This page shows changes to UCINET between versions. Just scroll down to see release notes in reverse chronological order.
Note: To find revision information for really old versions of UCINET look in these two places:
Related Links
  • To see a list of known bugs and user wish list items, check the fix list
  • To download the latest version, click download


  • Two-day workshop on social network analysis using UCINET WHOTaught by Steve Borgatti under the auspices of Statistical HorizonsWHENFriday, April 27, 2018 9:00 AM –Saturday, April 28, 2018 5:00 PM (Eastern Time)WHERETemple ...
    Posted Jan 14, 2018, 12:35 PM by Steve Borgatti
Showing posts 1 - 1 of 12. View more »

Version 6.649 | 6 Mar 2018

posted Mar 6, 2018, 6:17 PM by Steve Borgatti   [ updated Mar 7, 2018, 8:06 AM ]

  • Added feature to the DL Editor. Suppose you have 1-mode data drawn from a roster-based survey, but not everyone fills out the survey. Now you set an option that fills in the missing rows with NAs, so you can distinguish between who filled out the survey (but had no outdegree) and who simply didn't fill out the survey. 
    • A less automatic way of doing this already exists in UCINET under Transform|Give non-responders missing rows. This older procedure requires you to have node-level variable in which 1 means the person is a respondent and 0 means they are not.

Version 6.648 | 12 Feb 2018

posted Feb 12, 2018, 6:43 PM by Steve Borgatti

  • Fixed bug in Transform|Impute missing values -- was failing to produce an output file
  • Fixed bug in Data|Import text file|raw matrix. Was not guessing number of rows and columns very well

Version 6.647 | 19 Dec 2017

posted Dec 19, 2017, 1:08 PM by Steve Borgatti   [ updated Dec 19, 2017, 1:09 PM ]

  • Changed DL editor so that it automatically displays your saved data in matrix form. 
  • Added feature to y-permutation QAP regression that enables a Preacher and Hayes' style mediation test

Version 6.646 | 12 Dec 2017

posted Dec 12, 2017, 12:23 PM by Steve Borgatti   [ updated Jan 3, 2018, 7:32 AM ]

  • Added Louvain method of clustering (aka community detection). The menu version can be found at Network|Subgroups|Louvain. The command line version has the following syntax: louvain(<network dataset> [<starting partition>]). For example:
    ->x = louv(campnet)
    ->dsp x
    ->draw campnet x   //opens campnet in netdraw and colors node by the first partition stored in x
    Note: the CLI version does NOT automatically symmetrize. The menu version offers various options on this. 
  • Added small world statistic to the CLI. Syntax is: Smallworld(<dataset>), as in:
    ->dsp smallworld(campnet)

Version 6.644 | 19 Nov 2017

posted Nov 19, 2017, 2:38 PM by Steve Borgatti

  • Fixed bug in the various 'match datasets' routines that was causing crashes when you ran the routine a second time.
  • Changed default p-values in node-level regression routine. Previously, the default was classical significance. Now the default is permutation-based (which prints the classical test as well).

Version 6.643 | 8 Nov 2017

posted Nov 8, 2017, 9:16 AM by Steve Borgatti

  • Added ability to export ego networks to an Excel file. See Data|Export|Export egonets to Excel

Version 6.642 | 4 Nov 2017

posted Nov 4, 2017, 1:54 PM by Steve Borgatti   [ updated Nov 4, 2017, 1:55 PM ]

  • Fixed bug in Tools|Consensus analysis. When using the rectangular matrices option, the program was producing the wrong answer key matrix.

Version 6.640 | 5 Oct 2017

posted Oct 5, 2017, 1:34 PM by Steve Borgatti   [ updated Oct 5, 2017, 1:35 PM ]

  • Added beta test version of new routine for imputing network ties. Find it under Transform|Znidarsic et al Imputation of Ties. The methods are drawn from 

    Anja Znidarsic, Patrick Doreian, and Anuska Ferligoj. 2017. Treating missing network data before partitioning. In P. Doreian, V. Batagelj, A. Ferligoj Advances in Network Clustering and Blockmodeling, June 22, 2017 12 : 04. John Wiley

Version 6.639 | 2 Oct 2017

posted Oct 2, 2017, 6:15 PM by Steve Borgatti

  • Upgraded Loadps (load pilesorts) facility in the CLI to handle non-numeric respondent IDs

Version 6.638 | 27 Sept 2017

posted Sep 27, 2017, 6:01 PM by Steve Borgatti   [ updated Sep 28, 2017, 6:49 AM ]

  • Fixed important bug in DL editor. When using the full matrix format and no column labels, the program was shifting the data up one row, leaving the last row blank
  • Update: fixed another bug in the DL editor. When reading in UCINET files in edgelist and nodelist formats, it was not reading all relations. Note that for fullmatrix it still only reads the first relation

1-10 of 229