All such tests (inductive verbal numerical, and personality trait) are pretty well worthless when applied to job selection (and probably everything else).

See how they get the data -- take 1000 programmers, give them a personality test, and discover that there is some correlation (small, but significant, usually) between a trait and some other, usually completely informal, assessment of the goodness of the programmers.  If you have a good number in your group, this significance may only account for 25% of the data, so it's pretty rough.  

But look at the logic -- can you say, if you have this trait, you'll be a good programmer?  No, all you know is that good programmers have more than usual of this trait.  So may many other people who are NOT good programmers.  So, you can't say, if a person has this trait, she will be a good programmer.  The conditional probabilities (Good old Rev. Bayes) just don't work like this.

I once did a personality test which had a scientific creativity subscale, and I scored in the 99th percentile.  Does this make me a Very Creative Scientist?  No, it just means that I share a characteristic with such people.  Doesn't make me a VCS, or even likely to become a VCS.

F99 is right.  It's phrenology because almost no one understands the conditional probabilities.