|
|
 |
| Sign of direct and indirect effects |
 
|
| Message/Author |
|
|
| Anonymous posted on Wednesday, April 22, 2020 - 6:50 am
|
|
|
Hello,
My question concerns the sign of the indirect effect. My model contains observed independent variables (delinquency), latent mediators, and a continuous (range: 0-4) normally distributed outcome variable (drug use).
I am exploring whether there are any indirect effects. I find on significant indirect effect, but I am concerned about the sign and how to interpret the result.
I have a negative direct effect from my IV to the mediator and a negative direct effect of the mediator to the DV. My indirect effect is positive and I am curious about how to interpret the result (i'm aware of MacKinnon's inconsistent mediation, but still learning about the theory and interpretation).
The results would suggest that the IV (delinquency) does operate through my mediator (school activities) to increase substance use even though the direct effects were negative. I'm just worried I'm misinterpreting the positive sign of the indirect effect. Any advice or reading recommendation is greatly appreciated. |
|
|
|
|
How about:
- when delinquency increases, school activities decrease (i.e. a negative slope)
- when school activities decrease, drug use increases (i.e. a negative slope) |
|
|
| Back to top |
|
|
