olderversions

Description

• • Fixed bug in new concentration measures, specifically, the densities of cores and peripheries

  • Added browse buttons to new Transform|CSS routine. This routine allows you to have missing matrices, but only if you have labels for the matrices that match the labels for the nodes

  • Added new "concentration" measures to core/periphery output. Also made coreness scores print in sorted order

  • Fixed bug in networks|egonets|composition|continuous that was printing the wrong values for "sum" (the sum of attribute values of the alters of an ego)

  • Fixed bug in Networks|Triad Census producing "Abstract Error" error message.

  • Added panel on right side that you can drag previous output window to dock to

  • Fixed bug in Tools|Clustering|Moca causing slightly off values in the calculations of Cohen's Kappa and Freeman's S for partitions with more than two classes

  • Made changes in Data|Import|VNA to allow import of unicode text files

  • Added new routine Networks|Ego Networks|Egonet Density, which accepts valued data

  • Fixed bugs in Tools|Frequencies. (1) implemented sorting of output tables by values and frequencies. (2) fixed proportions calculation so that it now applies to the dimension being selected. e.g., if you elect to get frequencies by rows, the proportions of proportions of each row, not the matrix as a whole

  • Fixed bug in Networks|Centrality|Beta Centrality. Was giving wrong answers when the iterated/approximate method was chosen

  • Added separate Get Diagonal and Set Diagonal routines to Transform|Diagonal

  • Changed registration routines to facilitate registering program in settings where user does not have full priveleges

  • Fixed Networks|Subgroups|Girvan-Newman to match NetDraw algorithm. Previously, when two ties had equal betweenness, one was selected for deletion and then betweenness recalculated. Now, both ties are removed before recalculation. Also made changes to minimize round-off error

  • Fixed recently introduced bug in matrix algebra which made it difficult to type complex commands

  • Fixed bug in current directory combo box which was showing blank and also making it difficult to type in new address

  • Fixed bug in matrix algebra|corr function which was treating missing values as valid values

  • Fixed Data|Remove|Isolates routine which was stopping with strange error message ("you should not be seeing this")

  • Fixed Data|IMport Text file|Multiple Dl files which was having trouble with labels due changes in last few months toward accepting unicode

  • Changed random number algorithm in Data|Random|Erdos-Renyi routine to yield higher-quality (but more slowly produced values)

  • Fixed Girvan-Newman routine to correct nonsense dendrogram

  • Changed default column widths in outputs from -1 to 0, which means "auto"

  • Fixed Networks|Subgroups|Factions routine so that modularity optimization works properly

  • Added option to Tools|Cluster|Hierarchical to allow Newman community detection algorithm

  • Changed Tools|Similarities routine so that choosing AVG Cross Products gets you just that, rather than the projected cross products you would have gotten if you hadn't had any missing data (i.e., avg cross products multipled by N).

  • Enabled Fill routine in spreadsheet to determine boundaries of area to fill by the current location of the cursor. So you can point to a cell, press Fill, and fill all cells left and up from that spot with zeros

  • Fixed bug in Tools|Correlate columns across datasets routine causing null output

  • Added option for centering attributes in continuous autocorrelation routine (Moran's I / Geary's C)

  • Made changes in Display function attached to the D icon on the main toolbar, in hopes of fixing problem with Windows XP machines which was resulting in a 'division by 0' error.

  • Changed installation package to include more recent NetDraw

  • Fixed bug introduced in 6.337 that was causing Networks|Cohesion|Density|Density by Groups to operate on the attribute one column to the left of the one chosen by the user

  • Fixed bug in Networks|Regions|Bi-components causing crashes on exit (data were still being produced correctly)

  • Fixed bug in Networks|Extras|PLI causing it to fail to produce output.

  • Fixed bug in Transform|Recode causing it to fail to recode certain cells when user selected only certain rows and columns to be recoded

  • Fixed bug in Tools|Test Hypotheses|Mixed dyadic-nodal|Categorical|Relational Contingency tables introduced 3 days ago that was screwing up expected values. Also added ability to force symmetric data to be treated as directed

  • Fixed Moran-Geary procedure so that it now puts "quotes" around filenames

  • Fixed bug in Networks|Core-Periphery|Categorical causing starting partition to be very badly chosen

  • Fixed bug in Tools|Testing hypotheses|Mixed node/dyadic|rct which was reporting wrong value for number of ties

  • Changed Transform|Dichotomize routine to avoid "." in output filenames

  • Fixed bug in Data|Extract|SubMatrix causing output matrix to have incorrect labels

  • Modernized Data|Permute routine

  • Fixed bug introduced recently causing older procedures that combine a network with an attribute to fail

  • Just changes in an experimental routine -- not for general use

  • Fixed recent bug in DL reading routines (having to do with unicode)

  • Replaced routines for reading and writing excel files in the matrix Spreadsheet with faster versions

  • updated help file

  • Fixed bug in new option Options|Unicode labels so that now it defaults to unicode

  • Additional support for unicode labels to ensure that datasets created by ucinet as part of an analysis (e.g., "cliquepart" datasets created by cliques procedure)

  • Fixed bug in Tools|Scaling|Eigenvectors that was causing non-convergence. This also fixed problem in GetBeta button of Networks|Centrality|Bonacich Power/Beta Centrality

  • Fixed bug in reading labels from old ucinet files such as Wiring

  • Fixed bug in Tools|Scaling|Correspondence analysis that was failing in the case of non-square matrices

  • Added some support for unicode data labels. All data entered via the spreadsheets can now have special characters. At present data entered via VNA or DL text files does not accept unicode, but this will change shortly

  • Added new routine Data|Export|Raw (new) for exporting raw data matrices to text files

  • Added Save As Multiple Files facility to the matrix spreadsheet. Now you can import an Excel file with multiple tabs of different sizes and shapes, and save each tab to a different ucinet file

  • Fixed bug in Data|Import Text File|Pajek which was causing crashes when importing 2-mode data

  • Fixed bug in Tools|Clustering|Cluster Adequacy. When multiple partitions were entered, the cluster sizes were being reported incorrectly for all but the first partition.

