data from other sources
Interviews provide not just one source of data but one kind of data. And the relationship between that and other kinds of evidence is not necessarily straightforward.
What is sometimes known as post-modern interviewing (Gubrium and Holstein 2003) takes the position that positivist notions of reliability and validity and 'triangulation' misconstrue the nature of what is constructed by interviewees in giving an account of themselves. The validity of such accounts is intrinsic: it doesn't have to be 'checked out'. An appreciation of this con- structivist perspective is a matter of contention. But constructivist or not, there is a theoretical challenge as to the purpose of people's accounts of themselves.
A starting point is to examine the notion of 'triangulation'.
This term is much bandied about by those who have some notion that it involves getting more than one 'fix' on something to locate or confirm it more precisely and accurately. The term has its main (and original) application in surveying, particularly for map-making: a pattern of triangles, covering the whole country, marked at each angle by a trigonometric point (or trig point). In open country this is indicated by a concrete construction rather like the base of a sundial; in cities usually it will be marked by a vertical arrow on the side of a wall. Imaginary lines between these points locate features within the triangles. But of course, in moving from one point to another we are dealing with phenomena which are of the same kind: distance, altitude, angles, and so on.
When the notion of triangulation is applied to human data, it is soon apparent in moving from one fix – for example, what people say about themselves – to other fixation points such as what they do, or what records show, that we are not working in the same dimensions: the 'triangulation' bearings locate something more or less different each time.