Family diagnosis: research methods and their reliability for studies of the medical-social unit, the family.

Abstract

An instrument to study family functions (called Family Diagnosis) was developed and tested for reliability. While standard social data were obtained with fair reliability, there was a large and variable degree of variation on all other types of social and attitude data. The model proposed seems sound, but the variability shown within this framework indicates further need to define reliability in each study of medical-social data.

On establishing the validity of ‘objective’ data: can we rely on cross-interview agreement?

“…However, they still make the claim that this agreement' in itself, is evidence of the validity of the marriage rating since it shows that ratings based on accounts from different marriage partners give much the same picture.' I would argue, on the contrary, that given the possibility of common bias in the accounts the measure can only be said to be reliable (see Haggerty, 1965); its validity has not been demonstrated one way or the other.…”

“…with regard to reading chest X-rays or blood pressure measurements. He quotes Johnson-Abercrombie's work (Johnson-Abercrombie, 1960) which illustrates 'how the same observer's perception may vary depending upon the circumstances during the observation, and how different observers vary after observing the same event depending upon their personalities' (Haggerty, 1965(Haggerty, , pp. 1521(Haggerty, -1522.…”

“…Haggerty suggests that agreement by ' different observers of quite different training, using dif-ferent methods and with a different type of relation to the family' should be distinguished from agreement obtained where equally trained observers use the same methods. The former approach approximates more to a measure of validity than of reliability (Haggerty, 1965(Haggerty, , p. 1529. Adopting Haggerty's position, it could be said that across-interview comparisons using household family members provide a measure of inter-informant reliability rather than an indication of the validity of any measure based on a particular informant's report.…”