  • Added new Networks|Cohesion|Geodesic Distances routine that is capable of handling multiple matrices within a dataset

  • Changed Transform|Reverse routine to have more options

  • Added options to Transform|Dichotomize routine

  • Added Reverse operation to spreadsheet editor

  • Fixed bug in Tools|Correlate columns across datasets routine that was creating empty columns

  • Fixed bug in Networks|2-Mode|Centrality causing eigenvector scores to be blank when the input dataset had more columns than rows

  • Fixed bug in Data|Remove|Isolates in which nodes with reflexive ties were not considered isolates

  • Fixed bug in Data|Extract|Subgraph which was sometimes failing to produce output

  • Added options to Transform|Union routine

  • Added different models to the experimental power routine Networks|Extras|Political Independence

  • Added option to Networks|Egonets|Composition|Continous to allow filtering out alters with certain attribute values

  • Same program, but updated installation package. Hope to solve this spate of “The procedure entry point wcsncpy_s could not be located in the dynamic link library msvcrt.dll” errors.

  • Added Freeman Segregation measure to Tools|Cluster|Cluster Adequacy

  • re-uploaded current version to see if that cures installation problem.

  • Changed Networks|Egonets|Longitudinal|Brokerage Elasticity routine to add additional measures, namely the number of holes at T1 and T2, and these same figures expressed as a percentage of the possible, given the number of ties in the ego net at each time point.

  • Changed Networks|Subgroups|Girvan-Newman to automatically symmetrize the input dataset

  • Fixed bug in Networks|Centrality|Hubs and Authorities routine. Was sometimes returning the singular vectors associated with singular values other than largest.

  • Fixed major bug introduced just a few versions ago that was putting grey non-editable regions in the dl editor

  • Fixed major bug introduced just a few versions ago that was preventing the DL editor from starting up

  • Added experimental new Networks|Cohesion|Transivity routine

  • Fixed bug in Transitivity:Strong routine

  • Added new specialty routine Transform|Egotize. Takes a network and a dyadic matrix X and creates N versions of X, one for each ego, with values only for alters of ego

  • Fixed bug in Beta Centrality routine where, on pressing GetBeta button, you didn't necessarily get the largest eigenvalue

  • Fixed bug in installation program causing program to look for missing file mia.lib

  • Experimental version UCINET recompiled with new compiler

  • New installation program. You should probably uninstall your old version of UCINET

  • added option to networks|egonets|composition|continuous. Can now ignore values of valued ties

  • Added routine Networks|Extras|Colleen's Multiple CP. It runs continuous core/periphery model for multiple datasets

  • Some options in Networks|Regions|Components|Valued disabled until they can be tested

  • Ported NetDraw's Girvan-Newman routine into Ucinet (Networks|Subgroups)

  • Fixed bugs in Networks|Regions|Components|Valued which was giving wacky partitions

  • Added new experimental routine Networks|Trajectories that does simply analyses of trajectory data

  • Fixed bug introduced in 6.258 in the continuous core/periphery routine

  • Updated Transform|Multiplex Coding procedure to print a legend and allow larger datasets

  • Updated Networks|Centrality|Fragmentation routine to give choice of two methods, and provide some explanation of outputs

  • Added a number of new helpfiles

  • Fixed bug in networks|centrality|degree. It was calculating centralization incorrectly when data were directed and included self-loops

  • Fixed bug in networks|roles & positions|max regular|catrege which was causing random answers when 'Convert to geodesics' option was on

• Updated installer to include more recent version of Netdraw

• Fixed bug in Data|Remove|Pendants that was causing output matrix to have duplicate rows

• Fixed bug in Networks|Regions|Components|Simple which was saving a components-by-actors matrix instead of an actors-by-components matrix.

• Fixed bug in Data|Export|Pajek|Network which was causing empty output files. Another product of the Great Bug of .258

• Fixed major bug in Networks|Ego Networks|Longitudinal|Brokerage elasticity that was causing nonsense answers.

• Changed output of Networks|Ego Networks|Longitudinal|Egonet Change to generate more useful measures

• Fixed bug in Data|Remove|Isolates that was preventing dropping of any isolates

• Changed Networks|Egonets|Structural Holes procedure to output dyadic redundancy and constraint for both whole netowrk and ego network models. Also, ego network model runs faster now. Also eliminated option to dichotomize data

• Changed interactive Concor procedure to output node labels in addition to node numbers

• Fixed bug in EI index dialogue window causing it to lose the initial quotation mark in a filename

• Changed LBK procedure to disallow directed data

• Fixed bug in Data|Extract|Subgraph by Attribute vector. This was a bug introduced in 6.258 that caused the program to always output the complete graph

• Fixed bug in Tools|Cluster|optimization which was causing the program to exit before saving or printing any results. Also introduced in 6.258

• Changed Data|Export|Excel to permit export of large datasets (previously, due to limitations in Excel, the program was limited to 255 columns. now that Excel can handle more columns, this restriction has been lifted. Note: for large datasets, this function is better than cut-and-paste from UCINET spreadsheet.

• Bug fixed in Data|Join which was causing crashes when joining by rows

• Fixed bug in double dekker qap regression which was giving wrong significance when the X variables predicted Y perfectly (as might occur in a hypothetical dataset)

• Added file browsing to Networks|Balance Counter attribute dialogue

• Bug fixed in Transform|Block which was causing the program to exit without producing output. Part of the Great Bug of 258.

• Fixed bug in Data|Display which was causing first row/col of a matrix to be repeated when user specified a range, such as "1 to 6"

• Fixed bug in Data|Import text file|Multiple dl files that was causing crashes. Part of the Great ... oh never mind.

• Bug fixed in Transform|Matrix Ops|Within|Aggregations that was causing crashes. Part of the Great Bug of 258.

• Bug fixed in non-valued pilesorts format in the DL editor

• Bug fixed in the calculation of the Jaccard coefficient in Tools|Testing Hypotheses|Dyadic|Relational Cross-tabs

• the cross tabs routine was also modified to include the observed data as one of the permutations in the significance tests

• Same version re-uploaded

• Fixed important bug in Networks|egonet|egonet change routine, which was giving wrong answers for the first node in any network

• fixed bug in valued pilesorts format in the DL editor

• Added to help file

• added Quick Start guide

• More fixing of the Great Bug of 6.258. This fix should eliminate the "Strange error ... " message that was popping up.

• More fixing of the Great Bug of 6.258. Should eliminate crashes in multiple centrlaity measures (old), closeness centrlaity, Bernoulli randoms, and more.

