Here is a common statement made in debates about psychiatric diagnosis:
"[psychiatric disorder x] is an invalid construct"
This "validity critique" of psychiatric diagnosis arose after the technical overhaul of DSM-III, when the psychometric properties of such classifications became a powerful way of understanding their flaws. Originally, it came from seminal work by Richard Bentall and Mary Boyle, both of whom queried the construct validity and predictive validity of DSM-defined schizophrenia. This was an original and useful way to raise problems with schizophrenia-talk, and it has happily found its way into the mainstream of psychiatric discourse. Unfortunately these contributions get kind of watered down in the endless repetition of a "invalid construct" claim as quoted above. What gets left behind is the veneer of apparent truth, without any substantive meaning.
To say "valid" or "invalid" is not very helpful in itself. Those are words that have both a specific technical psychological meaning and a broader lay meaning. Validity in psychometrics is a term used to describe the extent to which a test or checklist measures something that is actually there or successfully predicts some other event. To understand what is meant by "valid" (or "invalid") we need to ask valid in relation to what? Thus a given questionnaire could be an invalid predictor of suicide, but still be a valid indicator of severely depressed mood.
Psychometric validity is difficult to nail down in the realm of diagnosis because it is not clear what might count as a validating criterion. Bentall and Boyle point out several facts about the diagnosis of schizophrenia, including the fact that symptoms do not cluster together and that functional outcome is not uniform. Those are good things to know, and they undermined pervasive myths about the schizophrenia diagnosis.
But in the broader, lay sense of the terns, it is all but impossible to meaningfully say whether diagnosis [x] is really "valid" or "invalid". What, after all, do those words really mean in this context? The dictionary definition simply suggests "grounded in fact", or "reasonable or cogent". On this definition, surely it is valid to say "I feel ill" (if your first person subjective experience tells you as much), or "I have schizophrenia" (if the technical language using community agrees as such). If I have a set of experiences that feel subjectively like illness, cause me to meet DSM-criteria for a disorder and that DSM-diagnosis provides a constructive narrative for helping me to live the sort of life I want for myself, is it still an "invalid" diagnosis? In the technical sense of "invalid", it is arguable. In the broader lay sense of "invalid" it seems the answer is a resounding no!
Another way that psychologists increasingly talk of validation is in terms of recognizing the reality of people's experiences. Thus we validate people's sadness or anger by joining with their perspective and agreeing that anger or sadness was a reasonable thing to feel under the circumstances. Failure to do this is called "invalidating", meaning that it seems to undermine the reasonableness of an emotion. Invalidate someone's emotions and you show them they are wrong to have felt them. Invalidate someone's experience of reality and you hint to them that they are crazy.
Talk of lack-of-validity has been a valuable addition to understanding psychiatric diagnosis. But when it gets isolated from its underpinning arguments, validity talk in this context can be invalidating.