If one box was "you get the job" and the other was "rejected" -- I can see why people would lie to achieve one result or another.
Again, which is why I mentioned in my post that there are scientific problems with the field as a whole, so I don’t understand the undue targeting of MBTI compared to everything else. You realize people are prescribed or not prescribed controlled substances on the basis of their responses to other questionnaires, right? You can look up the DSM criteria and get a diagnosis for just about anything based on your responses.
That said, the DSM has its own problems.
But what is it for then?
Ok, so what is it for? And how do you justify its use?
I've seen what it's being sold for/as.
I have not bought it, but I have used it. I justify using it because it is useful to me: It helps me understand how people very different from myself interact with their world. These other people are not wrong, just different. Remarkably often, the differences are quite comprehensible. It helps to organize those differences. The precise details of how to organize them are not especially important, because it is just a way to break up what would be a lot of confusing detail into chunks that minds can grasp more easily. If it matches how other people organize, then we can talk about it.
Certain ways of organizing differences satisfy more criteria than others, and so have attracted more people to use them. MB is one that does, and so has. That doesn't make other systems wrong; and that other systems exist doesn't make MB wrong, even if different systems are more useful in one place than another.
If just reading the descriptions is enough for you, do that. Most people seem to need more structure.
Maybe somebody who could never in a million years do what you do, but finds things easy that you could never do. You could hope they are better at figuring out how to communicate with very different people than you are. Or, you could learn it yourself. It turns out not to be so hard, with the right tools. MB is one. It's not the only one, and not the best for any particular thing, but it's quick and easy to pick up.
Some people have said the same of astrological signs. (EDIT: As per @simias' comment on confirmation bias. This is why "text blurb profile"-oriented reasoning is especially suspect. We already have ample evidence of that reasoning being as unreliable guide to truth as is unreliable a belief that your birthday determines your personality.)
Honestly, the more "words" you add to try to summarize a pile of question answers, the further away you are getting from "science" and the closer you get to astrology. If there is only one lesson I can convey here, it should be that. The science is the 50 or however many questions dimensional questionnaire and the 4 or 5 dimensional distributional summary of the full high dimensional distribution. You project "down from the questions" to the summary and then people using English "back project" but are mostly just filling in poetic detail from anecdotal experience. The bigger that back projection to paragraphs and chapters the more like astrology or psychology by and for poets things have become.
However, as mentioned a couple of times, these summaries may be telling you very little (not "probably says" as you say a few times, but "probably says nothing") about the 99% neurotypical people (yes, my estimate is hand-wavy based on the distribution along just one axis from the article which says the other are "similar"). The tests may be only useful to identify outlying personality extremes which is very much not how amateurs apply them. Amateurs overconclude..reading profiles and basically acting out the epistemological play of astrology over again (but backed by "science" so they feel less guilty; EDIT - and yes I have known several such people in real life).
Incidentally, this also explains "correlations" with other things (like IQ or engineers or etc.) because linear statistics like correlation are notoriously dragged around by outliers. To whatever extent these things are true, the correlations (unadjusted for individual noise as already mentioned) will have much weaker statistical power than you would naively expect. In effect, your sample size may be only 0.2% to a few % what you thought it was. So, instead of 1000s of people you have 10s with thin population of each of the 16 categories. This would show up in the reproducibility of the correlation studies, but honestly require a more careful meta-analysis than I am prepared to offer for free. So, these studies, too, may have a grain of truth, but perhaps only a grain and this grain is probably much smaller than what personality test advocates in this comment thread seem to understand.
At least one real world problem with taking these measurements too seriously is, as others in this thread have mentioned, that these thinly evidenced divisions - even if real - can be as ripe for abuse as race/whatever designator is. In-group preference [1] is a strongly replicated result in psychology. This is another reason why axes-only "a% I, b% N, c% T, d% P" are better than classification, and why the big 5 (which at least resists categories) is less harmful. a +- A, b +- B, etc. would be even better.
Have you or anyone on this whole thread ever heard of a p-value for your personality assignment or error bars on your personality axes before I mentioned them? (Yes, I know the problems with p-values. Those problems are not the point here.) No? Yet everyone has heard of same-person reproducibility problems. These are obviously deeply related issues.
So, the problem is not that "MB is totally not science". It's that personality measurement in general is weak enough science to be far more likely to be misapplied than to be correctly understood, especially by lay practitioners like HR depts/mgmt/people on first dates/people in emotionally heated scenarios. It may have a real and even positive role in diagnosing true outlier children in need of interventions, though "profiles" of all the super extreme 16 cases will probably sound bad and may be obvious sans any testing.
You can always lie, about anything you like. You can put salt in the sugar bowl. You can substitute decaf in for the coffee at the office, and give everybody caffeine-withdrawl headaches. You can have your dog crap in the sandbox at the park. What do you imagine you would prove, by that? (I give you exactly one guess.)
All you did lying on the test was to land in a different box than if you had not lied. But one box is not better than another box, so what good did that do? It means somebody who looks and believes it will find it harder to communicate with you. What you achieved was to throw sand in the gears. Congratulations.