HRGuru

Well, they can be pretty involved.  Some of them are long, intimidating and off-putting.  I know that you can run the risk of having good people decide not to bother to apply if they feel like they are having to go through too much, and some of these tests can be quite invasive and personal.  Service Merchandise had a horrible testing system, with hundreds of questions to go through.  Oddly enough, this testing did not have positive results.  You don't want people to start feeling a bit like a mouse in a maze, or as though you are picking their brains looking for where they're messed up.  These tests are usually administered in house, with the interviewee alone in a room with the computer, which is not the best feeling in the world.  I must say that even though I support the theory of testing, I advocate making it a much more pleasant and productive experience.  What I would suggest is having a few questions chosen from the tests that most match what you're trying to accomplish, and incorporate them smoothly into the interview.  This is going to take some thought and practice, but the interview is as much for the new hire's information as for yours.  If the person you are interviewing feels interrogated or over-examined, they will get a poor impression of your company and management.  But if you make it light, interesting and even fun, it will promote the idea that you just want a great team that will work well together in a positive environment.  Which is exactly the point of the tests.  I would also suggest not having more questions than it would take an hour at most to answer comfortably, as the interviewee is not likely to accept an offer from you if he found the interview exhausting or invasive.  The idea is to get them to open up, and you'll be more likely to do that if you're in person, friendlier and more open than you will if you have the appearance and demeanor of someone who is going to look them over with a microscope.  Put them at their ease as much as you possibly can from the minute they walk in.  If you're talking about just letting them access the tests over the internet, I would say that the drawback to that is that it makes the test far less effective.  One of the reasons that Service Merchandise failed is because they just left these people with a computer, as earlier stated.  People who have been through enough of these tests--and they were quite popular in the 80s--get used to answering a certain way.  Most people will think about it and tick off the answer they think you want to hear.  But if you're face to face with them, you can gauge their immediate reaction, which is the most honest one.  You can watch their facial expressions and body language.  To be honest, there's nothing like the personal touch. 

We utilized an outside source for pre employment assessments for sales candidates.  There was only one time the candidate did not want to do the testing which in my eyes puts up a red flag anyway...

Pre-employment personality measures based on the Five Factor Personality Theory (Big 5 for short) have been shown to be the most valid and reliable of all personality measures. One example, albiet an expensive one, is the NEO PI-R. This is one of those massive personality inventories everyone is speaking of here. It is long, and quite agonizing to finish. (I remember having to do this one in a class and thinking I'd rather hear a lecture.) 

However, a better alternative would be to use one of Goldberg's inventories. They are short, valid, and reliable. The best thing about Goldberg's inventories are that they are Open source, meaning as long as you reference where you got it the test is free! You can go to The international personality inventory pool to read all about the Big 5 and the free assessments.

A little background on the big 5 for everyone. This measure is based on 5 dimensions: Openness to new experience, Concientiousness, Extraversion, Aggreeableness, and Neuroticism. By measuring these areas one could make predictions about job performance (only if you have done a thorough VALID job analysis and have made your predictor and criterion from such an analysis. This will keep you legal.)

For example someone low in extraversion would probably not make the ideal salesman, likewise someone high in extraversion would not need to be an accountant sitting in an office in front of a computer all day.

Cognitive pre-employment tests have shown to be some of the strongest predictors of job performance. The Wonderlic is an example of a cognitive ability test. These tests usually test one's spatial ability, math, form perception, and other cognitive functions. For example you'd want a car mechanic, or assembly line technician to be high in spatial ability.

However, the final hiring decision should not be based soley on findings from a personality measure, but several factors based on the job description.

There are generic job analyses out there for free. The USDOL funds a site called O*Net where all they do is poll subject matter experts and compile KSAO lists on any occupation out there. (This replaces the DOT, although it is still available as well.) Just go click on O*net and you can play around with it.

Further reading on the Big 5 can be found anywhere (Google) If your looking for specific research I'm sure I can dig through my research and find you some articles.

dmiller, I agree with you.  You are right on target!