• Fixed bug from 6.258 causing crashes when user tried to open a non-existent file.

• added new data format to DL editor. Can now read pilesort type of data where each record is a set of items placed in the same pile together

• Fixed bug in Data|Describe that was causing crashes on using it a second time in a session

• Fixed bug in Networks|Triad Census causing problems with undirected graphs.

• Fixed bug in Data|CSS related to systemic bug of version 6.259. It was causing crashes

• Systemic bug introduced after version 6.258 hopefully removed

• Added routine to compare changes in ego network triples over time, with attributes of alters and ego

• Systemic bug introduced after version 6.258 hopefully removed

• Added routine to compare changes in ego network triples over time, with attributes of alters and ego

• Added Tools|Correlate Columns Across Datasets. Similar to Tools|Similarities, but allows correlating columns belonging to different datasets

• Added matrix algebra button to toolbar

  • Added ability to re-run most recent command by pressing Alt-Backspace. You can also select from a list of recent commands, stored in a drop-down list on the menu bar.

    • • Note: not all commands can be re-run yet.

• Some additional measures added to total centrality decomposition

• Added new routine for decomposing a node's "total centrality" into its direct or "endogenous" contribution, and its indirect or "exogenous" contribution.

• Fixed bug in Networks|Centrality|Fragmentation. Was not doing any calculations unless you explicitly set the method -- i.e., was ignoring the default setting.

• Added additional help topics

• Added option to ignore isolates in the computation of group degree centrality

• Changed Networks|Ego Networks|Basic measures so that the normalized brokerage measure is now doubled

• Removed Alpha option in Networks|Subgroups|Factions routine, and replaced with Modularity. Also removed errant diagonal option

• Replaced old group centrality routines with one that will eventually be more general.

• Modified Networks|Centrality|LBK procedure to allow diagnostic output and to limit length of paths

• Made fixes in the experimental LBK power routine

• Also made a fix in the matrix spreadsheet to try to get it to obey the "symmetric" sign

• Changed Networks|Cohesion|Homophily procedure to handle valued data more fully. Previously, only the Corr measure made use of tie strengths. Now, the E-I index does as well (instead of E and I being the number of ties between and within groups, respectively, they are now defined as the sum of tie strengths for ties between and within groups

• Fixed bug in Networks|Centrality|Multiple Measures that was causing intermittent crashs.

• No changes -- just new install package incorporating new version of netdraw

• Added some help files for various newer routines

• Fixed bug in Tools|Hypothesis Testing|Node Level|T-Test in which files selected via browsing tool were not surrounded by full quotes

• Improved output on Networks|Cohesion|Geodesic Cube and eliminated useless option

• Fixed bug in Networks|Cohesion|Simmelian ties that was producing valued output even when binary output was specifically requested

• Fixed bug in Transform|Matrix Ops|Between|Boolean Combinations. It was not closing output file properly, causing the program to see the file as "busy" when you tried to do any analysis on it

• Fixed bugs in dialogue window of Networks|Cohesion|Geodesic Cube. Browsing buttons were not working, and an inappropriate item was appearing as an output option. Also changed the default suffix for output file

• Fixed bug in Networks|Cohesion|Density by Groups which was preventing program from saving a file containing within-group densities

• Fixed bug in Networks|Cohesion|Simmelian ties that was producing wrong answers when "weighted" and "outgoing" options were chosen together

• Removed some unnecessary output from the Networks|Cohesion|Point Connectivity routine

• Fixed important bug in new routine Networks|Cohesion|K-local bridges. Was giving wrong answers. Oy.

• Fixed internal bug in a data structure, causing intermittent failures.

• Fixed bug in k-local bridges routine, causing lock-up on some datasets

• Note: observed similar bug in Networks|Centrality|Multiple measures (new). This has not been fixed yet.

• Added k-local bridges routine under Networks|Cohesion. For each tie linking A to B, it gives the distance from A to B that would result if the tie were removed

• Fixed bug in 2-mode centrality routine that was sometimes preventing output of eigenvector scores

• Added new 2-mode routine for calculating basic cohesion measures.

• Fixed important bug in creating multi-relational files. E.g., when dichotomizing a multirelational dataset like Sampson, a new multirelational dataset is created.

• Added 2-mode eigenvector output to Networks|2-mode|Centrality

• Changed normalization of output in tools|non-sym eigenvectors routine (only affects real-valued eigenvectors)

• Fixed bug in Data|Export|Pajek|Quantitative Attribute that was preventing user from choosing output file names

• Added standard errors of coefficients for the double-dekker qap regression program.

• Fixed major bug (probably introduced in version 6.232) which was causing most procedures to read files containing multiple matrices to screw up and just read the first matrix.

• Fixed bug in Data|DL Editor that prevented distinguishing between missing data and the absence of a relationship

• Added new routine Data|Sort Alphabetically. Allows you to sort rows/columns/levels of a dataset based on labels of the rows/columns/levels

• Added new function to spreadsheet editor. Under "file" in the spreadsheet menu, there is an entry called "insert columns from file". This lets you read in a new dataset and join it to the dataset you already looking at. This is handy for continually adding new variables to a dataset, say, one called "attribs".

• Added new function to matrix algebra package. Called "recode", it is used to recode values of a matrix. Syntax is as follows:

• -><outfile> = recode(<infile> <oldvalue> <newvalue>)

• for example, this changes all zeros to missing values in a dataset called campnet:

• ->campnet2 = recode(campnet 0 missing)

• Fixed bug in DL Editor which was causing crashes when saving files

• Made some minor improvements to matrix algebra command window

• fixed bug causing Networks|Egonet|Egonet Change routine to crash

• Added new routine to compute multiple types of centrality. Find it at Networks|Centrality|Multiple measures (New)

• Added new interactive CONCOR routine to Networks|Roles & Positions|Structural|Concor

  • Added Markov Clustering algorithm to Networks|Subgroups. I was skeptical but it works really well. Does a great job of telling you how many clusters there are.

    • Implemented both a regular and sparse matrix version. They get different answers for large datasets -- presumably a rounding issue

• Replaced Tools|Scaling|Eigenvectors routine with a new one. The old would only produce first eigenvector. The new produces all, and handles non-symmetric matrices (which may result in complex eigenvalues and vectors).

