In a first-order multiple linear regression model with
two predictor
variables, X1 and X2,
an interaction term may be used when it is believed that
effect of the
predictor variable X1 on the
mean of Y is influenced, or depends upon, the
value of X2.
the sum of the predictor variables S = X1 +
X2 should be used in a simple linear regression
model instead of using X1 and
X2 individually
relationship between the dependent variable and the independent
variables is linear
effect of the predictor variable X1 is
uncorrelated with the predictor variable X2
effect of the
predictor variable X1 on the
mean of Y is influenced, or depends
upon, the value of the error term