|
|
 |
|
|
|
|
system, a question mark, which we noted by its symbol on our code sheets whenever we could not hear what was going on between two patients. As the purgatory of writing numbers on pieces of paper lengthened, more and more transcripts were passed in with question marks on them. (Roth 1966, 190) |
|
|
|
|
|
|
|
|
Lest one think that this would be picked up and corrected at a later point, he continues: |
|
|
|
|
|
|
|
|
To ensure the reliability of our coding, the research design called for an "inter-rater reliability check" once every two months. We learned to loathe these checks; we knew that the coding system was inadequate in terms of reliability and that our choice of categories was optional, subjective and largely according to our sense of what an interaction is really about, rather than the rigid, stylized, and preconceived design into which we were supposed to make reality fit. (Roth 1966, 191) |
|
|
|
|
|
|
|
|
He goes on to describe how the coders conspired to come up with an inter-rater reliability coefficient of .70 on checking days to be able to keep the research going (Roth 1966, 191). |
|
|
|
|
|
|
|
|
Roth argues that this behavior is not unethical, but an inevitable consequence of delegated, large-scale alienated survey research labor"hired hand research." Recent studies of ICD-using coding practices, which are also highly delegated from the point of view of the WHO and the U.S. Public Health Service and largely unimportant from the point of view of certifying physicians, appears to highlight the same sort of phenomenon. |
|
|
|
|
|
|
|
|
Mick Bloor (1991) studied the practice of death certification by physicians. He notes that the practice is low status, isolated work and it is not checked or queried very much at all, even though there are legal provisions to do so. Even where autopsies are performed to check diagnoses, the hospital pathologist does not review the death certificatehis or her job is a clinical or research one! Certification is also unevenly distributed among medical practitioners. Out of 482 doctors in one Scottish city Bloor studied, 31 doctors had signed nearly a third of all the death certificates. He found that there were enormous variations. There was only 61 percent agreement on the diagnosis of the underlying cause of death between clinicians and pathologists; other studies have found that inter-rater reliability varies with the deceased's age and condition, their social class, the practitioners' nationalities, and their ages. |
|
|
|
|
|
|
|
|
Nicolas Dodier has shown how medical judgments, including various ICD-encoded diseases, are transcribed in a fashion that reflects the values and contingencies of the coders' workplace. Again, as one |
|
|
|
|
|