¿ªÔÆÌåÓý

ctrl + shift + ? for shortcuts
© 2025 Groups.io
  1. Ucinet
  2. Topics

Composition (heterogeneity/diversity) between and within groups #help

 
Edited

Hi Everyone,

I am calculating each ego's composition of networks by Ego Networks | Egonet Composition | Categorical Attributes. I have work and idea-sharing relationships between actors from a whole network. The procedure runs fine and I have generated IQV and Blau's index. However, I wanted to know the diversity of roles within one group such as ego belongs to finance staff, what is the composition (heterogeneity/Blau's index) of the finance people within his egonet (within diversity), and what is the composition of others (between ego's group and other roles - between diversity) (e.g., marketers, sales, etc.)
?
Blau's index for within groups will be, I think, using the proportion of his groups' ties only.
For between groups, other proportions can be used to calculate in excel, I reckon. I am unsure.

Is there a way of doing these in UCINET? If not, would UCINET team be able to incorporate these into UCINET and provide these in the output as separate columns: ego's category, within heterogeneity and between heterogeneity, perhaps IQV would be appreciated as well?

I went through the forum's chats but I couldn't come across a solution. Any thoughts or comments are appreciated. Thanks!
 

¿ªÔÆÌåÓý

Yes, that could be added. But one way to do half of what you want right now is this:

?

Go to data|filter|multiple subgraphs from partition. Fill out form like this:

?

?

Note that the network is pv504, the attribute dataset is pv504attr, and I have selected ¡°deptid¡± as the specific attribute of interest. The result is:

?

EXTRACT SUBGRAPH VIA ATTRIBUTE VECTOR

--------------------------------------------------------------------------------

?

Input Network dataset:????????????????? pv504 (C:\Users\sborg2\Dropbox\data\ucinet data\pv504

Input Attribute dataset:??????????????? pv504attr (C:\Users\sborg2\Dropbox\data\ucinet data\pv504attr

Prefix for output files:??????????????? pv504

Column of attribute dataset:??????????? DeptID (C:\Users\sborg2\Dropbox\data\ucinet data\DeptID

Match Nodes by:???????????????????????? Labels

?

?

Saved pv504-1

Saved pv504-2

Saved pv504-6

Saved pv504-4

Saved pv504-30

Saved pv504-20

Saved pv504-10

Saved pv504-3

Saved pv504-7

Saved pv504-5

?

Each of these saved datasets contains the network within one department. Then run ego net composition using each of these networks (and some other attributes, like gender).

?

Steve.

?

From: [email protected] <[email protected]> On Behalf Of ozlemozkok via groups.io
Sent: Tuesday, January 12, 2021 02:05
To: [email protected]
Subject: [ucinet] Composition (heterogeneity/diversity) between and within groups #help

?

[Edited Message Follows]

Hi Everyone,

I am calculating each ego's composition of networks by Ego Networks | Egonet Composition | Categorical Attributes. I have work and idea-sharing relationships between actors from a whole network. The procedure runs fine and I have generated IQV and Blau's index. However, I wanted to know the diversity of roles within one group such as ego belongs to finance staff, what is the composition (heterogeneity/Blau's index) of the finance people within his egonet (within diversity), and what is the composition of others (between ego's group and other roles - between diversity) (e.g., marketers, sales, etc.)
?
Blau's index for within groups will be, I think, using the proportion of his groups' ties only.
For between groups, other proportions can be used to calculate in excel, I reckon. I am unsure.

Is there a way of doing these in UCINET? If not, would UCINET team be able to incorporate these into UCINET and provide these in the output as separate columns: ego's category, within heterogeneity and between heterogeneity, perhaps IQV would be appreciated as well?

I went through the forum's chats but I couldn't come across a solution. Any thoughts or comments are appreciated. Thanks!
 

Oh thanks very much for that Steve. I'd been wondering about precisely this but hadn't gotten?around to posing the question here.
Now I won't need to!
thoughtfully
Starling


On Tue, Jan 12, 2021 at 4:35 PM Steve Borgatti <sborgatti@...> wrote:

Yes, that could be added. But one way to do half of what you want right now is this:

?

Go to data|filter|multiple subgraphs from partition. Fill out form like this:

?

?

Note that the network is pv504, the attribute dataset is pv504attr, and I have selected ¡°deptid¡± as the specific attribute of interest. The result is:

?

EXTRACT SUBGRAPH VIA ATTRIBUTE VECTOR

--------------------------------------------------------------------------------

?

Input Network dataset:????????????????? pv504 (C:\Users\sborg2\Dropbox\data\ucinet data\pv504

Input Attribute dataset:??????????????? pv504attr (C:\Users\sborg2\Dropbox\data\ucinet data\pv504attr

Prefix for output files:??????????????? pv504

Column of attribute dataset:??????????? DeptID (C:\Users\sborg2\Dropbox\data\ucinet data\DeptID

Match Nodes by:???????????????????????? Labels

?

?

Saved pv504-1

Saved pv504-2

Saved pv504-6

Saved pv504-4

Saved pv504-30

Saved pv504-20

Saved pv504-10

Saved pv504-3

Saved pv504-7

Saved pv504-5

?

Each of these saved datasets contains the network within one department. Then run ego net composition using each of these networks (and some other attributes, like gender).

?

Steve.

?

From: [email protected] <[email protected]> On Behalf Of ozlemozkok via
Sent: Tuesday, January 12, 2021 02:05
To: [email protected]
Subject: [ucinet] Composition (heterogeneity/diversity) between and within groups #help

?

