You are quite right that accurate diagnosis is very important to the care and treatment of any condition. Sometimes it is useful to treat symptoms alone (as with most kinds of headache or mild anxiety); much of the time, especially when the disorder is serious or the symptoms are a harbinger of such a disorder, misdiagnosis can be a real problem. Current diagnostic knowledge isn't perfect, nor is its application by all practitioners (in psychiatry or anything else), but it is extremely useful.

There have been a great many studies of psychiatric diagnosis, of diagnosing specific disorders, of the validity and reliability of various diagnoses in various settings, and of the methods used in diagnosis. The studies have involved many, many thousands of both diagnosed and previously undiagnosed patients over the years, and thousands of clinicians, statisticians, and related researchers and staff.

For summaries of such studies, in a clinical "textbook" context, consult any general psychiatry or psychiatric diagnosis textbook (there are many), or pick up a Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV-TR), at a local bookstore or Amazon.com. They make fascinating reading. :-)

For primary sources of such studies, consider searching three specific areas:

(1) The reference material for the DSM-IV and prior editions. You want the reference support information, which is in volumes separate from the DSM-IV-TR itself, and available from American Psychiatric Publishing, Inc., and maybe at places like Amazon.com. Every edition of the DSM since DSM-II, at least, has such written support material.

(2) The reference material for the APA's Practice Guideline series. These references are included in the Practice Guidelines for each disorder or group of disorders, each of which has several hundred pages. The texts are available at the same places as the DSM texts.

(3) The National Library of Medicine's online search service, called PubMed. PubMed is a huge listing of articles in peer-reviewed journals (research, clinical, etc.) in virtually all medical and medically-related fields (including psychiatry/psychology/mental health) which indexes only journals which have passed their test for scientific merit (there are hundreds, maybe thousands, of different journals. The URL for a PubMed search is www.ncbi.nlm.nih.gov/entrez/query.fcgi. It takes a little investigation to learn how to do a good search and interpret the results, but there are plenty of help links. If you don't learn how to do it, you are very likely to end up either with very little useful information or with thousands of irrelevant articles to sift through -- sort of like a complicated GOOGLE.