*****  To join INSNA, visit  *****

Dear Gueorgi Kossinets,

Thank you very much for replying to my query. Many thanks also to everyone
else who replied to my beginners question!!

> Not knowing your problem well, I could only
> guess that much will depend on how your network has come
> about -- e.g. whether it is an affiliation graph or a result
> of some other underlying social process.  For a random
> graph, estimated bias caused by missing data varies with
> average degree and network size.  If you require that
> nominations are reciprocated then the non-response effect
> should be much stronger.  Conversely, you might want to add
> a link to your network if it is confirmed by any one of
> respondents.

I try to study cliques and communication and support networks within a
centrally organised monastic organisation of 870 mendicants of four
categories (monks, nuns, male and female novices) which are divided into
some 110 subgroups whose membership is rotated by the head of the order.
These subgroups roam independently all over India and about half of them
manage to meet the head of the order annually to be re-grouped. The rules
of the order severely restrict communication, particularly between nuns
and monk. The development of permanent subgroups is systematically
prevented by the rotion of personnel between subgroup-leaders. The
question is therefore, to what extent the is the order socially
integrated, which informal cliques continue to exist and how do the
restrictive rules shape the network.

Out of 10 network questions only 2 produced a sufficiently high response
rate of ca. 250: 'who do you know personally' and 'who did you meet lat
year'. The questions relating to gifts and letters were generally not
answered since 'exchanges' are officially not allowed though asymmetrical
'gifts' and 'messages' are communicated via the mediation of the leader.

Being a beginner in sociocentric network analysis, I wonder (1) whether
such a response rate is generally regarded to be sufficient for a
sociocentric analysis, and (2) how to prepare the data for further
quantitative analysis, if commended, which I will learn as I go along. I
am thinking of basic step-by-step procedures of data cleansing, checking
consistency and representativeness etc. to get started in UCINET which
will be routine for most of you. The good news is that I know most of the
group processes through long-term contacts.

I received mixed responses so far regarding the first question and already
some good advice regarding the second. It seems to assume symmetry is
acceptable with regard to the two questions mentioned?

Compared to other studies 250 is a very high sample but I had reasons to
believe that I get a full response at least on the two questions
mentioned. For reasons of finance and time (one year) data collection had
to be stopped however (to contact small mendicant groups constantly on the
move over a large territory is extremely difficult and time consuming).

(3) A more general question relates to possibility of mapping the
geographical dynamic of a network of people on the move. Has anyone worked
on such a complicated problem before?


Peter Fluegel

SOCNET is a service of INSNA, the professional association for social
network researchers ( To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.