Interpreting an interaction term

#1
Hello,
I am doing a linear regression with an interaction term which I am unsure how to interpret correctly. Lets say I have categorial variables: First variable has A and B AND second has x,y and z as categories. If I choose A and x as default/ommited as far as I know you interpret it as follows: b1x is the effect x has if A is true. If I want to know what effect the combination of x and B has I have to add: b1x + b2B + b3(x*B). This is a bit confusing as you have to add b1x in the second combination (x and B) even though the term is interpreted to be the effect of x when A is true.
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Post the output from the model and we can help. But all coefficients in in reference to your base case (e.g., intercept). Some programs will let you define effect estimates to get at the other terms.

SAS: LSMEANS
R: emeans
 
#3
so A3 is the reference point. If I want to calculate the output for scneario B4 : const + B + 4 +B4 ? This is confusing because as far as I know the effect of B is conditional on the A3 scenario however you still add it in the B4 scenario. Am I wrong about this?
 

Attachments