Because some personality traits are disadvantageous to people who have them, and some people have personality traits that they would self-modify away from if they were rational actors, etc.
On a meta level, classification systems that have "not labeling anyone as better than anyone else" as a selling point sometimes sacrifice precision or accuracy to do so.
classification systems that have "not labeling anyone as better than anyone else" as a selling point
They would avoid doing that, but they're told that not labeling anyone as better or worse is just as worthwhile as precision and accuracy.
While psychology wonks have been going on for years about the statistical rigor and calibration of the Big Five, most people have just carried on using the Myers-Briggs type indicator (MBTI), which may not be statistical or scientific but is able to categorize people without insulting them.
A serious critique of the MBTI is the Myers-Briggs entropy distribution paradox (or, "Why are there 16 personality types when everyone I know is an INTJ?") A new personality test which has been gaining ground recently, the MLPTI, does not break up the INTJ into multiple categories; but does reduce the number of bothersome non-INTJ personality types and thus ameliorates the entropy paradox. For those not yet familiar with it, here is a rough translation between MLPTI and MBTI types.
The loss of half of the MBTI categories is not a serious problem, as demonstrated by the fact that you can't even name the ones that were left out without going back and looking. Seriously, when was the last time you met an ENTP?