Hi John,
As far as I recall de Vaus (1991) suggests that you only need to use an
ordinary correlation coefficient to test for criterion validity. So the
only question is whether you use a Spearman's or Person's, based mainly on
whether the distributions are normal, and perhaps on whether you consider
your questionnaire to yield ordinal or interval data.
Hope this helps.
Regards,
John.
On Jul 16 2009, John Brennan wrote:
>Hi everyone,
>
> For my MSc dissertation I'm developing a new
>questionnaire. As part of the validation process I want to
>measure correlations between the subscales of the
>questionnaire I have developed and other scales that measure
>similar constructs. I was going to use multiple regression,
>which will generate a lot of data, and I was wondering how I
>should present the data (as in, what elements of the SPSS
>output I should include). Thanks a million for any advice
>received.
>
>John.
>
|