Is an offset appropriate for this application?

Here is what I am trying to do. My outcome variable is "the number of
call attempts made to a household until an interview is completed." I
am attempting to model this outcome using a poisson or negative
binomial model. I have a covariate that is "number of days that at
least one call attempt was made". I am wondering whether I should use
this variable as an offset (and change the interpretation into a rate)
or leave it as a covariate.

It doesn't really matter to me what I do, but I would like to be
methodologically correct in my decision and be able to defend it,
because I will be presenting my results to an academic crowd. Is there
a test or method that I can use to help decide whether or not to use an
offset?

Thanks.
Jake

.

Relevant Pages

  • Re: Is an offset appropriate for this application?
    ... call attempts made to a household until an interview is completed." ... I have a covariate that is "number of days that at ... this variable as an offset ... methodologically correct in my decision and be able to defend it, ...
    (sci.stat.consult)
  • Re: Is an offset appropriate for this application?
    ... this variable as an offset ... methodologically correct in my decision and be able to defend it, ... be used when a predicted rate is being compared to an external standard ( ... fixed numbers of person years at risk (where offset = log(person-year at ...
    (sci.stat.consult)