Yes, Wassim, it's called scale reliability analysis. You should not simply add them together and hope for the best. Please see my study guides.

Is the context that of stepwise regression? If so, the current model is to be compared with the new more inclusive model by Akaike information criterion AIC or Bayesian information criterion BIC. If the new model has a smaller criterion the inclusion of the new variable is justified. Otherwise it isn't.

Hi Wassim

>is there a statistical check to do in order to use this variable as moderator in a multiple regression analysis?

I think there is some statistical work to be done irrespective of how you intend to use the variable, be it a moderator, mediator, Iv, DV etc. Most importantly you will need to assure yourself that these three binary items are all measuring the same underlying construct. This is usually done by using a technique such as factor analysis, but things are not straightforward in your situation. First, with three items in a factor analysis you will not be able to assess how well your model fits the data (there's a RG thread by Jafar Batiha on this issue). You may have to add other variables to the model to 'borrow a few degrees of freedom'.

Second, the binary nature of the responses make things a little more complicated. You'll need to decide if the responses are truly categorical (e.g. question answered correctly or incorrectly) or just a crude indicator of an underlying continuous variable (e.g. I like chatting with people, yes/no - the yes/no could be considered a very crude measure of an underlying continuum of extroversion).

Once you have made your decision then you specify the factor analysis model with categorical variables (this will produce an IRT-type model) or as crude indicator of an underlying continuous variable (sometimes this is called an ordinal factor analysis).

Good luck

Mark

Thank you all.

You mean that a factorial analysis will be done (KMO, % of the variance) and Cronbach's Alpha value be calculated ?

What's about other statistical checks for a created variable from the three dummy variables?

Thank you for the help. I need more clarification.