What you'll learn

• How to identify nominal, ordinal and interval data.
• Why the level of measurement affects which statistical test you can use.
• How to apply levels of measurement to AQA-style research scenarios.
• Common traps, especially when numbers look more “scientific” than they really are.

Why levels of measurement matter

In psychology, researchers collect data, meaning recorded information from participants, observations, questionnaires or experiments. A variable is anything that can vary, such as anxiety score, attachment type, reaction time or whether a participant obeys an instruction.

The level of measurement tells you what kind of information your data contains. This matters because it affects:

• what summary measure is appropriate, such as the mode, median or mean;
• what kind of conclusion you can draw;
• which inferential test you should choose. An inferential test is a statistical test used to decide whether a pattern in data is likely to be due to chance.

A useful way to remember the topic is as a ladder: each level gives you more information than the one below it.

AO1: the three levels of measurement

1. Nominal data: named categories

Nominal data consists of categories with no natural order. The word “nominal” comes from “name”: you are naming groups, not ranking them.

Examples in psychology include:

• male/female/other gender categories, if used in a study;
• diagnosis type, such as phobia, depression or OCD;
• yes/no responses;
• whether a behaviour occurred or did not occur.

Nominal data is usually counted as frequencies, meaning how many people or observations fall into each category.

For nominal data, the most suitable descriptive measure is usually the mode, which is the most common category. You cannot calculate a meaningful mean for categories such as “yes” and “no”.

2. Ordinal data: ordered data

Ordinal data can be placed in order, but the gaps between points are not necessarily equal.

A common example is a Likert scale, which is a rating scale where participants choose a numbered response, such as 1 = strongly disagree to 5 = strongly agree. The numbers show order, but the psychological gap between 1 and 2 may not be the same as between 4 and 5.

Examples include:

• ranking participants from most anxious to least anxious;
• rating stress from 1 to 10;
• finishing positions in a task;
• questionnaire responses such as “never”, “sometimes”, “often” and “always”.

For ordinal data, the median is often appropriate. The median is the middle value when scores are placed in order.

3. Interval data: equal intervals

Interval data uses a numerical scale where the gaps between points are equal. This means the difference between 10 and 12 is the same size as the difference between 18 and 20 on that scale.

Examples in psychology may include:

• reaction time in a cognitive task;
• score on a standardised memory test;
• total score on a well-designed psychological scale, if the stem treats it as a numerical score;
• IQ score, often treated as interval at A-Level.

For interval data, researchers can usually calculate the mean, which is the arithmetic average. They may also calculate measures of spread such as the range or standard deviation.

AQA Psychology does not require you to use ratio data as a separate category. Some real-world measures, such as reaction time, have a true zero, but for AQA test-choice purposes they are normally handled as interval-level quantitative data.

Matching level, aim and design to a test

Level of measurement is important, but it is not the only decision. You also need to consider the aim and design.

The aim is what the researcher is testing: a difference or a relationship. The design is how the participants are arranged. In an independent groups design, different participants are used in each condition. In a repeated measures design, the same participants are used in all conditions. In a matched pairs design, different participants are paired on relevant characteristics.

AO2: applying levels to scenarios

In exam scenarios, focus on the actual data recorded, not just the topic of the study.

If a researcher asks, “Did the participant conform: yes or no?”, the data is nominal. If the researcher asks participants to rate how much pressure they felt from 1 to 10, the data is ordinal. If the researcher records a standardised anxiety score from a validated scale and treats score differences as equal, the data may be interval.

A useful rule is: you can lose information, but you cannot magically gain it back. Interval scores can be converted into ranks, and ranks can be converted into categories, but each conversion loses detail. For example, exact memory scores can be ranked from highest to lowest, but once you only know the rank, you no longer know how large the score differences were.

AO3: evaluating measurement choices

A strength of understanding levels of measurement is that it improves validity, which means whether the study measures what it claims to measure. If a researcher uses a suitable level and test, their conclusion is more justified.

A second strength is transparency. Clear operationalisation — defining exactly how a variable is measured — makes research easier to replicate and evaluate.

However, psychological variables are often difficult to measure cleanly. A 1-to-10 “anxiety” rating may look precise, but one participant’s 8 may not mean the same as another participant’s 8. This reduces confidence that the scale has equal intervals.

Another limitation is that simplifying behaviour into categories can lose important detail. For example, coding responses as simply “aggressive” or “not aggressive” may ignore intensity, context and frequency.

There is also an ethical angle. When data includes sensitive categories such as diagnosis, trauma history or mental health status, researchers must protect confidentiality, meaning participants’ personal data should not be identifiable. Clear measurement should not come at the cost of participant dignity or protection from harm.