• New routine under Networks|Egonets|Brokerage Elasticity. The dyadic output is meant to show what would be the effect on ego's brokerage score of adding/removing each possible tie from an ego network. Node level output shows the net effect on brokerage of each tie that was in fact added/dropped at T2

• A few fixes in Networks|Centrality|Alpha Centrality (Bonacich Power)

  • Changed "get alpha" routine to double precision to avoid rounding error

  • Added warnings when diagonal values are not constant or not zero

  • Changed the way row sums are calculated in the routine so as to include the diagonal

• Added Networks|Egonets|Change routine. Given network at two time periods (stored in a single dataset with two matrices), it counts number of ties gained, lost, kept etc. and records whether contacts in each of these categories had ties with the alters in the T1 ego network.

  • Fixed GINV to not require square matrices.

• Fixed bug in Networks|Egonets|Honest Broker|Longitudinal so that both output matrices are now being saved.

• Added longitudinal honest broker routine

• Small changes in the registration routine

• Changed name of tools|similarities routine slightly to make it clearer

  • Added experimental power routine to Networks|Centrality|LBK power.

• Updated excel import to include .xslx files (Office 2007)

• Fixed bug causing problems when importing excel files that contained empty sheets

• added clarifying language to registration dialogue

  • New registration system installed.

• Program will now ask for User id. If you don't have one, just enter your name.

• Added additional data format to the set that Data|Import via spreadsheet|DL type formats can handle. The new format is a 2-mode, 3-way edgelist

• Fixed bug in Networks|Egonets|Structural Holes which was causing the calculation of ego-betweenness to blow up.

• Changed the way browse buttons deliver filename. Now, if the file's filepath matches the current folder, the program presents only the filename. But if the user had browsed away from current folder, then it retains full pathname.

• Fixed bug in the writing of filenames in the logfile. It had been repeating part of the filename

• Fixed bug in Networks|Ego Networks|Structural Holes that was causing program to run out of memory with networks of around 12K nodes. Now it runs larger networks quite fast. Also fixed bug preventing egobetweenness output in the same program

• Fixed bug in Options|Helper Applications, preventing retention of the setting for the text editor across ucinet sessions.

  • Fixed another bug in Helper Applications that was losing the location of Mage and Pajek

• Fixed bug in Visualization menu that prevented Mage and Pajek from running.

• Added new routine Networks|Triad Census. This counts the frequency of all 16 possible triads in a digraph

• Added new routine Data|Extract|Subgraph Via Attribute Vector. This provides a convenient way to extract a sub-network based on a node property, such gender = 1, or betweenness > 0.

• Fixed bug in new routine Networks|Ego Networks|CSS Brokerage. When the user chose to define ego networks in terms of both the "true" and the perceived networks, the program was incorrectly assuming that there were ego-alter ties in both networks

• Added Cohen's Kappa for measuring intercoder reliability to Tools|Similarities routine

• New routine Networks|Ego Networks|CSS Brokerage. Given both a CSS and a "true" network, counts number of structural holes that exist in both true and perceived networks, just true network, just perceived network, and so on. Concept developed by Scott Soltis

• Modified Data|CSS routine to add appropriate suffix to output file name

• UCINET unchanged, but the install package contains updated NetDraw

• Fixed bug in tools|dissimilarities|gps/gis. Now the browse buttons work

• Fixed important bug in networks|cohesion|distance which was getting the wrong number for average distance when analyzing large matrices

• Changed Data|Export|DL to allow pagewidth to exceed 32767 (and changed the default width to 1,000,000 effectively preventing the output file from having multiple lines per record)

• Changed default tie strength from 0 to 0.0001

• Fixed bug in main toolbar that disabled the Netdraw button

• Fixed bug in networks|ego networks|structural holes procedure which was sometimes causing it to refuse to run.

• Added button to toolbar for importing text via spreadsheet interface

• Fixed important bug in the Data|Import via spreadsheet|DL type formats procedure. It had been ignoring the "force symmetry" option. No longer.

• Change in the Options|Register procedure to allow users without administrator rights to register the program

• No change in program itself, just a change in the installation procedure. Should enable administrators to set up UCINET for users

• Fixed bug in Networks|Ego Networks|Structural Holes // Restricted Model procedure. It was giving a floating point error for egos with only one alter.

• Fixed bug in handling of scratch folders causing "Can't open file xxxx\output.log" error. Problem was fixable by user by going to options|scratch folder and changing to any folder user has permission to write to

• Fixed bug in Data|Import Excel|Multiple DL which was giving errors when inputting edgelist data with only 2 columns.

  • Replaced textual output (logfile) browser with user's preferred text editor (which defaults to the Windows Notepad).

• Changed default scratch folder (where logfiles are written) to the Windows TEMP directory. This should make things smoother for people who don't have write access to their Program Files folder

• Added new data import procedure Data|Import Excel|Multiple DL type formats. It allows you to import Excel files that are formatted as nodelists, edgelists, etc. You can also simply cut and paste or enter data directly into a spreadsheet.

• Added 2-mode centrality measures: Networks|Centrality|2-mode

• Numerous small fixes

• Changed DATA|JOIN procedure to allow resizing window and to remove distracting ".##h" from file names.

• Changed Networks|Cohesion|Reciprocity to add two new columns of measures, as well as correct the documentation printed in the log file

• ALT-Backspace key now sends you to the last run routine. At least, most of the time.

• Now Options|Helper Applications includes an entry for specifying what text editor to use (e.g., when the user presses the "pencil" button on the toolbar). The default is Windows Notepad

• Fixed important bug in Data|Import|DL routine which was failing to import FULLMATRIX files that had no labels

• Added eigenvector routine for directed data

• changed name of Bonacich Power routine to Alpha Centrality / Bonacich Power

• Fixed bug in dialog window for Networks|cohesion|homophily

• added

  • Fixed bug in Networks|Centrality|Closeness in the reversed distances (Valente & Foreman) routine which was giving incorrect group centralization scores.

• No change, but someone reported not being able to download the program

• Fixed bug in Networks|Cohesion|Density|Density by Attributes which was not allowing choosing different attributes.

• Fixed bug in Networks|Cohesion|Density|Overall Density which did not have browse buttons implemented.

• Fixed bug in Transform|Union, which was allowing long filenames to wrap, causing the program to misunderstand filenames.

