Keyboard Shortcuts
Likes
- Ucinet
- Messages
Search
|
Network Diameter and Radius
In UCINET I can easily choose 2-mode analysis to show network cohesion values such as diameter and radius. However, with 1-mode analysis of the converted matrix I cannot seem to locate where one can find diameter and radius although there are several others displayed (28), i.e., Network-->Whole Network-->Multiple Measures. Any advice is welcome. |
|
Binarizing Continuous Data
I want to binarize some continuous data (latitude, longitude) such that the datasets are all binary. It also seems that network analysis software is more "friendly" towards binary data.in my testing so far. The method I am considering is simply coding for presence absence of either the mean+/1SD or not, which should capture 2/3 of similar values. One could break each continuous attribute into several categories but that would greatly add the number of attributes when originally there was only one attribute. Any other ideas would be welcome. Thanks. |
|
Re: Spherical Distances from GIS/GPS Coordinates
Thank you that helped me think about SNA and spatial analysis which led to thinking about SNA and GIS which led me to searches on those topics, A fair amount of work has been done in areas such as epidemiology/disease research using both and I assume this is an important part of what the big tech companies are now doing. I was just curious why the distance tool was in UCINET as I did not see how it linked to the rest of the software.
|
|
Re: Spherical Distances from GIS/GPS Coordinates
I tried to post this reply yesterday but it apparently disappeared. Apologies if this is a repeat post. This is one of the things I did in a recent study (currently under review). ? We wanted to determine whether geographical proximity influences the decision of local councils in England to collaborate. I calculated the distance between each pair of councils from their published latitude and longitude coordinates. I had to transform the coordinates to account for the different ‘sizes’ of degrees lat and long in England (that is, converting the lat and long coordinates to miles relative to a convenient origin). Then I used UCINET’s matrix operations to calculate the Euclidean distances between each pair of councils. (Perhaps the ‘tools’ command does this more conveniently – I didn’t try that.) ? We used the matrix of council-council partnerships as the dependent variable and the matrix of inter-council distances as an independent variable for MR-QAP (in fact we found that administrative boundaries were more important than simple proximity). We also looked at the distribution of distances between councils in various types of partnership. ? This paper gives some examples and references: Clio Andris (2016) ‘Integrating social network data into GISystems’ International Journal of Geographical Information Science, 30:10, pp.2009-2031? ? I'm not an expert on SNA, so hope this is useful! Ruth ? |
|
Re: Spherical Distances from GIS/GPS Coordinates
I tried to post this reply yesterday but it apparently disappeared. Apologies if this is a repeat post. This is one of the things I did in a recent study (currently under review). ? We wanted to determine whether geographical proximity influences the decision of local councils in England to collaborate. I calculated the distance between each pair of councils from their published latitude and longitude coordinates. I had to transform the coordinates to account for the different ‘sizes’ of degrees lat and long in England (that is, converting the lat and long coordinates to miles relative to a convenient origin). Then I used UCINET’s matrix operations to calculate the Euclidean distances between each pair of councils. (Perhaps the ‘tools’ command does this more conveniently – I didn’t try that.) ? We used the matrix of council-council partnerships as the dependent variable and the matrix of inter-council distances as an independent variable for MR-QAP (in fact we found that administrative boundaries were more important than simple proximity). We also looked at the distribution of distances between councils in various types of partnership. ? This paper gives some examples and references: Clio Andris (2016) ‘Integrating social network data into GISystems’ International Journal of Geographical Information Science, 30:10, pp.2009-2031? ? I'm not an expert on SNA, so hope this is useful! Ruth ? |
|
Re: Spherical Distances from GIS/GPS Coordinates
This is one of the things I did in a recent study (currently under review). We wanted to determine how much geographical proximity influences the decision of local councils in England to collaborate. I calculated the distance between each pair of councils from their published latitude and longitude coordinates. I had to transform the coordinates to account for the different ‘sizes’ of degrees lat and long in England (that is, converting the lat and long coordinates to miles relative to a convenient origin). Then I used UCINET’s matrix operations to calculate the Euclidean distances between each pair of councils. (Perhaps the ‘tools’ command does this more conveniently – I didn’t try that.) We used the matrix of council-council partnerships as the dependent variable and the matrix of inter-council distances as an independent variable for MR-QAP (in fact we found that administrative boundaries were more important than simple proximity). We also looked at the distribution of distances between councils in various types of partnership. This paper gives some examples and references: Clio Andris (2016) ‘Integrating social network data into GISystems’ International Journal of Geographical Information Science, 30:10, pp.2009-2031 I'm not an expert on SNA, and hope this is useful! Ruth |
|
Re: Spherical Distances from GIS/GPS Coordinates
Hi Scott, I used distances (and UCINET) for this paper: Best, Jeremy
On Friday, August 9, 2019, 03:46:24 PM GMT+2, scott.thomas99@... [ucinet] wrote:
?
