a non-metric scale that, while not offering a direct numerical value, does allow a substantive and ascending ranking over the characteristic manifestations. With this scale, statements about greater-smaller relationships (education level, vehicle class, etc.) can be made. However, the ordinal scale cannot say how much better or higher the third answer is than the second answer, since the exact intervals between two characteristics cannot be measured.

Examples: school grades, age groups, education level, military rank
 

 

3. Interval Scale (for ranking with equally large intervals, without natural zero point - "differences")

The interval scale is a metric scale where the intervals between two answers in the form of consecutive numerical values are equally large. With this scale, statements about equally large characteristics (temperature measurement, years) and their differences can be made.

Examples: Celsius temperature, seasons, IQ, exam grades
 

 

4. Ratio Scale (for ranking with equally large intervals and natural zero point - "ratios")

The ratio scale is the highest metric scale level that not only has a directly usable numerical value with equally large intervals between the individual units, but also quantitatively shows the ratio of their distance values to one another. With this scale, complex statements about relationships (body height, monetary units, physical measures) between different characteristics or measurement units can be made.

Examples: age, income, body height, body weight, area size

 

Sometimes these four types are also combined into three variants, where the cardinal scale (= also called metric scale) includes both interval and ratio scales. In principle, the highest possible scale level is always aimed for, because then the information content of the examined data is also greater. The higher the scale level, the more mathematical procedures can be used in data evaluation. The following overview summarizes the most common scale levels in ascending order of their complexity, from the simple nominal scale to the most comprehensive ratio scale.