On 3 February 2010 03:25, Cherie Armour <[log in to unmask]> wrote:
> Hi All,
> I am working on a database but would like preliminary advice before moving
> forward. The database is representative of a European country. However, it
> contains two main groupings, the first is a random sample where interviews
> were conducted via the telephone. The second is a random sample of children
> who are in care (interviews conducted face to face. All were asked the same
> questions (there are no standardised measures used but some items come from
> standardised measures). So obviously the childcare cases have been
> oversampled. There is a weight variable included to adjust for the
> oversampling, however each group member has the same weight value so there
> are only two weight values.
> My questions are
> Should each case have a unique value in the weight variable/
No, some will have the same weight.
> If I want to conduct frequencies for my variables in SPSS how do I do this
> while including the weight values?
I forget. Is it under complex surveys?
> If I do not employ the weights do I have to treat the two groups as
> independent samples?
Yes.
> If I want to look at only the childcare cases and their associations do I
> just disregard the weight values and the cases in the other group?
> This is very complex (at least I am finding it so) but any help would be
> greatly appreciated. Oh and to complicate things further I have extensive
> documentation but it is in a foreign language. I have
> a colleague translating but we are struggling to find the relevant info for
> the above questions.
Depends on how representative you want to be, and if the weights make
them representative. You should be able to do (almost) everything you
can do without weights with weights, using the survey tools in SPSS.
J
--
Jeremy Miles
Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
|