I noticed in UCINET Tools there is the possibility of producing a matrix of distances from latitudes and longitudes. I'm naive but curious -- how has that been used in network analysis and does anyone know of some references? Thanks. |
|
Re: 2-Mode Core/Periphery Data
I downloaded UCINET version 6.683, although it loaded version 6.682. I now can get core-periphery analysis for 2-mode although some of the headings are not printed right the start and final core-peripheries are there. I now notice in the UCINET program that when choosing 2-Mode Networks-->Core/Periphery it actually says "Categorical Core/Periphery" and I have been using a mixture of categorical (97%) with continuous normalized data (3%) and clearly that was a mistake although preliminary examination of the results mostly look reasonable. |
|
Residual plots for MRQAP
Hi all, Please, I wish to know if computing residual plots is necessary for regression in networks such as MRQAP and if yes, please what are the values I should use to compute the residual plots e.g standardized coefficients, unstandardized coefficients, p values or what values ? I intend to do this having read that R squared is not enough to indicate fitness of the model. Yours, Oscar |
|
Re: Board Interlock heterogenity
开云体育yes ? From: ucinet@...
Sent: Thursday, August 8, 2019 07:02 To: ucinet@... Subject: Re: [UCINET] Board Interlock heterogenity ? ? Dear Professor, ? Thank you for reply. Does that mean that the first matrices would be 2-mode employees x companies?? ? Kind Regards, Robert ? pt., 2 sie 2019 o 22:28?steve.borgatti@... [ucinet] <ucinet@...> napisa?(a):
|
|
Re: Board Interlock heterogenity
Dear Professor, Thank you for reply. Does that mean that the first matrices would be 2-mode employees x companies?? Kind Regards, Robert
|
|
Re: Output in Data ->Export -> Excel seems to be wrong
开云体育That’s annoying. I have half a mind to just output CSV files, which tend to be trouble-free. The option to choose xlsx type files relies on Microsoft’s COM tools, which always seem to have issues. ? Can you send me the data? Steve.borgatti@... ? steve ? From: ucinet@...
Sent: Sunday, August 4, 2019 18:38 To: ucinet@... Subject: [UCINET] Output in Data ->Export -> Excel seems to be wrong ? ? Hi there,? I do not know if it is something wrong or just the way it is. ? I was trying to export to excel seven joined matrices (Data ->Export -> Excel) When appears the options to save the files: 1) If I choose "Xls - Older excel type" the output excel files is correct. It will have seven matrices,? one in which tab. 2)If I choose "Xlsx- newer Excel type", the output excel files is incorrect.? It will have eight matrices; The first two matrices will be the same (with the same results, but with their own names) and it will appear a new one (plan 8). ? I have 2013 excel installed, maybe it is old for this function.? ? Thank you ? Vanessa ????????? ? |
|
Keyplayer
I am missing something when using keyplayer in UCINET. No matter how many nodes I select, 3, 5, 7, 10 keyplayer selects the first 3, 5, 7, or 10 individuals in the UCINET dataset (which are in alphabetical order). I am using version 1.45 and studying fragmentation. I though keyplayer would identify 3, 5, 7, etc names of the keyplayers that when removed would cause fragmentation. Any ideas of what I doing wrong would be appreciated. |
|
Re: CATEGORICAL AUTOCORRELATION: RCT analysis with 4 groups
开云体育Can you send me the data? Steve.borgatti@... ? From: ucinet@... <ucinet@...>
Sent: Friday, August 2, 2019 03:37 To: ucinet@... Subject: [UCINET] Re: CATEGORICAL AUTOCORRELATION: RCT analysis with 4 groups ? ? Thanks for the reply Steve.? I've managed to do a RTC analysis once - with a NodeAtt file and selecting the relavent column.? My problem seems to be that UCINET wants to convert my multi-category variable into four dummies, even when I import it as two columns and tell it to treat the data as NodeAtt (actor attribute [value]):? ?
