Dynamic content analysis.

An experiment was performed to demonstrate that methods of linguistic content analysis suitable for the study of intra-interview processes are both possible and practicable. Six hundred thirty-eight randomly selected psychiatrist were sent rating protocols of machine-mediated interview transcripts (transcripts of interviews conducted by means of remotely located teletypes) and were asked to rate each patient communication on one or more of 13 diagnosis-related rating dimensions. Rank correlation statistics on the raters' levels of agreement were found to be significantly high in 11 of the 13 dimensions. In contrast to the commonly held opinion that psychiatrist lack consistency in their application of diagnostic terms, this study suggests that psychiatrists generally agree in their usage of these terms, provided that the behavioral samples being evaluated are appropriately delimited.