37

# In this example, SES has an indirect effect on academic achievement by way of the

effect it has on parental involvement (Allison 1999:60). Allison’s example goes on to

state that the “regression model only estimates the direct effect of each variable,

controlling for all the other variables in the model” (1999:60). This model is especially

useful in understanding the limitation in Jeynes’ operationalization of the relationship

between SES, parental involvement, and academic achievement by utilizing models with

and without SES. In reality, SES becomes the independent variable which acts as a

driver for parental involvement — parental involvement moderates the effect of SES on

academic achievement. It is important to remember that the non-SES model does not

prove that SES has no effect on academic achievement, but rather that the model is only

representative of the effect of parental involvement when SES is not used as a control

variable. In fact, it is quite likely that both SES and parental involvement influence

academic achievement, but because SES and parental involvement are related to each

other, it is difficult to ascertain the individual effects of either on academic achievement.

# According to Allison, another way to look at this relationship relates back to the diagram

above, and the sum of the indirect effect variable, SES, plus the direct effect variable,

parental involvement, is equal to the total effect of the independent and the control

variable upon the dependent variable, academic achievement (1999:61).

# Another, more extensive, explanation of the complicated relationship between the

three variables comes from the field of social psychological research and suggests that

the relationship could be clarified by making a distinction between mediator and

moderator variables and their effects on the dependent variable. According to Baron and