What, you mean the theoretical approach?
Rather than checking to see if the theory still matches up with reality, people are merely looking to see if the theory has internal consistency and using that as evidence that the pattern is valid?
Sure, that would be a logic flaw. (incidentally, one that shows up in ANY type of belief system, whether religion or politics or whatever -- mistaking internal consistency for external validity.)
External validity is contingent upon internal consistency. In other words, we do not know if what we observe is sound unless we have thought about it. Thinking about it is necessarily an internal process.
Internal consistency must hold primacy over the external observations for this reason. As an example, take the theoretical notion that Se manifests as an external perception of sensation.
We look at a person who claims to be an Se, who is in his 60s, for some reason he seems to be using Ni in this situation, so we ought to conclude that Se does not manifest in terms of external perceptions.
A more plausible way to look at the situation would be to say functions are justified by virtue of internal conceptual framework, but the way they manifest in people is a whole another matter.
Type in itself is an unconscious tendency. It has nothing to do with human behavior. Human behavior represent various manifestations of type, but not type itself.
For example, consider the idea of how different a Chinese INTP would be from an Australlian. If we had the fortune of spending time with both of them, our initial hunch (based on the mere positivistic external observation) would be to conclude that they are not INTPs at all. If external validity holds primacy over internal conceptual framework, the whole system shall crumble as it is the internal conceptual framework that makes the system possible. External observations are only part of the system.
Hence, in recapitulation we get that pure typology could be figured out based on the tendencies of mind without the observations of people. (Jung figured this out more from inquiring into how people think, rather than how they behave in social situations. This is why he studied philosophy, literature and biographies, and in therapy focused more on the deeper thoughts of his patients rather than the observations of their basic behaviors in social situations.)
For the purposes of applied typology, carefully controlled empirical studies are necessary. If we are asking for example, how do Se people tend to behave. We need to know what kind of Se people we are dealing with, and in what context. Based on pure typology and information concerning the circumstances we could concoct a hypothesis, yet reliable knowledge requires observations of people.
But once more, that is not the heart of type. If we wish to understand the archetypal quiddity of Se, we need not observe human behavior at all, we need to observe the internal conceptual framework and human thought to continue re-working the aforementioned system. Human behavior will only throw us off as it often shows only very superficial traces of one's thought and how one's mind works. This in effect leads to the confusion of type for personality. That is the error that has led many, especially the new students of the subject to believe that they are their type.