Validity

Validity is the extent to which a research instrument actually measures what it has been designed to measure. There are two kinds of validity that can be gauged statistically. These are concurrent validity and predictive validity.. Both can be determined using Pearson r. So, calculate Pearson r using SPSS.

Thank you for your answers. However maybe I was not clear what I needed - I was looking for ways of calculation these measures. Now I found the formulas e.g. here - http://www.ltrc.lsu.edu/TRB_82/TRB2003-001883.pdf - on page number 4. Of course I can calculate the indicators myself, but I wondered whether there is such predefined procedure in SPSS or SAS.

I use SPSS.  The only way I know of to get SPSS to do these calculations is to do the following:

1) Run the regression you want, and SAVE the predicted values (click the "Save" button and select whether you want the unstandardized or standardized values, or both).

2) For equation 1 on your ref., run your Pearson correlation between your observed and predicted values.

3) For equation 2, you can use Transform -> Calculate to get a new column for yhat-yobs, then a simple descriptives for the mean of this new column of data.

4) For equation 3, do the same transform but use the absolute value function in your calculation, then take the mean.

5) For equation 4, square the calculation from equation 2, take the mean of this column of (Yhat-Yobsv)squared, then divide that mean by n again

Validity is truly the extent to which a research instrument actually measures what it has been designed to measure as stated by Eddie. There are two kinds of validity that can be gauged statistically. These are concurrent validity and predictive validity.. Both can be determined using Pearson r. therefore, if you use Pearson r, it means you assumed that your data is normal. But if not, I think Spearman can be used.