Research Methods / Investigations Guidance

Levels of Measurement

The levels of measurement below are in order from weakest to strongest. When trying to establish the level of measurement used, start with Nominal data and work your way through.
For example, if you are measuring continuously, then you can assume that your data is 'at least ordinal' - this is the basic assumption for the Mann-Whitney U Test, Wilcoxon's T Test and Spearman's Rho.
You can go further than this and establish whether it is ordinal, interval or ratio level by looking at the scale you are using (time in seconds, temperature in degrees centigrade, number of words found, etc.) and how these scales work.

Nominal

Data is measured discretely in named categories.
For example, participants' responses are measured using a descriptive term rather than a numerical score (true / false, better / worse).
Alternatively, participants' responses are grouped together into a category (number of participants who report seeing broken glass).

Ordinal

Data is measured continuously.
For example, the time it takes participants to finish a wordsearch can be measured continuously in seconds.
Scores can also be put in rank order (i.e. it would make sense to say that participant one came first, participant four came second, and so on).

Interval

Data is measured continuously.
For example, the ambient temperature at which participants feel their concentration is affected can be measured continuously in degrees centigrade.
Scores can be compared using intervals (i.e. it would make sense to say that participant one was affected at a temperature 10 degrees higher than participant four, but 10 degrees is not twice as hot as 0 degrees, so it wouldn't make sense to say that participant one was able to withstand a temperature twice as high as participant four).

Ratio

Data is measured continuously.
For example, the number of words found could be measured continously by counting them.
Scores can be compared in ratios (i.e. it would make sense to say that participant one found twice as many words as participant four).