? Thanks for your help with this. ? Best wishes, Mathew
|
|
Re: Board Interlock heterogenity
开云体育Suppose you have two matrices of 0s and 1s. ? W, where W(I,j) = 1 if person I works at company j S, where S(I,j) = 1 if person I sits on board of company j ? Then T = W’S is a company by company matrix in which T(j,k) gives the number of employees of company j that sit on the board of company k. In ucinet’s CLI, you can type this: ? ->T = prod(transp(w),s) ->dsp T ? steve ? From: ucinet@...
Sent: Thursday, August 1, 2019 8:15 AM To: ucinet@... Subject: [UCINET] Board Interlock heterogenity ? ? Hello, ? What would be the best method to create a directed board interlocks network, where the direction goes from company A to company B if they both share a director, AND the director serves as an executive at A and a regular member at B (according to Brennecke & Rank 2017 Tie heterogeneity in networks of interlocking directorates: a cost–benefit approach to firms’ tie choice). I have a database with companies and assigned to them directors together with their positions, the network would obviously be a matrix of companyXcompany relations. Thank you in advance. ? Kind Regards, Robert Magnuszewski |
|
Re: CATEGORICAL AUTOCORRELATION: RCT analysis with 4 groups
Thanks for the reply Steve.? I've managed to do a RTC analysis once - with a NodeAtt file and selecting the relavent column.? My problem seems to be that UCINET wants to convert my multi-category variable into four dummies, even when I import it as two columns and tell it to treat the data as NodeAtt (actor attribute [value]):?
 Thanks for your help with this. Best wishes, Mathew |
|
Board Interlock heterogenity
Hello, ? What would be the best method to create a directed board interlocks network, where the direction goes from company A to company B if they both share a director, AND the director serves as an executive at A and a regular member at B (according to Brennecke & Rank 2017 Tie heterogeneity in networks of interlocking directorates: a cost–benefit approach to firms’ tie choice). I have a database with companies and assigned to them directors together with their positions, the network would obviously be a matrix of companyXcompany relations. Thank you in advance. ? Kind Regards, Robert Magnuszewski |
|
Re: CATEGORICAL AUTOCORRELATION: RCT analysis with 4 groups
开云体育So, I have a network called campnet and an attribute dataset called campattr where the third column has 3 categories: ? ? ? ?????????????? 1 2 3 ???????????????G R C ???????????????e o o ???????????????n l m ???????????????d e b ???????????????e?? o ???????????????r???? ???????????????- - - ????1?? HOLLY? 1 1 1 ????2? BRAZEY? 1 1 1 ????3?? CAROL? 1 1 1 ????4???? PAM? 1 1 1 ????5???? PAT? 1 1 1 ????6? JENNIE? 1 1 1 ????7 PAULINE? 1 1 1 ????8???? ANN? 1 1 1 ????9 MICHAEL? 2 1 2 ???10??? BILL? 2 1 2 ???11???? LEE? 2 1 2 ???12???? DON? 2 1 2 ???13??? JOHN? 2 1 2 ???14?? HARRY? 2 1 2 ???15??? GERY? 2 2 3 ???16?? STEVE? 2 2 3 ???17??? BERT? 2 2 3 ???18??? RUSS? 2 2 3 ? When I run the RCT procedure I tell it that the attribute is “campattr col 3” and it gives the results below. I gather it’s not working that way for you. I’m guessing that your nodeattr dataset contains two columns of actual data (not one column plus row labels). If so, you should be telling it “nodeattr col 2” ? steve ? RELATIONAL CONTINGENCY TABLE ANALYSIS -- DIRECTED NETWORKS/UNDIRECTED MODEL -------------------------------------------------------------------------------- ? Network dataset:??????????????????????? campnet (C:\Users\Steve Borgatti\Dropbox\data\ucinet data\campnet) Attribute:????????????????????????????? campattr col 3 # of Permutations:????????????????????? 