1
$\begingroup$

Assume I have following hypotheses:

  • X is negatively related to Z
  • Y is positively related to Z
  • W is a moderator for the relationship between X and Z

And I have some control variables:

  • age
  • health
  • marital status

To test my hypotheses, I conduct a regression analysis. However, I do not know which approach is the correct one.

  1. I could do multiple regression as follows: The dependent variable is Y. For the independent variables I first add my control variables to the model (age, health, marital status) as independent variables, then I enter both the X and Z as independent variables.

  2. Another possibility to apply multiple regression is following: The dependent variable is again Y. I could also conduct 2 regression analyses. For the first regression I first add my control variables to the model (age, health, marital status) as independent variables and then I enter X as independent variable. For the second regression I first add my control variables to the model (age, health, marital status) as independent variables and then I enter Z as independent variable.

Obviously the obtained results are rather different in both approaches. In the first approach only X can be significant whilst in the second approach both X and Z can be significant.

Online, I find different opinions on the subject making the situation even more confusing. What is the most convenient approach to test these hypotheses?

Many thanks in advance for clarifying this issue to me. I generally don't know anymore what to do...

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.