• Fixed bug in Networks|Cohesion|Density|by Attributes that was causing a crash after finishing all calculations

• Finished changes to Networks|Cohesion|Krackhardt GTD routine so that all measures are appropriate for disconnected graphs.

• Changed Networks|Cohesion|Krackhardt GTD routine so that LUB measure works with multiple components. MOre to be done here

• Added routines under Data for matching datasets via node labels. For example, if you have a network dataset and a corresponding attribute dataset, but the rows of the attribute dataset are in a different order than the rows/columns of the network dataset, you can run either Data|Match Network and Attributes or Data|Match Multiple Datasets to sort them both in the same order.

• Data|Match Multiple Datasets can also be used to sort a single dataset into alphabetical order based on node label

• Changed text area in double-dekker mrqap procedure to stop wrapping filenames

• Added ability to enter "+" in default folder field at bottom of main window to indicate your "my documents" folder. For example, if you have a folder called Data under My Documents, you could type this:

+\data

instead of having to type something like this:

C:\Documents and Settings\Username\My Documents\data

This works in Vista as well.

• Changed help viewer to html-based viewer that works with Vista.

• Fixed bug in Transform|Matrix Ops|Within Datasets|Cell Transformations that was saving the output ##d file in a different folder from the associated ##h file

• Changed algorithm used to calculate eigenvalue in the GETBETA function of Networks|Centrality|Power procedure. The old algorithm failed when a matrix had multiple eigenvalues with the same absolute value. The new algorithm always works.

  • Fixed bug in Networks|Regions|K-core routine which was giving wrong values for coreness (it was delivering 1+ the correct value in most cases).

• Added feature to Networks|Egonet|Structural Holes to enable defining ego network according to incoming, outgoing or both kinds of ties. Also fixed bug causing Efficiency scores to not be missing values when degree was zero.

• Improved dialog box for selecting default folder

• Fixed bug in the way UCINET opened NetDraw

• Added procedure Networks|Ego Networks|Concentration. Given a network and a 1/0 attribute vector, the procedure calculates, for each ego, the number of pairs of alters that (a) are tied and (b) involve a given type of alter (based on the attribute vector). For example, if men are coded 1 and women 0, it measures the extent to which an ego's alters are connected to the men in the ego network

• Fixed bug in Data|Extract dialogue window which was inserting file information in the wrong edit box

• Added options to Transform|Bipartite so that row and column entity labels can be given distinctive prefixes

• Added new procedure Networks|Ego Networks|Honest Broker that computes the Kilduff and Mehra honest broker index. Essentially, an honest broker is a node who is trusted by two third parties that don't trust each other. The measure is a count of the number of pairs of alters for whom ego serves as an honest broker

• Fixed bug in the way UCINET interfaces with MAGE. Previously, there were problems when there were long filenames containing spaces

• Temporary fix made in Tools|Consensus analysis to allow calculation of probabilities of right answer in answer key when there are a large number of respondents

• Added new betweenness procedure called Proximal Betweenness. A person has high proximal betweenness if they are often the final intermediary along a short path -- in other words, a person that others receive information directly from

• Fixed bug in Transform|Matrix|Between files|Statistical summaries. When adding filenames via the browse button, long filenames were wrapping, making them impossible to read. This no longer occurs

• NOTE: The "AVERAGE" function in Tools|Matrix Algebra as been renamed AAVG

• Changed Networks|Egonets|Egonet Strength to allow defining ego network in terms of outgoing ties only, incoming ties only, both outgoing and incoming, and reciprocated ties only.

• Fixed bug in UNPACK procedure which was putting the wrong matrix label in the output datasets

• Fixed bug in new procedure Networks|Egonets|Egonet Strength which was incorrectly including reflexive ties in the calculations, and (for non-valued) data, was incorrectly averaging by dividing by the total number of nodes rather than degree

• Fixed bug in matrix algebra DIAGONAL function

• Added new procedure Networks|Egonets|Egonet Strength. It calculates the average attribute value of each ego's alters. For example, you can calculate the average centrality of each person's neighbors.

• Added simple Frequencies procedure to Tools menu

• Fixed bugs in Core/Periphery|Continuous program so that (a) it now allows users to change name of output partition file, and (b) includes node labels in the partition file

• Fixed bug in "count combinations" procedure which was causing output file to overwrite input file.

• Changed Data|Egonet procedure so that it would regard missing values as non-ties

• Changed networks|egonets|structural holes procedure:

  • Now outputs node degree and ego-betweenness

  • Default option is now ego model rather than whole network model, which affects the constraint procedure

  • added option for determining what value is given for constraint measure when degree is zero

• Added INTERSECTION procedure to Transform menu. A complement to the UNION procedure, it takes multiple networks as input, finds the nodes that are present in all of them, and outputs networks with just those nodes

• Added TIME STACK procedure to Transform menu. This procedure takes a collection of separate datasets with nodes in common (such as a network sampled at multiple points in time in which nodes appear and disappear), builds a single multidimensional matrix with all nodes that have ever appeared.

• Added some capabilities to the spreadsheet editor to facilitate selecting, inserting and deleted rows and columns

• In the spreadsheet editor, removed the ability to highlight a column by simply clicking on its first row

• Added "DIST" command to matrix algebra procedure. Allows you to compute geodesic distances on unvalued graphs & digraphs. If the dataset you give it has multiple relations, it will compute distance matrix for each relation. Example:

• -->distmat = dist(sampson)

• This would run geodesic distances on each relation in the Sampson dataset, and place the results in a dataset called "distmat".

• Fixed bug in Double-Dekker regression which caused crashes on some computers

• New installation program for UCINET. With this version of UCINET, the directory locations for UCINET and other Analytic Technologies programs have changed. The new locations are a bit more rational. However, there is a down side, which is that if you want this new version to replace your old one, you must uninstall the old one first. If you don’t uninstall the old one, you will end up with two versions of UCINET, and any shortcuts you may have created will call up the old version rather than the new. So, we recommend uninstalling the old version before installing this one.

• Fixed bug in Data|Subgraphs from Partitions. The automatically generated default output filename prefix was wrong -- it included a file extension and shouldn't have.

  • Fixed recently introduced bug in Random Graphs|Erdos Renyi that was placing all ties in upper left quadrant of network.