10000 Random seed:??????????????????????????? 10291 ? Input data is directed. Warning: Attribute vector has been recoded. ? Here is a translation table: ? ? Old Code??? New Code?? Frequency ? ========??? ========?? ========= ????? 1??? =>???? 1?????????? 8 ????? 2??? =>???? 2?????????? 6 ????? 3??? =>???? 3?????????? 4 ? Number of ties: 54.000 ? Cross-classified Frequencies ? ?????????? 1? 2? 3 ?????????? 1? 2? 3 ????????? -- -- -- ??? 1? 1? 20? 2? 2 ??? 2? 2?? 5? 9? 4 ??? 3? 3?? 0? 3? 9 ? ? Expected Values Under Model of Independence ? ???????????? 1??? 2??? 3 ???????????? 1??? 2??? 3 ????????? ---- ---- ---- ??? 1? 1? 9.88 8.47 5.65 ??? 2? 2? 8.47 5.29 4.24 ??? 3? 3? 5.65 4.24 2.12 ? ? Observed/Expected ? ???????????? 1??? 2??? 3 ???????????? 1??? 2??? 3 ????????? ---- ---- ---- ??? 1? 1? 2.02 0.24 0.35 ??? 2? 2? 0.59 1.70 0.94 ??? 3? 3? 0.00 0.71 4.25 ? ? ? Average permutation frequency table ? ?????????? 1??? 2??? 3 ??????? ---- ---- ---- ??? 1?? 9.88 8.48 5.63 ??? 2?? 8.46 5.33 4.21 ??? 3?? 5.64 4.23 2.14 ? ? Observed chisquare value = 50.061 Significance = 0.000200 Number of iterations = 10000 ? Cross-classified observed frequencies saved as lltab Expected values saved as llexp Ratio of Observed to Expected values saved as llrat ? ---------------------------------------- Running time:? 00:00:01 Output generated:? 31 Jul 19 16:09:05 UCINET 6.682 Copyright (c) 2002-19 Analytic Technologies ? ? ? ? From: ucinet@...
Sent: Tuesday, July 30, 2019 23:56 To: ucinet@... Subject: [UCINET] CATEGORICAL AUTOCORRELATION: RCT analysis with 4 groups ? ? Dear UCINET group,? ? I am working on a relational contingency table analysis looking at the?associations between four groups.? ? One type of node (Y), split into two groups based on their position in the network (core and periphery) and another two groups based on attribute X and non-X.? These groups are mutually exclusive.? I want to test if the association from Y to X is consistent from the core to the margin of the network or if there are different patterns of?association based on Y's position in the network.? ? I created a NodeAtt file with two columns, the ID and the attribute column, recoding each node as either 1, 2,3,4 (i.e. core, periphery, X, non-X).? ? ? Problem: UCINET automatically turns each category of the variable into 4 separate dummy variables and only considers the first dummy variable in the contingency table.? ? How do I get UCINET to consider all the categories for the RCT analysis? And secondly, am I going about this in the right way, in choosing RCT analysis for this type of question?? ? Thanks for all your help UCINET group! ? Cheers, Mathew |
|
Re: 2-Mode Core/Periphery Data
toggle quoted message
Show quoted text
From: ucinet@...
Sent: Tuesday, July 30, 2019 22:57 To: ucinet@... Subject: [UCINET] 2-Mode Core/Periphery Data ? ? Can one use a mixture of continuous data with binary data in a 2-mode core-periphery analysis? UCINET crashes when I do that but if I remove the continuous data (all ranged from 0-1) UCINET is fine. However, I do not want to abandon those continuous data. Any suggestions would be appreciated. |