Dear Gaetano,
Thanks for your reply. To clarify, and perhaps think aloud. Apologies
to all for the long post.
My specific interest is in gathering information about informal
networks in organisations, and later linking it to other information.
In a department you can gather network data using a questionnaire
format but beyond about 60 people it is too time consuming. Therefore,
what I am after is a tool that will help the clever gathering of
data.
In answer to your specific questions:
i) The HR department of most organisations will have basic demographic
information and organigramme (formal network) data. This can be used
to identify people who are functionally related to each other, but
also people who are similar to each other in some way (age gender
ethnicity etc.)
ii) The ties can be processed later, and values dichotomised into an
adjacency matrix using R or similar tools (R is good of course because
it is OS). So what is needed is a clever way of building up a long
potentially quite sparse vector that could be combined to make a NxN
matrix.
iii) I don't see this using private information from other
sources(e.g. contact list) than the organisation and what the
respondent discloses.
This is where companies like Facebook and LinkedIn have put a lot of
effort. I was wondering whether there was something simple that we
could use that would assist in gathering these data.
An initial approach would be to enable people to select concentric
groups based on a theme - such as office location, job type.This would
enable a pro-active identification - e.g. I still know some people in
the Bujumbura office so I select that office location.
Also, a reactive data collection, and this is where the social network
sites are so good - A wizard might then be able to help through using
an algorithm based on people they might know, this is vital in
gathering weak tie information.They went to the same school, etc., but
increasingly, they have a tie in common or some other commonality.
The first step would be:
Who are the people you have at least a passing acquaintance with?
Having built up a list of people that they know you could then ask
them relational information like the following.
How would you describe your relationship with X (perhaps indicating
using a star system as on ebay purchase rating, although it would need
to be more than just 1-5, each star would need to have a categorical
meaning that was clear)
Do you know X well enough to share something you would not want made
public within the organisation?
And so on.
This is standard questionnaire format information.
These would be valued data, but could be dichotomised and processed
later on. Each node would have a unique i.d. so that the data base
could be built up safely. So if I imagine for example a 6000X6000
matrix in a study, it would be being populated incrementally, and the
suggestions would become better informed over time.
There would also need to be some means which would identify when the
suggestion tool was firing blanks, and say - I think we're done.
Does that make sense?Is it technically feasible?
Thanks for your patience,
Nick
···
On Jun 6, 10:50 pm, Gaetano Borriello wrote:
> Nick:
>
> this sounds interesting. I don't know of anything right off-hand for
> this. Could you be more specific as to how you see the demographic
> data being connected? For example, would you ask people if they know
> others that have reported similar GPS coordinates? work in the same
> facility? have the same surname? or have specified some group
> affiliation? And then, how would you verify that the relationship is
> two-way, that two people have connected to each reciprocally? Or were
> you thinking something that would actually tie in to the Facebook
> friends' list?
>
> Gaetano
>
>
>
>
>
>
>
> On Wed, Jun 6, 2012 at 1:26 AM, Nickdunc wrote:
> > Dear All,
> > I came across ODK today, and it looks full of promise.
> > I can see that it would readily provide a solution to questionnaire
> > data acquisition, which is great as I do that in developing countries.
> > What I have been looking for is a means of gathering social network
> > data which extends beyond the scope of a questionnaire.
> > For example, in a large organisation spread across the country, you
> > might know people in your department, your local (or previous
> > locales), people from your home region or your ethnic group and so on.
> > So with some demographic data in place a facebook/linked-in process
> > might take place in which people would be able to efficiently identify
> > ties across a large dataset.
> > Are you aware of something of this sort generally? Or is there perhaps
> > some package compatible with ODK that would enable this sort of data
> > collection in the field and integrated into the ODK database. Given
> > that this is for use in development settings, I'm very keen that it be
> > open source if at all possible.
>
> > Many thanks,
> > Nick
>
> > --
> > Post: opendatakit@googlegroups.com
> > Unsubscribe: opendatakit+unsubscribe@googlegroups.com
> > Options:http://groups.google.com/group/opendatakit?hl=en