• Modified installation program. More modifications to come. This may result in the program not being able to find NetDraw. If this happens, go to Options|Helper Applications and select Netdraw entry. Use browse button to find the netdraw program. It is probably located here: c:\program files\analytic technologies\netdraw. ONce you find the folder, just click on netdraw.exe and from then on UCINET should be able to find it again.

• Fixed bug in Qap Regression interface that was preventing choosing multiple independent variables, one after another.

• Added additional output to Transform|Union to save vector indicating which original network a node was drawn from

• Changed options in hierarchical clustering to include two kinds of average link clustering -- one that is weighted by cluster size and one that is not.

• Fixed bug in the weighted average clustering. It had been weighting by the new size of the cluster (after merging) when it should have weighted by the old size.

• Fixed bug in spreadsheet that caused occasional "read past end of file" messages

• Fixed bug in Data|Attribute that created a default output dataset name that was not sensible.

• Fixed bug in Transform|Recode that caused intermittent failures to create output file

• Changed spreadsheet so that clicking on top left cell would not highlight entire column

• Changed Transform|Block and Networks|Cohesion|Density to accept "Identity" as a response to a prompt asking for row or column partition. This makes it convenient to calculate things like the number of ties that each node has to women and (separately) to men.

• Fixed bug in Data|egonet procedure, which was failing to save the ego net on disk (instead, it was saving the whole network).

• Fixed bug in help system that was preventing program from finding its help file

• Added frequency counts to the output of the Networks|Cohesion|Distance procedure

• Fixed bug in handling of filenames with spaces in them. Was causing two sets of full quotes to put around filenames.

• Fixed bug in Data|Join procedure creating output datasets with thousands of matrices.

  • Fixed bug in Networks|Regions|Components|Valued which caused incorrect dendrogram when a solution resulted in the identity partition (where each node is in a class by itself). However, the saved HIPART matrix was always correct.

• Fixed calculation of the average permutation table in Relational Contingency table procedure

• Fixed full quoting of pathnames in several procedures including Data|Attribute

• Added option in Simmelian ties routine to allow outgoing ties only or require reciprocal ties

• Added output options to the QAP correlation routine

• Minor changes to export routine

• A few changes to dialogs in order to encapsulate filenames in full quotes.

• Added new facility for cross-tabbing matrices (e.g., social relations) against each other. Includes qap-based chi-square and correlation coefficients. Find it under Tools|Testing Hypotheses|Dyadic|QAP Relational Crosstabs

• Added new QAP Correlation routine to the main menu. Missing values are handled differently with respect to significance (p-values) and standard errors. In the old qap routine, the significance was based on Hubert's Gamma, which gives identical significances as correlation in the case of no missing values, but gives different values when missing values are present. Another advantage of the new routine is that it accepts more than two input matrices (and computes correlations among all possible pairs).

• Added QAP routine to Matrix Algebra facility. Syntax is as follows:

• --> qap "dataset1" "dataset2" "dataset3" ...

• Notice that if you include more than two datasets in the list, it will run qap correlation between all pairs of matrices. Also, if the dataset names contain spaces, you should enclose them in full quotes. Otherwise, the full quotes are not necessary

• Cosmetic changes to Dichotomize and Symmetrize procedures

• Changed Networks|Cohesion|GTD procedure to automatically disregard self-loops (diagonal of adjacency matrix)

• Fixed Permute function. When permuting order of matrices in a dataset, it was ignoring permutations of the rows and columns.

• Fixed bug in edge betweenness program which caused it to round all figures to nearest integer on saving. Printed values were unaffected as was the hierarchical decomposition

• Changed export to Pajek to include error message when attempting to export 2-mode data

• Changed name of default output file for bipartite procedure

  • Added routine to matrix algebra to intersect graphs. Syntax is: <outdataset> = INTERSECT(<indataset1> <indataset2> ....). For example: Y = INTERSECT(TIME1 TIME2 TIME3). The input is a set of networks. The output is a single dataset with multiple relations. The nodes are just those nodes that are found in all of the input datasets.

  • Changed installation package. Thanks to Rich DeJordy, I think we have finally ironed out issues with upgrading to a new version so that you do NOT need to uninstall the old version (what a pain!).

  • Changed method of scaling factor analysis scores to match SPSS. Also corrected a minor bug in their calculation.

• Enabled factor analysis routine to handle missing values

• Changed default output names for most procedures in the Data and Transform menus

• Fixed problem in cohesion|reciprocity routine that caused crashes when using a partition.

• Changed structural holes routine so that it would output dyadic statistics for large datasets as well as small ones

• Fixed bug in data|attribute causing it to prepend "from" to the default output file name, which screwed it up sometimes

• Some cosmetic changes

• Fixed bug in export to Pajek that caused crashes when not all fields where filled in.

• Added consensus analysis to Tools menu

• Fixed interface to double-dekker qap regression so it would have less trouble with pathnames containing spaces

• Fixed bug in the reading of dataset titles that caused some crashes

• Altered Dichotomize to output a little more information

• Fixed bug in factor analysis program which was putting case labels where variable labels should have been when inputting 2-mode data

• Added feature to VNA import so that user could identify missing values

• Installation package now allows upgrades as well as new installations.

• Fixed Non-metric MDS routine when using Metric starting configuration. Thanks to Rich DeJordy for pointing this out.

• QAP regression updated to reflect the latest method by Dekker, Krackhardt and Snijders. It is called "double-Dekker semi-partialling".

• Changed Data|Attribute dialogue box so that it no longer prefixes "from-" in the output file name

• Fixed bug in E-I index procedure. In printing output from permutations, the values for internal and external ties were swapped. This was only a printing problem and did not affect the calculation of significance. Thanks to "analysisrussell" for pointing this out.

• This installation package includes NetDraw 2.9

  • Some routines modified so that browsing for files automatically puts full quotes around the file name in dialogue windows. Without the full quotes, routines that expect input in the form of filename-space-dimension-space-integer get confused if the filename contains spaces. So if you put quotes around the filename, the program knows it is all one object. IMPORTANT NOTE: if you type the filename in yourself, the program will not automatically supply quotes.

• New installer implemented. Before downloading this version, you should uninstall the old version by using the Windows uninstaller (go to start|control panel|add/remove programs). Datafiles will not be removed by the uninstall.

