This is popular journalism, with some of the ticks of that craft — nearly every chapter begins with a you-are-there anecdote — but Murphy Paul is actually very good at explaining how the different personality tests she covers were devised, how they are supposed to work, how they spread, and the profound lack of good reason to believe that they do, in fact, work.
(The only point I would elaborate on for that is that she is still too soft on the "Big Five" personality factors test. That theory, if you can call it that, actually fails confirmatory factor analyses, i.e., the factors do not, actually, correctly predict the correlations among items on personality tests. This is acknowledged even by the founders and pushers of the Big Five, who cover by — rejecting the hypothesis test! [I am not making this up or exaggerating.])
One point which Murphy Paul raises repeatedly, and which I think is fairly shrewd, is that tests which seem to tell us what people are suited for are useful to organizations which have to sort people into different roles, and justify this sorting, when there is really very little information on which to do so. Even if the test is completely bogus, then, it may give legitimacy to the organization and its division of labor and of power. She would prefer that oranizations would actually be flexible enough to let people find out what they like doing and are good at. She doesn't consider, however, that this is surely going to be very costly in time, attention, and lost opportunities while people experiment. (You can make your own jokes about the exploration-exploitation trade-off.) Suppose a test is nearly useless, but not quite useless. It might be rational for those running the organization to use it, because it provides some information and some legitimacy cheaply, rather than either bluntly arbitrary decisions (cheaper, no legitimacy, no information) or full responsiveness. If people knew that the tests being used were mostly noise, however, I doubt that they would provide much legitimacy, so as usual this would involve a Big Lie.
Of course, all this is supposing that Murphy Paul is correct about why personality testing has been so widely embraced. If she is, I'd expect that other advanced industrial societies should either make comparable use of tests, or functional equivalents as people-sorters, or some reason why this problem doesn't arise for them. I don't know, and she doesn't address this --- this book is very US-centered.
The use of these invalid personality tests in custody disputes and other legal proceedings, however, is without even this marginal justification. It is simply a disgrace and an injustice, which Murphy Paul rightfully denounces.
300 pp., a few photos
Debunking / North America / Popular Science / Psychoanalysis
Currently out print as a hardback, ISBN 0743243560 [Buy from Powell's]. I presume that this is the same as her The Cult of Personality Testing, (paperback, US$14, ISBN 9780743280723 [Powell's]), but haven't compared the texts.