I've been reading a lot of papers lately about perceptions of social relationships. You might guess that what I think of your relationships is not nearly as relevant as "the truth" of what you think of your relationships. But many sociologists would argue that this desire to get at "the truth" is often misguided.
For example, Martin Kilduff and David Krackhardt argue persuasively that your organizational influence depends more on who people think you are connected to than on who you are actually connected to. Clearly, what I think of your relationships does matter.
Of course, the relationship between perception and reality does count for something. In another study, David Krackhardt reveals that those with more accurate perceptions of advice networks (who asks whom for advice) are rated by their colleagues as more powerful. Interestingly, the same power boost does not come to those with accurate perception of friendship networks. For some thoughts on why some people perceive networks more accurately than others, see this paper by Tiziana Casciaro.
When you get right down to it, even your own perception of your relationships is open to interpretation. One person's sense of "friend" may be equivalent to another's sense of "acquaintance." Allowing each person in an organization to rate his own relationships (0=don't know this person .... 5=close friends) can easily create a matrix of Babylon rife with translation errors, though in many respects it is still the best way to go. Another alternative is consensus. Instead of having each individual determine the ratings for his relationships, compute the ratings based on a collective poll.
Thankfully, not all relationships are so open to subjective interpretation. Sex, for example. Clintonesque beatings around the bush aside, either you sleep with someone or you don't. Not only that, but it's a perfectly symmetric relationship. Unlike friendship, not to mention love, unrequited sexual relationships cannot technically exist.
This, The Economist points out in a lovely article on social networks, is the serious reason why social network analysts are fond of studying sex. Theoretically, it makes for cleaner data less open to argument and interpretation. Funny how sex doesn't seem to work that way in most other contexts.
For an example of a social network analysis of sexual relationships, I recommend this study of Swedish sexual networks. This paper includes a couple provocative pictures -- not the kind you might be imagining, but probabilistic charts describing the likelihood that the entire adult population of Sweden is one giant sexually connected cluster. Conclusion: let's be careful out there!