I've often seen people post their Myers-Briggs test results, as if it would somehow explain who they are to the rest of the world. Salon has an interview with the author of a book that it's BS, that many personality the test don't match up to reality.
[Q: ] Or when a test indicates someone would be a successful employee, you'd check to see whether that person turned out to be successful or not?
[A: ] Yes, you'd want to compare what the test said to real-world results.
The other criterion, reliability, is when a test delivers consistent results when given to the same person on a repeated basis. There again, a lot of personality tests are very unreliable, in part because people are complicated and dynamic and do change over time and according to situation. Proponents of the Myers-Briggs will tell you that your personality type, as represented by the four dimensions, is inborn and unchanging. But in fact, research shows that when given more than once to the same person, as many as 75 percent of test takers will get a different personality type on the second administration.