4.2.3.2 - Levels of measurement
The same psychological construct can produce nominal categories, ordinal ranks or interval measurements depending on how it is operationalised. You will classify data by what comparisons the numbers genuinely permit, then explain how level of measurement affects summaries and statistical tests. Applying the distinctions will prevent treating labels as quantities or assuming that gaps between ranked positions are equal.
The measurement decision
Psychologists do not just collect "data" in a vague way. They operationalise a variable, then record it in a particular form. The level of measurement describes how much information that form of data contains.
Level of measurement
A level of measurement is the type of scale or classification used to record data. For AQA Psychology, the three required levels are nominal, ordinal and interval.
This matters because the same psychological idea can be measured in different ways. Suppose a researcher is studying anxiety before public speaking:
| Operationalisation of anxiety | Level of measurement | Why |
|---|---|---|
| Whether the participant says they felt anxious or not anxious | Nominal | The data are named categories with no order. |
| A self-rating from 1 = not anxious to 7 = extremely anxious | Ordinal | The scores can be ordered, but the gaps between ratings may not be equal. |
| Reaction time in milliseconds to threatening words after the speech | Interval | The units are equal, objective units of time. |
An AQA-style question may ask you to identify the level of measurement used in a study and explain your choice. The answer should be based on the actual data recorded, not just the topic being studied.
The quick decision route is:
| Ask yourself | If yes, the level is usually... |
|---|---|
| Are the data just categories or names? | Nominal |
| Can the data be ranked, but the gaps are not known to be equal? | Ordinal |
| Are the data numerical scores with equal units? | Interval |
One useful habit: describe what each participant actually contributes to the data set. A participant might contribute a category, a rank/rating, or an objective numerical score. That tells you the level.
The same records, different information
Suppose four participants complete a search task in 12, 18, 21 and 33 seconds. These original records use equal time units. If the researcher replaces them with ranks 1st, 2nd, 3rd and 4th, the records retain the order but lose the unequal time gaps: the gaps were 6, 3 and 12 seconds, although each rank step is one place. If the researcher instead records only “finished within 20 seconds: yes/no”, the resulting variable is categorical. Neither recoding creates new information, and the original seconds cannot be recovered from the categories alone.
Decide the level of the variable actually analysed. Choosing a statistical test may also require the aim and whether the scores are related or independent; the level alone does not choose the whole test.
Nominal data: named categories
Nominal measurements identify which category each case belongs to.
Nominal data
Nominal data are data placed into named categories with no inherent order or numerical value.
Nominal data are the simplest level of measurement. The researcher counts how many cases fall into each category. The categories may be labelled with words, letters or numbers, but those labels do not mean "more" or "less".
For example, these are nominal:
| Psychological context | Nominal data recorded |
|---|---|
| Social influence study | conformed / did not conform |
| Attachment observation | secure / insecure-avoidant / insecure-resistant |
| Clinical questionnaire | has a phobia diagnosis / does not have a phobia diagnosis |
| Treatment preference survey | CBT / drug therapy / no treatment |
The exam trap is that nominal data often appear beside numbers. If a results table says 18 participants conformed and 12 did not conform, the frequencies are numbers, but the underlying measurement is still nominal because the data are counts in categories.
You can sometimes use the mode with nominal data because it identifies the most frequent category. Percentages can also summarise category frequencies. However, it would not make sense to calculate a mean of "CBT, drug therapy and no treatment" because the categories are not points on a numerical scale.
Nominal data are useful when the variable genuinely consists of categories. They are less sensitive when a researcher compresses a richer behaviour into a yes/no outcome. For example, "aggressive/not aggressive" may lose information about how intense, frequent or varied the behaviour was.
Ordinal data: order without equal gaps
Ordinal measurements preserve order without establishing equal distances between values.
Ordinal data
Ordinal data are data that can be placed in rank order, but the intervals between points on the scale are not known to be equal.
Ordinal data tell us who or what is higher, lower, stronger, weaker, first, second, third, and so on. What they do not tell us is the exact size of the difference between each position.
Common psychological examples include:
Continue this lesson.
You’re previewing NovaLearn without an account. Create a free account to read the rest.
Create free account