Hard and soft data: a semiotic point of view.

Correspondence: Prof. Dr. med. Rolf H. Adler Medical Division Lory University of Berne Inselspital CH-3010 Berne e-mail: [email protected] Attitude and bias towards hard and soft data, although often concealed, play a role in clinical medicine, teaching and research. In the literature, the problem of hard and soft data has only occasionally been discussed [1, 2]. It is found in the area of qualitative and quantitative research [3, 4], of test-sensitivity and specificity, and of clinical biostatistics [5]. However, the terms qualitative, quantitative, sensitive, specific are free of value judgement and the pejorative connotation which is attached to the hard-soft differentiation. As Feinstein wrote: “... hard data have been excessively venerated to an extent far exceeding their inherent importance or actual reliability; and soft data have been not merely de-emphasized, but deliberately excluded or eliminated from consideration. For the clinico-statistical worshipper ... soft data are not just ‘dirty’ and sinful; they are scabrous horrors, to be expunged from civilized numeracy.” [5] Perhaps the subject matter is of no importance; anyone dedicated to a serious scientific approach – and who is not – swears by hard data. The questions are seldom posed as to what distinguishes hard data, do they exist at all, how do they differ from soft data, or is there room for soft data in “serious” medicine.