Effect of Suppressor Variable.
Before we look at suppressor variable and its effect, let’s briefly know the meaning of a variable. A variable is an entity that changes or we can say variables are measurable characteristics that changes or varies either from group to group, person to person over time. There are different types of variables but our focus is on suppressor variable.
Horst was the first to bring the concept of a suppressor variable into the psychological literature in 1941.
In multiple regression, Horst described examples in which a variable that was uncorrelated with the criterion but substantially correlated with another predictor enhanced the overall predictive power of a model.
Horst explained this perplexing phenomena by claiming that the suppressor (the uncorrelated variable with the criterion) adjusts for criterion-irrelevant variance in the other predictor, improving its predictive value.
An independent variable that has no components in common with the dependent variable but does have irrelevant elements in common with the predictor is referred to as a suppressor variable.
A suppressor variable, according to Horst (1941), is a predictor that has no connection with the dependent variable but, paradoxically, contributes to the test battery’s predictive validity.
Effect Of Suppressor Variable On Regression Model
The original link between the predictor and the criterion variable is changed when a suppressor variable is added to a regression model.
The suppressor variable can make the association stronger, weaker, or have no effect, but it can also change the relationship’s direction, switching it from positive to negative.
If the predictor and suppressor have a positive correlation, the suppressor will have a negative regression weight after being included in the regression equation.
Practical Illustration Of Suppressor Variable And Its Effect
In order to truly understand the concept of this suppressor variable and its effect, i will be using a real life example involving a married man(X1) and his wife(Y1) taking a walk together and a passer-by girl with big ass(X2). Take a look this picture below.
X2 is the suppressor variable
X1 is the predicator
Y1 is the dependent variable.
Notice that X1 and Y1 are the married couple and X2 (suppressor) is the passer-by girl that has no relationship with the Y1 (dependent variable) but has an effect on X1(predicator). Now this effect caused by this X2 (suppressor variable) can either strengthen, weaken, and even change the nature of the relationship between X1 and Y1.
Mainly looking at the picture above, it is obvious that X1 is already in a big trouble from Y1. So that is what suppressor variable does when added to a regression model, it can either impact it negatively, positively or no impact at all.
Application Of Suppressor Effect
The suppressor effect is the result of a suppressor variable, and one of the most common applications of this effect is the establishment of validity scales to adjust for response biases including social desirability and acquiescence.
Suppressor effects are operating when the addition of a predictor increases the predictive power of another variable.
Suppressor effects can help explain the construct validity of symptom measures by bringing to light opposing features that are present but mainly concealed in the total score of the measure.
How To Deal With Suppressors In Regression
- Combine predicator variables that are highly correlated
- Run a factor analysis on the predictors and use their factor scores as predicting variables.
- Find the suppressors and remove them from the analyses
How To Deal With Factor Suppressors In Structural Equation Modeling
- Re-examine your factor structure to make sure it has a good fit.
- Compare your factor structure with Haman’s single factor
- Combine factors that are highly correlated with each other
- Find the factor suppressors and remove them from the model
- Examine the factor suppressors in isolation in their own model
Although suppressor variables can be surprisingly and seemingly problematic, researchers should not ignore them. Understanding what suppressors are is a valuable tool for the researcher because it points out important information about the relationship between multiple variables, and they offer potential causes for problematic regression models