You are here:
An important part of a well-designed analysis is to be aware of the types of data that are available, so that the appropriate analytic techniques are employed, and inappropriate ones avoided.
Nominal data are data that cannot meaningfully be placed in rank order – they are typically labels (names) for categories. Some examples are: eye colour, biological species, students’ ethnicity or gender, educational course of study. There is no underlying progression in the data – no order of merit or size.
Ordinal data are data that can be placed in rank order, but for which the magnitudes of distances on the underlying scale cannot meaningfully be compared.
Some examples are: rating scales (Likert etc.), some educational test scores
A typical, and false, assumption made by researchers, educationalists and lay people alike, is that it is meaningful to say that the difference between a test score of, say, 25 out of 50 and one of 30 out of 50, is equivalent to a difference between 40 out of 50 and 45 out of 50. While, in both instances, the difference is five marks, it is not usually the case that the difference in achievement level is directly reflected in the difference in marks. This is because it is very rare that all marks in a test are equally easy, or difficult, to obtain. For example, a teacher will construct an informal end of topic test containing 20 items. Because we cannot say that each item is of equal difficulty, the marks cannot be meaningfully compared, one against the other. This data is ordinal data.
Test scores can be used to place students in a rank order, albeit with some uncertainty resulting from error of measurement. However, they usually cannot meaningfully be compared in the way described above.
Interval data are data that can not only be placed in rank order, but can also be compared in terms of magnitudes of difference.
Some examples are: time, weight, some educational test scores
In other words, a difference on a scale corresponds to the same change regardless of its location on the scale. For example, weight is an interval-scale measurement; the difference between 80kg and 90kg is the same as the difference between 90kg and 100kg. This is because the change in the property underlying the measurement (mass) is the same in each case.
Educational test scores which are weighted and scaled for difficulty fall into this category. Examples are e-asTTle and PAT data.
It is important that educators understand the difference between ordinal and interval data, so that they do not make false assumptions when reading data.
Return to top