Hi Gemma,
the advantage of effect coding is you actual refer to the average of all the
categories instead of a specific reference category. With effect coding you
can give a value for every category within a comprehensive table, with
reference coding you can't. The advantage of reference coding is obviously
the calculation of odds ratios. In practice I always use effect coding for
nice output tables (so all the categories can get a specific value) and
reference coding for answering specific policy or research questions (with
odds ratios).
Hope this helps a bit.
Dr. Rik Van de Kerckhove
On 18/05/2009, Gemma Robson <[log in to unmask]> wrote:
>
> Hi All,
>
> I am trying to find a list of advantages/disadvantages for using Effect
> coding vs using Reference coding within Logistic Regression when
> using categorical predictor/explanatory variables.
>
> Any advice would be greatly appreciated.
>
>
> ________________________________________
> Gemma Robson
>
> Education Consultant ” SAS Customer Services
> Tel: +44 (0)161 888 2032 ” Mobile: 07795 442 774 ”
> [log in to unmask]
> SAS UK & Ireland ” T3 Quay Plaza ” Salford Quays ” Manchester M50 3BA
> www.sas.com/uk ” www.sasprofessionals.net ”
>
> SAS(r)... THE POWER TO KNOW
> Official data and competitive intelligence partner of the RBS 6 Nations
> Championship
>
>
> * Please consider the environment before printing this e-mail
>
> The information in this e-mail and any attached files is confidential. It
> is intended solely for the use of the addressee. Any unauthorised disclosure
> or use is prohibited. If you are not the intended recipient of the message,
> please notify the sender immediately and do not disclose the contents to any
> other person, use it for any purpose, or store or copy the information in
> any medium. The views of the author may not necessarily reflect those of the
> company.
>
--
Dr. Rik Van de Kerckhove
18 Academy Court
Kirkwall Place
London E2 0NQ
|