Jeremy, thank you.
It took my brain a while to work it through but now I've got it this
does appear to be the third (most straightforward) way I was looking
for.
General question for all... I find Julie Pallant's SPSS survival guide
very useful for basic 'how to' guide to SPSS. Can anyone recommend a
book that covers the same material but offers more detail when things go
wrong or are not straightforward?
Thanks again,
Caroline Wilson, PhD student, Institute of Energy and Sustainable
Development, De Montfort University
-----Original Message-----
From: Jeremy Miles [mailto:[log in to unmask]]
Sent: 03 February 2009 14:36
To: Caroline Wilson
Cc: [log in to unmask]
Subject: Re: data management query
Hi Caroline
My advice is to always enter the data as it appeared on the
questionnaire or survey, and give it the number that was used on the
questionnaire. Anything else leads to potential confusion.
Then recode the variables, and give them a new name. All variables
that go together to make one scale are recoded together.
So you might start with variables
Q2a
Q2b
Q2c
Q2d
If these were four point LIkert scales, measuring depression, to
recode them you can use:
For non-reversed items:
compute dep1 = Q2a.
For reversed items:
compute dep2 = abs(q2b-5).
And if it's reversed in one group:
if (group=1) dep3 = abs(q2c-5).
if (group~=1) dep3 = q2c.
You can then sum them, average them, calculate alpha, etc.
Always use syntax, that way you can record what you have, save it, and
correct it when you make a mistake.
Jeremy
2009/2/3 Caroline Wilson <[log in to unmask]>:
> Hi,
>
> Advice would be appreciated from anyone with experience of reversing
the
> codes of variables.
>
> I've got six groups of subjects to work with.
> Adults in five groups were asked a series of questions which contained
some
> negative worded items. These are being entered into SPSS and the
reverse
> coding has yet to take place.
> Now I've got a set of children's data to add. For this group, the
questions
> were not negatively worded, in order to make the survey as simple as
> possible to understand.
>
> Should I
> 1. enter the children data and then reverse code them so they match up
using
> the 'if' key?
> 2. enter the children data in a separate data file, recode them so
they
> match up, then merge the data files when the answers are in a
consistent
> order?
> 3. go for an even easier option that hasn't occurred to me but which
you may
> have come across?
>
> Many thanks
>
> Caroline Wilson
> PhD Research Student
> Institute of Energy and Sustainable Development
> De Montfort University
> Leicester
>
>
|