Interaction effect

[floriisco]

Hey guys,

I would like to investigate effect of a treatment in randomized control trial. Also, I would like to investigate the moderating effect of motivation (of the participants who are included in the treatment group).

I wanted to use a multiple linear regression with an interaction term in this way:

y= alpha + beta_1 * treatment dummy + beta_2* motivation + beta_3* treatment dummy * motivation

Then, I realized that motivation will not be measured for the participants who are only in the control group.

Would it be possible to use the following model instead?

y= alpha + beta_1 * treatment dummy + beta_2* treatment dummy * motivation (setting motivation zero for the control group)

Cheers,
floriisco

 

[Dason]

How many treatments are there?

 

[floriisco]

Only one treatment. The treatment dummy variable would take value 1, if the participant is in the treatment group and 0 otherwise

 

[Dason]

If you're setting motivation to be 0 for the control group anyways then adding an interaction term in doesn't change the model at all. You don't need the interaction term. The two models you posted would be equivalent in this case.

 

[floriisco]

What I would like to find out is 1) the effect of the treatment itself and 2) whether motivation affects the magnitude of effect 1)? How can I do this?

Like this: y= alpha + beta_1 * treatment dummy + beta_2* motivation?