[Edited Message Follows]

Hi Everyone,

I am calculating each ego's composition of networks by Ego Networks | Egonet Composition | Categorical Attributes. I have work and idea-sharing relationships between actors from a whole network. The procedure runs fine and I have generated IQV and Blau's index. However, I wanted to know the diversity of roles within one group such as ego belongs to finance staff, what is the composition (heterogeneity/Blau's index) of the finance people within his egonet (within diversity), and what is the composition of others (between ego's group and other roles - between diversity) (e.g., marketers, sales, etc.)
?
Blau's index for within groups will be, I think, using the proportion of his groups' ties only.
For between groups, other proportions can be used to calculate in excel, I reckon. I am unsure.

Is there a way of doing these in UCINET? If not, would UCINET team be able to incorporate these into UCINET and provide these in the output as separate columns: ego's category, within heterogeneity and between heterogeneity, perhaps IQV would be appreciated as well?

I went through the forum's chats but I couldn't come across a solution. Any thoughts or comments are appreciated. Thanks!
 

Thanks for the detailed and quick response, Prof Borgatti.
 
Edited

Hi Prof Borgatti,

Thanks again. I ran this, this gave me 17 dept networks as you suggested. I am seeking an ego-level (ego network level) result for within/between compositions per dep.
After getting these 17 dept networks, I can run ego composition with another attribute (not dept attr in the second step as it all gave me zeros in the second step) in (within) them, but how do I get across/between composition in the second step? Is that what you mean when you said you can do half of what you are saying in your response?


On another note:
I tried the Egotize procedure and then run ego composition by dept but this did not seem to give me reliable results, as I tried it on campnet data with roles students and teachers. Most ended up in zero.
The solution you suggested is somewhat equivalent to this if the partition is done by nodeid (which is I think equivalent to egotize procedure), am I correct?

I wanted to see within and between department tie composition/diversity in one ego's network. So I am running the 203X203 node matrix as my network data along with attribute data using deptid as in the ego composition by category and I am not getting between and within composition (my initial question stands), but i get one composition value for the whole ego network. As you said, if this within/between step can be added to UCINET, we would find it useful.
Thanks again.
Regards.
Ozlem
 

Hi Prof. Borgatti and UCINET team,

I posted the idea about between-diversity calculations based on Blau's index and you responded that this could be added. I wonder whether this is implemented, yet. If not, can you please tell me whether you will implement this? I want to use it. I checked the version history but I could not pinpoint this feature addition. I appreciate your thoughts. Thanks.?
 

¿ªÔÆÌåÓý

Hi, I can talk a look at this next week. I believe our previous correspondence was:;

?

?

?

From: Borgatti, Steve
Sent: Tuesday, January 12, 2021 10:35
To: [email protected]
Subject: RE: [ucinet] Composition (heterogeneity/diversity) between and within groups #help

?

Yes, that could be added. But one way to do half of what you want right now is this:

?

Go to data|filter|multiple subgraphs from partition. Fill out form like this:

?

?

Note that the network is pv504, the attribute dataset is pv504attr, and I have selected ¡°deptid¡± as the specific attribute of interest. The result is:

?

EXTRACT SUBGRAPH VIA ATTRIBUTE VECTOR

--------------------------------------------------------------------------------

?

Input Network dataset:????????????????? pv504 (C:\Users\sborg2\Dropbox\data\ucinet data\pv504

Input Attribute dataset:??????????????? pv504attr (C:\Users\sborg2\Dropbox\data\ucinet data\pv504attr

Prefix for output files:??????????????? pv504

Column of attribute dataset:??????????? DeptID (C:\Users\sborg2\Dropbox\data\ucinet data\DeptID

Match Nodes by:???????????????????????? Labels

?

?

Saved pv504-1

Saved pv504-2

Saved pv504-6

Saved pv504-4

Saved pv504-30

Saved pv504-20

Saved pv504-10

Saved pv504-3

Saved pv504-7

Saved pv504-5

?

Each of these saved datasets contains the network within one department. Then run ego net composition using each of these networks (and some other attributes, like gender).

?

Steve.

?

From: [email protected] <[email protected]> On Behalf Of ozlemozkok via groups.io
Sent: Tuesday, January 12, 2021 02:05
To: [email protected]
Subject: [ucinet] Composition (heterogeneity/diversity) between and within groups #help

?

[Edited Message Follows]

Hi Everyone,

I am calculating each ego's composition of networks by Ego Networks | Egonet Composition | Categorical Attributes. I have work and idea-sharing relationships between actors from a whole network. The procedure runs fine and I have generated IQV and Blau's index. However, I wanted to know the diversity of roles within one group such as ego belongs to finance staff, what is the composition (heterogeneity/Blau's index) of the finance people within his egonet (within diversity), and what is the composition of others (between ego's group and other roles - between diversity) (e.g., marketers, sales, etc.)
?
Blau's index for within groups will be, I think, using the proportion of his groups' ties only.
For between groups, other proportions can be used to calculate in excel, I reckon. I am unsure.

Is there a way of doing these in UCINET? If not, would UCINET team be able to incorporate these into UCINET and provide these in the output as separate columns: ego's category, within heterogeneity and between heterogeneity, perhaps IQV would be appreciated as well?

I went through the forum's chats but I couldn't come across a solution. Any thoughts or comments are appreciated. Thanks!
 

Thank you.

1 - 8 of 8