Levels of Measurement

Nominal data

Data that can be put into categories/names but theres no order between them

• Categories don’t overlap

• You can’t rank them

Eg country of birth (UK, France, Japan), pet type (dog, cat, hamster) etc

*Nominal = name only

Ordinal data

Data that can be put into categories order but the gaps between the values aren’t equal

• You can rank them

• But you can’t say by how much one is more than the other

Eg places in a race (1st, 2nd, 3rd), happiness rating (from 1 to 7), shirt sizes (small, medium, large)

*Ordinal = order matters but not the exact difference

Interval data

Data with equal gaps between values. You can measure the exact difference

• No true 0 (0 doesn’t mean ‘nothing’)

• Continuous numbers (decimals allowed)

Eg temperature in celsius (20 degrees is 10 degrees more than 10 degrees), time in seconds

Can you covert between types of data?

It is possible to convert from a higher level of measurement to a lower level of measurement (but not the other way around). Interval can be converted into ordinal, and ordinal can be converted into nominal

Converting interval to ordinal:

• Start with participant interval scores, eg biological measures like galvanic skin response, reaction times or psychometric scores (eg IQ) on a standardised test

• Each participant is assigned a rank score to turn the interval measure into an ordinal measure. This is done by listing each participant from the highest scoring to lowest scoring (using the interval measurement to place each participant). Any participants with the same interval score share the same rank position

Converting ordinal to nominal:

To convert ordinal data to nominal data, separate categories are created. Eg fast reaction/slow reaction, intelligent/unintelligent, extravert/introvert, depressed/happy, aggressive/passive. The highest ranked half of the participants are assigned to one category, and the other half to the other category