Note: To find revision information for really old versions of UCINET look in these two places:
- To see a list of known bugs and user wish list items, check the fix list
- To download the latest version, click download.
- Changed netdraw button on toolbar to immediately ask for a file name before starting netdraw
- Added cutpoints routine to CLI (aka matrix algebra). Given a network, it outputs a vector V in which v(i) = 1 if node i is a cutpoint. Unlike menu version, works for both undirected and directed graphs. Syntax and example are as follows:
-><outfile> = cutpoints(<dataset>)
->cp = cut(campnet)
- Fixed bug in Double Dekker QAP regression. When coefficients were sufficiently small, the program was setting them to zero. As a result, when it compared the observed coefficient (zero) with the coefficient from a random permutation (also zero), it was always finding that the permutation value was as large as the observed, yielding a p-value of 1.0.
- Fixed bug in Transform|Symmetrize which was only symmetrizing the first matrix in a dataset
- Yet another install package for the 64-bit version -- previously, after installation the program would not start up
- Change in Networks|Cohesion|Alters in Common so that valued data would automatically be dichotomized. In addition, missing values are now ignored
- Bug fixed in Transform|Aggregate|CSS. Previously, the labels Pa and pA were reversed. Now, the value labeled Pa correctly gives the number of ties Perceived but not Actually there, while pA gives the number of Actual ties that the respondent failed to Perceive
- Uploaded new install package for the 64-bit version
- Added option to Gould & Fernandez brokerage to allow closed triads to count as brokerage instances as well as open triads
- Reverted minres routine in continuous core-periphery procedure to old version that seems to handle negative values and non-symmetric data better
- Minor change in Transform|Aggregate|CSS to make default output file names match the method of creating the "true" network
- Fixed bug in Networks|Cohesion|Density|Density by Groups which was causing crashes when the user selected the Identity partition instead of a categorical attribute
- Fixed another bug in Networks|Cohesion|Density|Density by Groups which was failing to retain node labels when Identity partition was used
- Added CLI command for aggregating columns of a matrix, much like Density by Groups. It is called AggCol and is used like this:
->gender = col(campattr "gender")
->dsp aggcol(campnet gender)
The result, in this case, would be counts of the number of male and female alters that each node has:
- Added routine in CLI to calculate Goodman and Kruskal Gamma coefficients (measures of rank-order similarity) on columns of a matrix. Example usage is:
->coefs = ordinalassoc(campnet)
- Added att2mat command to CLI. This converts an attribute to a matrix. The syntax is <outfile> = att2mat(<vector> [<method>]) where the choices for method are: 'match', 'diff', 'absdiff', 'sqrdiff', 'prod', 'sum', 'identity', 'rows', and 'cols'. The default is 'match'. Example:
->gen = col(campattr gender)
->samegender = att2mat(gen match)
- Added mat2vec command to CLI. This vectorizes a matrix so that it looks like one long column. If the input dataset contains multiple relations, then each becomes a column in the vectorized dataset. The syntax is <outfile> = mat2vec(<dataset> [<method>] [diagok]) where the choices for method are: 'whole', 'lowerhalf' and 'upperhalf'. The default is 'whole'. By default, the diagonal is omitted, but if want the diagonal included, add 'diagok' to the list of parameters. Example:
->pvec = mat2vec(padgett lowerhalf)
->svec = mat2vec(sampson whole diagok)
- Fixed bug in Matrix editor. It had been crashing when the user saved a dataset after having cleared the editor using the New button
- Fixed bug in Transform|Symmetrize in which it was failing to heed the user's choice of method (for the methods lower down in the list).
- Fixed bug in Tools|Univariate which was failing to save the output statistics for all but the first matrix in a dataset.
- Fixed bug in Tools|Hypothesis Testing|Node level|Regression which was putting in a nonsense number for the standardized beta coefficient of the intercept (which is supposed to be 0). Nothing else was affected by this.
- [64-bit version only] Have added warnings to most (but not yet all) routines which do run in 64-bit mode. This is better than simply crashing ...
- Fixed bug in Network|2-mode|Categorical Core/Periphery Model that was causing crashes. Note: this is a routine that does not run in the 64-bit version.