- Nominal data
- Nominal data are used in statistics to identify groups. These variables are used in for instatnce a t-test, ANOVA and as dummies in regression.
- Examples of Nominal data
- Dichotomy
- How can a nominal variable be analysed?
- Related topics to nominal data:
- Significantly improve your knowledge about research with this manual
- Mission

## Nominal data

**Nominal data are used in statistics to identify groups. These variables are used in for instatnce a t-test, ANOVA and as dummies in regression.**

**Examples of Nominal data**

Examples of a nominal variable are gender, eye colour, married, used medicine, place of birth, type of furniture, attended course and so on and so on. Don’t try to value the answers, because it is impossible to say if a kind of eye colour is better, worse or more highly valued than another eye colour.

It is okay to code the answers with numbers. Almost always this has to be done because statistical programmes are only working with numbers. So if you have an answer like ‘blue eyes’ then code this answer as a number. Grey eyes should be given a different number. And a combination of blue and grey gets another number. If a whole lot of answers are possible and you want to distinguish every single answer, you really need a whole lot of numbers. Imagine for instance how many codes you might need for all professions.

In surveys respondents usually can choose an answer from a list. This is called a multiple choice question. Normally you code the list of answers starting with number 1 for the first alternative and 2 for the second and so on until the end of the list is reached. In an online survey, the computer programme will assign these numbers automatically.

**Dichotomy**

A special type of nominal data is a dichotomy. A dichotomy is a nominal variable with only two values. For instance gender has only two values: male or female. Another example is: are you married? Yes / no. Notice that it is not allowed to give answers like divorced, widow or going to get married soon.

When a multiple response question is used in a survey, the respondent is asked to tickle aspects that are applicable to him. For instance:

Tickle the objects you have at home:

[ ] a television

[ ] a radio

[ ] internet

[ ] a washing machine

[ ] a boiler

This isn’t one question, but five. And all questions are a dichotomy. So you need fives columns in your database and the answers should be coded as 1 (= tickled) and 0 (= not tickled). Most programmes for online surveys code not tickled as blank. These blanks should be replaced with zeroes because a blank means missing and that is not the same as not tickled. I hope providers for surveys are willing to repair this fault in their programme.

**How can a nominal variable be analysed?**

Now what can be done with nominal data? Which types of analysis can be adjusted?

One thing that is always allowed is to count the answers. So it is okay to make frequency tables. And when frequencies are allowed, also the percentages can be computed.

Second, the mode can be provided.

And then?

Well, nominal variables are used in many other tests. They can be split into another nominal variable. Then a chi square test can be performed. It can also be used for splitting an ordinal variable and then a Mann-Whitney test or a Kruskal-Wallis test can be applied. When a nominal variable is used for splitting a continuous variable (that is an interval or ratio variable) a t-test or an ANOVA can be performed. To conclude, though a nominal variable in itself is not very interesting for a research, it has a lot of applications in combinations with other variables.

**Related topics to nominal data:**

All these tests are clearly explained in in our **SPSS-tutorials**.