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

Dear Milanov,

Assuming that you don't have the dates when alliances end, you will have to
use a rolling x-year window (as Gulati and some others have used).

I created networks out of longitudinal multi-party alliances by utilizing MS
Access. The following procedure is not as complex as it may first appear. It
is relatively easy to get a grasp of using Access if you have somebody with
a little experience nearby (just in case you haven't used it earlier).

1. Create two access tables with formats:
Table1: alliance_years
        Column1: ID (integer)
        Column2: Year (integer)

Table2: alliance_participants
        Column1: ID (integer)
        Column2: Participant (Text)

2. Add an increasing index number as the first column in your excel file
(just 1,2,3... If you create it by using a formula, make sure numbers are
non-formatted by copying the column and using paste special->values
3. Copy-paste two first columns (id, year) to table alliance_years
4. Make a new version of the excel file to work with, then delete column 2
5. Copy-paste two first columns (id, company) to table alliance_participants
6. Delete column 2 (the company participants you copied to access already)
7. Sort by column 2 (so that empty lines will be on the bottom)
8. Goto step 5 and repeat until all columns with alliance participants have
been copied to Access (along with alliance id).
9. Create a query matching alliance participants dyadically for given year
range. You can do this by query designer, or enter the SQL query manually:

SELECT ap1.participant, ap2.participant FROM alliance_participants as ap1,
alliance_participants as ap2, alliance_years WHERE = and = and year < 1991 and year > 1985 and
ap1.participant < ap2.partipant

(Please note that you need to change the year range for each individual

10. Copy-paste results to excel
11. Save as a text file with hyphenation around cells (company names).
12. Edit the text file by adding the following header (please note that it
doesn't matter what n is as long as it is at least as high as the number of
individual companies in the data set):
        dl n=5000 format=edgelist1 labels embedded type=symmetric
13. Open UCINET and import the text file by pressing ctrl+I. Now you have
the cumulative alliance network for your given year range, and it is easy to
calculate centrality values and other measures.

The benefit of utilizing Access is that you can sort & manually check out
the table alliance_participants for typos or inconsistencies in company
names (e.g. you might want to fix "Merck India" to equal "Merck & Co.").
Also, you will most probably want to test different time windows for
alliance networks if you do not have the ending dates for alliances to make
sure your results are robust.

I hope this helped. Also, it would be interesting to know whether others
have better methods for manipulating data.

Best regards,
Henri Schildt

Helsinki University of Technology

-----Original Message-----
From: Social Networks Discussion Forum [mailto:[log in to unmask]] On
Behalf Of Milanov, Hana
Sent: 28. tammikuuta 2005 19:38
To: [log in to unmask]
Subject: amateur question

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

Dear all,

I am a complete novice in network analysis, and at my very beginning of
using UCINET software. I was wandering if anyone could help me with
inputting and handling data in the program - namely, I have a
longitudinal dataset of some 14000 alliances in a 20 year time span
(thus, non-directed ties). The data is in Excel now, and listed in the
following manner:

Year of alliance              partner1            partner2
partner3 and so on

1980                             xxx                   yyy

1980                             xxx                   aaa

the problem (or not?) is that in the same data:

1.      some companies formed multiple alliances (with different
partners) in the same year (so we have them as a special entry, as in
the example above
2.      some companies have multiple partners in ONE alliance in ONE
year, although rare cases; most of them have just one partner per

our interest is getting degree centrality (for a start) measures for
each company for each year. Could you give me any guidance regarding the
best way to input the data into UCINET, and the possibility of using
centrality measures (I am not sure if UCINET in the process of creating
a matrix would interpret some of the above as missing data, as most of
the companies have 1 partner, but I cannot get rid of those having
multiple partners. Also, should I perform any matrix transformations or
such prior to centrality measures?)

another problem is that some of the companies are not only linked to one
another via alliances, but also have their own subsidiaries as part of
the network data. I believe we should code that somehow as
parent-subsidiary ties are important (and stronger!) but I am not sure
how to approach that.

I know I am full of questions, and probably did not make it clear what
exactly the problem is, but if anyone could help me, I could email part
of the data for easier communication. Thank you!!!!


[log in to unmask]

Hana Milanov

PhD Student

Management Department - BU630A

Kelley School of Business

home phone: 812-333-7285

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.

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.