- see what is inserted, below.
> Date: Wed, 24 Apr 2013 10:47:23 -0700
> From:
[hidden email]> Subject: Not strictly SPSS- question on scores for independent variables
> To:
[hidden email]>
> Hello,
>
> Whilst this is not strictly a question regarding SPSS, I thought you guys
> might be able to help.
>
> I've sought advice from various forums and unfortunately, my quests have
> been far from fruitful!
Which was the case? - There wasn't any answer where you asked?
Or, there was some answer that did you can't summarize in sentence,
along with why it won't do? (Why mention that at all? And, by the way,
you have just insulted whoever did answer. Are you sure they don't read here?)
> I'm in the process of sorting out my last two variables for a regression and
> am having a little bit of trouble...
> They are customer score and expert score and utilise a 5 star rating system.
>
> Am I best turning them into percentages and then using the acrsine
> transformation? Or should I dummy code them? Or is there a more simple way
> to include them in my model!?
I don't think anyone has used arcsin to transform dichotomous data since
computers wiped out their single advantage (you don't have to compute the
variances by hand). I don't know why you would use them for this. (I assume
you meant percentiles of the grouped scores, like one would adapt Logits.)
The simple way to include them in the model, which is not obsolete, is to
use the scores (0 to 5) or (1 to 5) to predict.
Do you have a *set* of scores from each expert? or each customer? or both?
Just one expert?
>
> Any pointers/guidance would be really helpful! I've looked at papers and
> read other forum posts but can't seem to get a clear answer!
>
...
--
Rich Ulrich
Locked