• The new version handles multiple users of the same machine better, through making better choices of where to put things in the Windows registry. However, installation does require having administrator rights.

  • Ok, this time I actually did change the default value of DisplayFullPathNames to True.

• Added support for three new DL formats for 3-dimensional datasets:

  • EDGELIST3. This is like edgelist2 except there are three columns that identify nodes: the row ids, the col ids, and to relation ids, as in

    • labels embedded

    • matrix labels embedded

    • data:

    • george event1 year2005 22.5

    • george event1 year2006 0.0

    • billy event3 year2006 1.0

  • EDGEARRAY2. This is like edgearray1 except for 2-mode data, as in:

    • matrix labels:

    • year2005, year2006

    • data:

    • george event1 22.5 0.0

    • billy event3 na 1.0

  • LAURA. This format was built for Laura Savoia. It's like this:

    • data:

    • year05 year06

    • george team1 team1

    • billy none team3

• In Jan 2005 degree centralization calculation was modified to yield values between 0 and 1 for valued data. Now (thanks to prodding by Giancarlo Oriani), I've extended the same change to the normalized degree centrality values. So all of these are just their old values divided by the largest edge value in the network.

• Fixed bug in structural holes dialog window in which it would only calculate full network constraint if the user specifically chose full network option. Otherwise it was calculating ego network constraint even if the dialogue window showed "full network". However, all outputs were always correctly labeled.

• Changed default value of DISPLAYFULLPATHNAMES option to true. This allows ucinet to easily navigate to different folders without getting confused.

  • Added normalization option to affiliations procedure to allow dividing number of co-memberships by smaller of the two events. Also added ability to run multiple matrices at one time

• Fixed bug in Data|Remove Pendants which caused a crash when removing pendants removed all nodes in the network. Also hooked up the browse buttons.

• Reorganized Networks menu. Still changing.

• Modified Relational Contingency Table analysis procedure to allow non-symmetric data

• Fixed bug in Degree Centrality (asymmetric model) which was scrambling the labels when writing out to file.

• Fixed bug in hierarchical clustering (introduced yesterday) which caused crashes.

• Minor bug fixes in displays of tools|cluster|moca and tools|cluster|hierarchical clustering

• PERMUTE. Changed permute so that if you set the row permutation to RANDOM and set the column permutation to SAME AS ROWS, you permute both the rows and columns in the same random order (previously the rows and columns were randomized independently)

• UNION. Added option to allow adding up ties that are in common across two networks that share some nodes. So if network A has a tie between Billy and Johnny, and network B also shows a tie between Billy and Johnny, the new network will have a 2 for the relationship between Billy and Johnny.

  • Changed structural holes procedure to (a) handle larger datasets, and (b) include an alternative method of calculating structural holes

• Changed how the degree centrality program calculates centralization on valued data. Now, the centralization value is divided by the maximum value in the input dataset, this ensures that it runs between 0 and 100%

• Fixed bug in spreadsheet -- previously, if you clicked on main window while in the spreadsheet, it reset the spreadsheet (clearing all cells).

• Added row and column indicator to spreadsheet. It can also be used to go to a particular cell.

• Fixed bug in data|export|excel that caused problems exporting datasets with multiple relations to excel on some systems.

• Changed dichotomize function to output real-valued matrices (to accommodate missing values).

  • Fixed bug in ucinet system file reader -- it affected newer procedures that were trying to read interger-type files created by equally new procedures (such as Erdos-Renyi graphs).

• Nothing worth reporting

• Cleaned up output of Transitivity program

• Changed clustering coefficient program to operate on multiple relations in a file

• Changed Import|Raw procedure to more accurately guess the dimensions of the input data

• Fixed bug in Random|Erdos-Renyi procedure. When user selected multiple graphs, the program was constructing them cumulatively so that the density increased with each graph. Now they are separate and the density is identical for all graphs created.

• Fixed bug in Krackhardt GTD routine. Previously, it was failing to output measures to a disk file.

• Added measures of clustering adequacy to Tools|Cluster|Cluster Adequacy. Used to judge whether a partition does a good job.

• Fixed bug in Hierarchical Clustering program which caused E-I index to be calculated on a modified version of the proximity matrix (instead of the proximity matrix itself)

• Fixed bug in Reach Centrality procedure which caused incorrect values for the normalization of in-reach

• Added ability to select number of decimals when exporting vectors to Pajek

• Fixed bug in Data|Import|Multiple Dl files that caused problems with filenames containing spaces and other punctuation

• More finished version of 6.68

• Added two experimental routines that await serious testing: (a) Krackhardt's Graph Theoretic Dimensions (GTD) measures (Networks|Network Properties|Krack GTD), and (b) Reciprocity between groups procedure (Networks|Network Properties|Reciprocity)

• Added Erdos-Renyi random graph routine (Data|Random|Erdos-Renyi). It can create either directed or undirected graphs

• Replaced text editor with windows notepad

• Changed scatterplot facility so that user can drag points around in the map. The scatterplot procedure is used by the MDS programs and the Correspondence analysis program

• Modified Import|VNA procedure to be more forgiving of errors in the data.

• Major bug in RECODE function which caused random wacky numbers has been fixed

  • Small but important bug in Tools|Similarities leading to slight inaccuracy in calculation of similarities between matrices has been fixed.

• Data|Export|Excel and Data|Import|Excel can now handle multiple relations/worksheets

• Added Export to Excel in main menu. Note that multiple relations/sheets are not yet supported

  • Fixed bug in 'Subgraph from Partitions' which caused problems with 0/1 partitions.

  • Added info to About box so you can tell whether a copy is registered or not.

• Browse button on Fragmentation now working

• Added ability to import VNA files (Data|Import|VNA). Only imports network and node attribute data -- node and tie properties are ignored.

• Added Fragmentation Potential metrics (drawn from KeyPlayer program) to Centrality menu

• Added "remove pendants" routine

• Added "spherical distance" routine under Tools|Dissimilarities in order to compute distances from latitude and longitude data, such as obtained from GIS/GPS systems

• Fixed bug in REGE which was converting all data with decimal places into integers

• Fixed bug in hierarchical clustering which was preventing it from saving cluster id variable. Also renamed the cluster id dataset from "tabuclus" to "clusterid"

• Changed starting cutoff value in dialogue box for "subgraphs from partitions" routine to 1 (was 3).

