- The question of reciprocity (and the realistic reporting of it) can be difficult to untangle. Bruce s two questions on capturing reciprocity are great. I alsoMessage 1 of 4 , Mar 31 8:51 AMView Source
The question of reciprocity (and the realistic reporting of it) can be difficult to untangle. Bruce's two questions on capturing reciprocity are great. I also discovered another way to look at the issue via a blog entry on Centrality which tries to address attention networks as opposed to social networks. In other words, it looks at the value of "fandom" where the relationship is asymetrical yet important. The real issue, in that case, is one of power, and whether or not it's one of the aspects you're measuring in the network in question.
Another point I'd like to make is the simple fact that two parties could report differently on the same relationship is significant in of itself. Each perception of the relationship is "real" to each party, no matter if you inform them otherwise, and their actions will be influenced by their perception unless they can be confronted with empirical proof otherwise (which still may not work). Such empirical proof could be gathered through data-mining techiques, but brings up other issues which I'm pleased to see have already been addressed in the Jump Start series.
So, the real trick is to know when it's important to use perception (opinion) data versus computational (data-mined) data in the network being analyzed.
--- In email@example.com, "brucehoppe" <Bruce@...> wrote:
> Anne raises an excellent question about reciprocity.
> Most discussion of reciprocity I have seen in SNA looks at one of two
> * What to do when a symmetric relationship between two people A and B
> (e.g., talk with) is reported differently by A and B
> * What to do when a particular network calculation (e.g., Eigenvector
> centrality) requires symmetric data but the relationship of interest is
> asymmetric (e.g., advice-seeking).
> Another question is that of perception vs reality. Sometimes the most
> important measure of A's relationship to B is not what A thinks or what
> B thinks, but what everyone else thinks. Here is a great paper on the
> topic by David Krackhardt: "Assessing the Political Landscape
> ng%20the%20Political%20Landscape.pdf> " in which he supports the claim
> that what A and B think of their mutual relationship does not matter as
> much as what others think about it. Krackhardt's discussion takes the
> idea of reciprocity and expands it to "cognitive social structures
> ve%20Social%20Structures.pdf> ."
> --- In firstname.lastname@example.org, "nnaelah" annehale@ wrote:
> > I would like to understand reciprocity as used within sna (ona)
> > My frame of reference is sociometry. Ex: Here is a one-way criterion:
> > "Whom do I go to for information on innovation within the supply chain
> > network?" It makes sense that a person you choose may not necessarily
> > choose you back; however, when sociometrist's explore choices we are
> > able to first identify on one data sheet (similar to force field
> > analysis) the strength of a pull to choose and pull not to choose each
> > person, and then indicate on the data sheet the choice whether to
> > choose, not choose, remain neutral toward (high to low range). All
> > relationships are viewed as reciprocal to some degree, and the data is
> > depicted in phase space (non-linear) before depicting the choice data
> > (linear). We are also able to ask for perceptual data in this
> > if we want to be innudated with information. What processes within
> > are similar and what data depiction programs are used which have this
> > degree of clarity about reciprocity?