Print

Print


*****  To join INSNA, visit http://www.sfu.ca/~insna/  *****

Nonrespondents in Communication Network Studies: Problems and Possibilities
Stork, Diana; Richards, William D.

Group & Organization Management; Jun 1992; 17, 2; ABI/INFORM Global, pg.
193-209

This article explores nonresponse problems in network research.
Following a general introduction to the analysis of incomplete data
sets, we focus explicitly on incompleteness caused by non-respondents.
We explain why nonrespondents cause different and theoretically more
complicated problems in network research than in research which uses
individuals or groups as the units of analysis. We provide an example
that demonstrates how much data is actually missing when, in a 60 person
network, 15 members do not respond to a sociometric survey. We discuss
approaches to the analysis of such incomplete data sets. We show how
they work by applying them to the analysis of work communication at
Ultra. Recognizing that response rates are rarely 100%, we conclude with
some suggestions for designing sociometric studies that will facilitate
the analysis and interpretation of network data.

Both researchers and consultants are finding network metaphors and
network methods to be useful ways to conceptualize and operationalize
patterns of communication, influence, friendship, and authority in
organizations.
:
:

RESPONSE RATES
It is easy to collect sociometric data, "since anyone can ask
sociometric questions" (ref). They are difficult to analyze, however,
particularly when data that describe network relationships are missing.
A relationship "is not the property of an individual [but rather] a
characteristic that is defined in reference to two ... people taken
together" (ref). Thus complete description requires information from
both individuals in a relationship. To analyze a network of
communication relationships, researchers "would like to collect data
from all [its] members" (ref). In practice, however, returned surveys
often represent less than a 100% response rate, and "missing data are
... a curse to survey network data [because] network analysis is
especially sensitive to missing data" (ref). Missing data pose a
particularly serious problem for network analysis at the system level
because they may create "huge holes in the who-to-whom data matrix"
(refs) that distort the system's communication structure. Yet response
rates reported in the literature suggest that network researchers are
often faced with having to analyze data sets with response rates between
90% and 65% (refs).

This article examines the problems caused by nonrespondents and
describes approaches to the analysis of incomplete data sets that may
lessen the impact of missing data. Suggestions are offered for the
design of network studies and data-collection instruments that will
improve response rates and provide the kind of information needed to
justify decisions about how incomplete data sets are analyzed.

----

If you would like to read the whole paper, please send an email message
to me at [log in to unmask] and I will send a pdf to you.

Bill Richards

Orly Tenne-Gazit wrote:

>*****  To join INSNA, visit http://www.sfu.ca/~insna/  *****
>
>Orly Tenne-Gazit wrote:
>
>***** To join INSNA, visit http://www.sfu.ca/~insna/ *****
>
>Hello All, I'm a PhD student working on networks-related research.  I have a methodological question that I hope you might have an answer to:
>
>When measuring full networks for work units (15-25 people each):
>1) How much missing data* is bearable?  And why?
>2) What can be done to deal with missing values when analyzing full networks?
>
>*By missing data I mean people from the work unit who did not fill out the questionnaire, so I don?t have their report of their network ties with the others in the unit.
>
>Thank You in advance!!!
>
>Orly.
>
>_____________________________________________________________________
>SOCNET is a service of INSNA, the professional association for social network researchers (http://www.sfu.ca/~insna/).  To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.
>
>

_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.sfu.ca/~insna/). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.