Levels of Measurement

When it comes to data and categorising it but unfortunately, it is a little bit more complicated than simply qualitative and quantitative data. Another way data can be categorised is by its levels of measurement. There are a total of four, and we'll try to break each one down so that you can not want to run away every time you see data.

• We will start by defining the measurement levels in research.
• Next, we'll delve into each level of measurement in statistics by looking at its definition and an example. For instance, we will define the nominal level of measurement and provide a nominal level of measurement example.
• Finally, we will look at how levels of measurement in psychology can be identified and what their uses are.

Level of Measurement

In data, there are four levels of measurement nominal, ordinal, interval and ratio.

Levels of Measurement in Research

When psychologists conduct their research, understanding the measurement variables in statistics is one of the most critical steps.

The four levels of measurement are scales used to measure variables in research.

Regarding data analysis, certain conditions must be met when conducting statistical tests. For instance, the dependent variables data should be ratio or interval if aiming to conduct a parametric test.

Define: Nominal Level of Measurement

The nominal level of measurement in psychology consists of 'named' or 'labelled data'. It is identified as a level of measurement that collects categorical data.

Nominal data is characterised by the following:

• No order between values– one answer in a questionnaire is as vital as the others, and this is because these data tend not to provide numerical value.

• Nominal values do not overlap– respondents can select only one answer (data that can take only specific values are called discrete data).

• They are not usually used for evaluation calculations but rather for grouping data or participants;

  • The standard calculations used to represent nominal data are percentages and mode.

Nominal Level of Measurement: Example

Typically questions in questionnaires that have a fixed response that doesn't involve you rating something generate a nominal level of measurement.

If we break down this example response, it can be identified that the data is split into categories (i.e. each sex). However, the data ranking is unimportant, meaning we can't determine if being born male or female is more important than the other.

Define: Ordinal Level of Measurement

Similar to the nominal level of measurement, ordinal data is identified as categorical. However, the ranking of the data is vital.

Ordinal data have the following characteristics:

• There is no way to measure the numerical value of one response to the next, e.g. researchers cannot determine how much the respondents who answered 3 differ in importance from respondents who answered 5.

• Data based on ranking – there is a difference between the ratings based on the order, but we cannot measure how big the difference is.

• The order of the data is essential, e.g., 1 may reflect a weaker response than 5.

• Likert scale responses are an example of ordinal data.

Ordinal data is usually qualitative because we cannot determine the numerical significance between values. It is typically used for data reflected in categories, i.e., ordinal data has limited use for quantitative data.

Ordinal Level of Measurement: Example

Let's see an example of ordinal data and how we can identify the response as ordinal.

In this example, although the order of the data collected is necessary, the differences between the values are not, making it an ordinal level of measurement example.

Define: Ratio Level of Measurement

We identified nominal and ordinal data as categorical data, but ratio data is categorised as the opposite of this as it collects continuous data, meaning it can have an infinite value,

Ratio data is characterised by the following:

• There is an absolute zero, i.e., the data collected cannot be 0 or less than 0.

• The measured data is continuous (data that can have any value).

• The distance between the values is the same, e.g. the distance between 3 and 5 and 7 and 9 is the same.

• Researchers can measure the difference between the values. E.g., the researcher can identify and quantitatively measure the difference between participants who responded to 1 and 50.

• The direction of change in numerical values is essential, e.g., 14 to 24 indicates an increase, while 30 to 17 indicates a decrease.

Ratio Level of Measurement: Example

Examples of data where ratio measurement is used are participants' height, age and speed. None of the examples listed can have a value of less than 0, and the data is continuous because the values reported can have an infinite number of values.

Let's break down a research example to highlight how the ratio level of measurement in psychology may be collected.

Height is clearly a ratio level of measurement example. The difference between height scores is quantifiable, e.g. someone with a height of 5ft is 1 foot shorter than someone who is 6ft tall, and you can't be measured at a value of 0 or lower.

Now, age can be a tricky one. But think about it we're never really 0 years old; we may be 0 and 1 second years old or older. So age does have an absolute value of 0, and the difference between ages is equally important. For instance, if you are six years old, you will always be identified as younger than someone over six years old.

Interval Data Pyschology

Interval data is a fixed unit, and the distance between the adjacent numbers is equal.

Similar to ratio data, interval data collect continuous data.

Interval Level of Measurement

Interval data are characterised by the following:

• The value 0 is not absolute; the collected data can be 0 or less.

• The measured data is continuous (the data can have any value).

• The interval between the values is equal, e.g. the intervals between 3 and 5 and 7 and 9 are identical.

• Researchers can measure the difference between the values; e.g., the researcher can identify and quantitatively measure the difference between participants who responded to 1 and 50.

• The direction of change in numerical values is essential, e.g., 14 to 24 indicates an increase, while 30 to 17 indicates a decrease.

Interval Level of Measurement: Example

An example of collected data that can be classified as interval data measurement is temperature since the temperature can be 0 or below.

Let's look at an interval level of measurement example in psychological research.

IQ scores are clearly a ratio level of measurement example. The difference between IQ scores is quantifiable, e.g. someone with an IQ score of 45 has a score 2x lower than someone who has a score of 90. Although it's heard of, you can get a score of 0, meaning this test score does not have an absolute 0 value.

Similarly, we can quantitively identify the difference between temperatures, and you can measure a temperature of 0 and below.

Levels of Measurement: Psychology

When conducting research, it is crucial to determine the data's level of measurement because this helps us understand how to interpret the data, what statistical test should be used, and what information the data can give us.

Look at the table below to see how we identify the type of data to use.

Levels of Measurements in Statistics

From identifying the level of measurement, researchers can determine how data was collected, e.g. were the methods qualitative or quantitative, how the data can be classified and what type of statistical tests can be used.

Not only does the level of measurement in statistics influence the type of test that should be carried out it also influences the inferences.

Typically, researchers can make generalisable inferences from ratio and interval data as these allow researchers to use parametric tests. The same cannot be said about nominal and ordinal data.