www.seeingstatistics.com

Descriptive Statistics for Categorical or Nominal Variables

The mode--most likely category--and the proportion or percent in each category are the most useful descriptive statistics for categorical variables.

Example

In an undergraduate honors statistics class taught in the psychology department, there were ten females and five males.

Summary

In a class of 15 students, females outnumbered males two to one. That is, ten (67%) were female and five (33%) were male.


Computer Examples

R

Usually, a variable with nominal values would already be available in the dataset. However, for this example, we create the variable.

> gen <- c("m","f","f","m","f","f","m","f","f","f","m","f","f","m","f")
> #make sure variable is a factor; start here is variable already exists
> gen <- factor(gen)
> #examine counts of each factor level
> summary(gen)
 f  m 
10  5 

> #then use R as a calculator to compute proportions or percentages
> 10/(10 + 5)
[1] 0.6666667
> 5/(10 + 5)
[1] 0.3333333
> #or compute odds, in this case of female to male
> 10/5
[1]  2
> #a bar graph is overkill for two levels but may be useful for more
> plot(gen,xlab="Gender",ylab="Frequency",col=c("pink","lightblue"))
Bar chart of gender frequencies

StatView

Menu: Analyze > Descriptive Statistics > Frequency Distribution



nominal description from StatView


© 2002, Gary McClelland