When doing research, it is important to understand how data are classified and how they are measured.
Understanding types of data and levels of measurement is important in many ways:
i) it helps with accurate coding of the data.
ii) it helps you decide on the appropriate data analysis techniques.
iii) it helps you interpret your results correctly.
This article discusses the various types of data and their levels of measurement.
Types of data
Data can be classified into two types: numerical or categorical.
Numerical data
These is data that is in numbers. They are of two types also: continuous or discrete data.
Numerical data is quantitative in nature.
Continuous data
It’s data that is infinite and impossible to count because it can take on so many values.
Continuous data can be sub-divided into various parts.
Discrete data on the other hand are numerical data that are finite and possible to count.
Discrete data cannot be sub-divided into different parts.
- Number of children a person has.
- Number of pet dogs a person has.
- Number of teachers in a school.
- Number of doctors in a hospital.
- Temperature.
- Weight in kilograms or pounds.
- Height in feet.
- Distance in kilometers.
- Time in hours.
Categorical data
Categorical data is data that is in categories or groups instead of in numbers.
Categorical data is qualitative in nature.
Examples of categorical data include: gender (male or female), race (Black, Caucasian, Native Indians, Asian, Hispanic etc), type of housing (apartment, bungalow, maisonette etc), highest level of education (pre-primary, primary, secondary, tertiary), tribe, religion etc.
Each type of data is measured differently.
Levels of measurement
There are four levels (scales) of measurement of data. The level of measurement used depends on what type of data it is.
Nominal data
In nominal data, numbers are assigned to objects with the sole purpose of differentiating one object from another. The numbers therefore have no real meaning.
This level of measurement is used for categorical data.
Nominal data does not have any ranking or order.
Examples:
| Variable | How it’s measured | Explanation |
|---|---|---|
| Gender | 1 – male; 2 – female | The numbers assigned to male and female are arbitrary and do not mean anything. It does not mean that females are more superior to males. |
| Religion | 1 – Christian; 2 – Muslim; 3 – Hinduism; 4 – Atheist | The numbers assigned to the four religions are also arbitrary and have no real meaning. |
| Number on basketball jerseys | 1, 2, 3, 4, …. etc | The numbers on the jerseys are used to differentiate the players. It does not mean that a player wearing jersey number 10 is better than a player wearing jersey number 2. |
Ordinal data
For ordinal data, numbers are assigned to objects to differentiate them but the numbers have a meaningful order and can therefore be ranked.
The placement of the number therefore matters.
However, the intervals between the numbers are not necessarily equal.
Examples:
| Variable | How it’s measured | Explanation |
|---|---|---|
| Birth order | 1st, 2nd, 3rd, 4th … last | The first born has a higher birth order than the second born; second born has a higher birth order than the third born etc. However, the interval between the 1st and 2nd born may not be the same as the interval between the 2nd and 3rd born. The spacing may be different. |
| Highest level of education | 1 – pre-primary; 2 – Primary; 3 – Secondary; 4 – Tertiary | A person with tertiary education is ranked higher than a person with secondary-level education; while the person with secondary education is ranked higher than the one with primary-level education. |
Interval data
For interval data, the numbers have order but the difference between the adjacent numbers is equal across the measurement scale.
Interval data also do not have a true zero (0), meaning that any 0 in the scale is part of it and does not imply the absence of the object being measured.
An often-cited example of interval data is temperature in degrees Celsius or degrees Farenheight. The 0 in these scales refer to a degree of temperature (which is quite cold) rather than the absence of temperature.
Ratio data
Ratio data have the characteristics of all the above three levels of measurement (nominal, ordinal and interval).
Like nominal data, ratio data have numbers that differentiate the various objects.
Like ordinal data, ratio data have a meaningful order.
Like interval data, ratio data have meaningful intervals between adjacent numbers. However, ratio data differ from interval data in that they have a true zero (0), which means the absence of the object being measured.
In addition to the above characteristics, ratio data can be sub-divided to get fractions of the numbers.
Examples of ratio data include: weight (in kilograms or pounds), height (in feet or inches), distance, age (in years), income (in absolute figures, e.g. in US dollars) etc.
Summary of the levels of measurement
The table below shows the main characteristics of the four levels of measurement (Y=Yes; N=No).
| Characteristic | Nominal | Ordinal | Interval | Ratio |
|---|---|---|---|---|
| Numbers are used to differentiate objects. | Y | Y | Y | Y |
| Numbers have meaningful order. | N | Y | Y | Y |
| There is equal interval between adjacent numbers. | N | N | Y | Y |
| Has a true zero. | N | N | N | Y |
| Can be sub-divided to get fractions/ratios. | N | N | N | Y |
In conclusion, it is important to understand the various types of data and their levels of measurement when doing research. Data types and levels of measurement determine how a questionnaire will be coded and how the data will be analysed and interpreted.
Related posts
A Comprehensive Guide for Designing and Administering a Questionnaire