If the road is not yet clear to you,

we will show you the way how to get through.

A dichotomy is a nominal variable with only two values.

A dichotomy is a special type of nominal variable

Nominal variables are being used for making groups. If there are only two groups to be distinguished the nominal variable is a dichotomy. For instance the variable gender has two levels: male and female. No other values are allowed. Be aware this is the operationalisation the scientist uses. Another way to operationalize gender is by asking whether people feel like they are male or female or something in between. Such a question can be used too and a standard list of answers is on a five point Likert scale: totally male – more male than female – male nor female – more female than male – totally male. The way gender is operationalized now, is not a dichotomy because it has five values.

Coding a dichotomy

The values in a dichotomy can have all kinds of codes. Normally 1 and 2 are used, but it does not really matter if you use 1 and 5, or 23 and 67. Even male and female would be okay but it will give problems when using these codes in almost every statistical program. Best of all, use numbers.

Dichotomy and statistical tests

If there are more than 2 codes, you should use the ANOVA. But if you want to compare group 1 with group 3 a t-test should be used because all other groups will be ignored.

Nominal variables cannot be used in regression analysis but a dichotomy is no problem. Be careful when interpreting the result of a dichotomy in regression, because the regression coefficient will be the opposite if the codes are changed from 1 and 2 into 2 and 1. Also the changes from 1 and 2 into 1 and 5 show a different regression coefficient.

When the code 0 and 1 are used the variable is called a dummy. It is possible to transform a nominal variable with more than 2 two levels in a list of dummy’s. Read all about dummy’s on our site page called Dummy-variable.

Related topics to dichotomy:

• Nominal data
• Dummy variable
• ANOVA
• t-test
• Regression

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

Return to the index of this dictionary