Many areas of science have declined substantially over the past several decades, and one of these is the field of personality psychology - where we have now given-up on the rigorous and multi-disciplinary approach best exemplified by Hans J Eysenck, and are back to the pre-theoretical indeed medieval - world of "that sounds like a good idea".
The thing about inventing categories of personality, is that it is trivially easy, because they all have some predictive validity - therefore there is no end to the proliferation of typologies and categorizations.
For instance: Young children can be classified into two groups by a ten minute test of whether they are able to defer a present reward (e.g. one biscuit) in order to get a greater reward a few minutes later (e.g. two biscuits) - and these groups have statistically different outcomes later in life.
People can be categorized into groups according to the four humours (phlegm, black bile, yellow bile, and blood), of by Jung's categorization developed as Myers-Briggs, or by the Big Five or any number of others - all with some predictive validity with respect to future behaviour.
A small research project I did easily showed that those with a tattoo had a significantly self-reported higher rate of drug and alcohol use than those without.
Another little project showed that those who possessed digital bathrooms scales had a higher rate of reported eating disorders.
The point is there are innumerable traits that are predictive.
Almost any characteristic you can think-of has some value of this kind; and when characteristics have multiple components - i.e. when one creates a "syndrome" of characteristics - the correlations will be greater.
For example; the typologies used in the Diagnostic and Statistical Manual (DSM of the American Psychiatric Association) used in psychiatry; or for that matter the typical star signs used in astrology, they will usually demonstrate statistical differences in outcome. (e.g. Eysenck reported astrological differences many decades ago).
The predictive validity of different schemes can be studied comparatively to seek higher statistical correlations, but this process is very sensitive to sample and population selection, and still lacks discriminative power.
But only Eysenck (and those influenced by him) properly recognized that, in order to make progress, the scientific implication was that personality psychology must go beyond lists of characteristics to include as many as possible cross-correlations drawn from other biological sciences; such as cross-species comparisons, biochemical distinctions (e.g. in brain, or of blood), response to drugs, psychiatric correlates, medical correlates, evolutionary theory - and so forth.
Eysenck made some progress in testing his categories (extraversion-introversion, neuroticism - and to some extent psychoticism) by such means; and more importantly he provided a programme by which this ought to be done.
But Eysenck's programme was abandoned; and the predictable outcome has been a proliferation of personality typologies - each of which has its group of "believers", but none of which go beyond the trivial and common sense knowledge that how people behaved in the past may have an influence on how they behave in the future.
Note: I invite readers to apply the above critique to currently common self- and other-categorizations (including their own favourites) relating to sexual typology and preference, lifestyle, consumer choices, race, ethnicity, residence, political and religious affiliation etc. etc.