In my intro Psychology text, it said that for what it was designed for, this test had a correlation coefficient of 0.33, which means that if you wanted to find the answer to a yes-no question, you’d get it right about 66% of the time. Of course, flipping a coin gets you the right answer 50% of the time. You wouldn’t want to say that it’s absolutely useless, but if you were trying to determine something, you’d probably use more than just this test.
I think I should point out that, IMO, there are a few things in Psychology that the average person believes would be absolutely useless, but that have been shown through scientific studies not to be.