I am investigating the effect of childhood trauma on mental well-being, through the mediating effect of resilience, and I want to control for age, ethnicity, gender, mental health disorders and trauma types. However the sum of trauma types represents the independent variable of childhood trauma. So, is it possible to control for trauma types?
In terms of the other variables:
- Mental health correlates with childhood trauma
- Being black and of mixed ethnictiy correlates with childhood trauma.
So, is it possible to control for these variables as well? given that they are significantly correlated with the variables included in the simple mediation model?
Thank you in advance.
I think it is possible to control for the variables you have mentioned, but controlling for trauma type may be difficult. If I understand correctly, your participants have indicated one or more types of trauma they have experienced, and you have summed them to create the IV. Is this right? In this case, you would need to create a dummy variable for each type of trauma and enter them all as covariates. This may cause a problem with power, depending on the number of trauma types and size of your sample.
If I understand your model and questions correctly, it sounds like you could control for mental health and ethnicity, assuming the correlation is not extremely high, and as long as there is no reason to assume that any of your covariates might actually be a mediator. If you control for those variables and there is a very high correlation, you may run into issues of multicollinearity.
Regarding the trauma types, here is another option: Let's say you are using three different trauma types (physical, sexual, psychological). You could run three separate models within the PROCESS macro in SPSS, entering one trauma type as the independent variable for each model, and including the other two types as covariates. For example, you could run one model where physical trauma is the IV, and enter sexual and psychological trauma as covariates. The next model would use sexual trauma as the IV, and use psychological and physical trauma as covariates, and so on. Doing this in PROCESS will give you the statistical output for the effects of each type of trauma on its own while controlling for the other trauma types. Conceptually, however, you have really just run one total regression model with three predictor variables. This would force you to change your IV to "type of trauma" rather than total trauma, but it would still hopefully answer your question of how trauma affects mental well-being via its effect on resilience. See also: Hayes, A. (2017). Introduction to Mediation, Moderation, and and Conditional Process Analysis: A Regression-Based Approach from Guilford Press for more information.
Hope this was helpful!
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude