# Suppression effect?

#### Xyz123

##### New Member
Let's say you are conducting a multiple regression analysis examining the impact of X1 on Y. Let's also say you do not find a relationship until you control for X2. After controlling for X2, X1 predicts Y. Am I correct in assuming that this is an example of a suppression effect? If not, what is the name for this effect? What exactly has happened here?

#### hlsmith

##### Not a robit
Depends. First you should draw out the relationship between the variables that will help you understand the source of your phenomenon (e.g., confounding, controlling for a mutual effect, etc.). So draw and post you graph and we can try to help.

#### Xyz123

##### New Member
It's more of a hypothetical scenario, but to give more details:

the goal of the regression analysis is to examine the impact of some measure of cognitive ability on performance in school. A relationship was not identified between these two variables until the student’s SES was controlled for. After controlling for SES, cognitive ability predicted performance.

#### Xyz123

##### New Member

I'm trying to understand what the diagram would look like (or a plausible scenario of how it would like) if the above scenario had happened. Like I said, the goal of this hypothetical regression analysis is to examine the impact of some measure of cognitive ability on performance in school. A relationship was NOT identified between these two variables UNTIL the student’s SES was controlled for. After controlling for SES, cognitive ability predicted performance. I am trying to understand what redundancy, spurious relationships, and suppression are and whether any of these scenarios would fit what I am describing.

#### Karabiner

##### TS Contributor
The bivariate correlations between the 3 variables are also relevant here.