Numerical Example of the Metric Effect


To help demonstrate the metric effect of a statistical interaction the previous example is utilized. The following represents the relationship of an interaction between sex and age on the response variable, salary. X1 represents age and X2 represents sex. The general model estimated is:
Y= 2 + 6 (X1 ) +1 (X2 ) +1 (X1 X2)

The metric effect of age (X1 ) on salary (Y) at values of sex (X2 ):

Sex Group 1 (Males): 6 + 1 (1)= 7
Sex Group 0 (Females): 6 + 1 (0)= 6

The metric effect of sex (X
2) on salary (Y) at values of age (X1):

Age 30: 1 + 1(30)= 31
Age 40: 1 + 1(40) = 41
Age 50: 1 + 1(50) = 51

The results provided above represent the estimate of the conditional effect of sex at specified values of age. It is possible to determine the relationship between sex and salary at values of age for both males and females. In this case, the independent variable of interest is a dummy variable. The metric effect results can be multiplied by values of the sex variable.
The metric effect of males on salary at the age of 30:

Age 30 and Sex Group 1 (males): (1 + 1(30))1 = 31

The metric effect of females on salary at the age of 30:

Age 30 and Sex Group 0 (females): (1 + 1(30))0 = 0

Substantively, 30 year old males average just over 3 thousand dollars more than do 30 year old females.


Back to Interpretation

Revised: August 17, 1999.