• Fixed bug in Extract routine which prevented usage of datasets stored (by ucinet) in a new internal format

• Modified UNION procedure to allow option of putting missing values (instead of zeros) for the links between graphs. Also modified the way it handles node labels when "match by labels" is not selected.

• Fixed bug in the Data|Import|Multiple DL files procedure that didn't allow selection of several files at one time in the file browse facility

• Completed addition of default output folder -- now all routines except one's that don't make sense for it should send outputs to the designated output folder.

• Added default output folder -- if non-blank, most routines will now direct output to that folder.

• Fixed bug in Recode dialog box causing an error message

• Fixed bug in reading Pajek net files that came in with version 6.37

• Fixed bug in exporting dl files in nodelist2 format (was writing "nodelist1" instead of "nodelist2")

• Fixed bug in EdgeArray1 data reader (thanks to Kate Lawrence for pointing it out). The problem was that for data with labels embedded and alpha=yes (the default) the program sorting rows and columns of the data matrix in alphabetical order, but not actually changing the labels, so the wrong labels were attached to each node.

• Changed hierarchical clustering to allow printing sideways textual dendrogram

• Fixed bug in hierarchical cluster analysis -- since version 6.43 it was saving the partition file (PART) incorrectly, resulting in bad graphical dendrograms

• Fixed major bug introduced in version 6.44 (April 9th, 2004). The import procedure was incorrectly reading dl files using fullmatrix format in which row labels were not embedded.

• Added facility for creating new networks (or matrices in general) based on boolean combinations of old ones. E.g., create matrix Y so that Yij = 2 if Xij > 0 and Zij = 34, and Yij = 1 otherwise. It is under Transform|Matrix operations|between datasets|boolean combinations.

• Enabled column/row delete functions in matrix spreadsheet

  • Fixed bug in IMPORT DL procedures causing first letter of row/column labels to be cut off when reading files using Fullmatrix format.

• Hierarchical clustering now runs on each matrix in a 3-way dataset. Also includes some additional diagnostic output.

• Metric MDS changed to accept datasets with multiple matrices -- it automaticaly does a scaling on each matrix in the dataset.

• Correspondence analysis similarly changed.

• Fixed Remove Isolates to work with non-symmetric data and multiple relations

• Changed Tools|Dissimilarities procedure to enable multiple relations and to permit computation of distances between relations. Also added Hamming distance measure.

• Union procedure added to the Transform menu. Use UNION to combine separate graphs into one larger graph -- taking the union of all nodes and lines.

• Fixed bug in Permute procedure

• Added Remove Isolates procedure to Data menu. Provides one-step way to generate network with isolates removed.

• Added Extract Main Component procedure to Data menu. Enables user to create a subgraph representing the main (weak) component of a network

• Added new output to Brokerage routine that shows, for each actor, who they are brokering between.

• Added ability to import Pajek 2-mode net files

• Added ability select multiple files in the Transformation|Arithmetic|between datasets procedure, as well as the Data|Import|multiple dl files procedure

• Added additional QAP regression methods, including Dekker, Krackhardt and Snijders k-step semi-partialling method which is very slow but very accurate. For my money, though the fast ordinary semi-partialling is best choice because it is way faster and its only fault is being too conservative (doesn't reach significance often enough). But I'm sure Dekker, Krackhardt and Snijders would consider this a very irresponsible thing to say.

• Overhauled dl import routines to handle embedded labels better (especially ones containing spaces)

• Added multiple dl import routine to handle reading many files at once.

• Added arithmetic transformation routines to Transform menu; now, things that in the past had to be done via matrix algebra, can be done via menu interface

• Overhauled matrix algebra routine

• Non-metric mds routine has been changed to run on all layers of a multi-matrix dataset. Also, the bug that prevented running on matrices greater than 255 rows or columns has been fixed.

• Replaced the spreadsheet with a new one that is not limited to 255 columns. It can also do some limited transformations of the data.

• Changed the output.log browser so that seeing large output takes less time. For example, running Data|Display on large datasets used to take forever. It is a bit faster now.

• Added a spreadsheet-like matrix browser that is faster than the new spreadsheet for viewing large datasets in matrix form

  • Changed input format for Recode procedure to facilitate more complicated recoding procedures.

• Adopted new installation procedure that is a little faster, a little simpler. Keyplayer is no longer included with UCINET installation.

• Changed output of geodesic distance procedure to suppress screen output for large datasets. Now processes a 5000 node dataset in 2 minutes.

• Fixed bug in UCINET versions 6.04 discovered by Melissa A. Schilling in which degree centrality saves a centrality dataset that has the wrong labels. The numbers are right, and are in the original order, but the labels for the nodes are in the wrong order.

• Fixed bug in line (edge) betweenness program, which was causing problems reading the output file

• Added BETA TEST version of new partialling approach to QAP regression. Seems to work fast and well.

• Updated the version of KeyPlayer that is automatically installed with UCINET. Apparently, old version was missing a file.

• Changed default on line betweenness routine to not do Newman-Girvan hierarchical clustering automatically. See also changes in NetDraw.

• Modified edge betweenness procedure (Networks|Centrality|Betweenness|Lines) to automatically output hierarchical clustering of nodes obtained by successively deleting most central link and then recalculating

• F-Group procedure from UCINET 4 reinstated. No documentation as yet.

• Changed maximum flow & lambda set procedures to accommodate networks with more than 255 nodes. Now there is no limit.

• Changed Factor Analysis routine to permit input of 1-mode association matrix (e.g., correlation matrix, co-occurrence matrix)

• Added Options|Helper Applications, which lets you tell ucinet where netdraw, pajek and mage are

• default name for netdraw program is now netdraw1.exe

• nodelist and edgelist readers changed to allow names enclosed in quotes as in "John Smith" or "Smith, John Jr."

• Improved help file added.

• More recent version of Netdraw added that allows text attributes (e.g., gender can have values of "male" and "female" and "not clear").

• More speed improvement on reading nodelist and edgelist files.

• Group centrality|Degree|Test routine fixed (was giving crazy answers and hanging up).

• Fast routines for reading nodelist and edgelist data files reinstated.

• Button added to Data|Egonet routine that allows selection of the focal nodes by choosing node names from list.

• Reach centrality procedure changed to (a) eliminate eccentricity (it will be added back to the closeness procedure later), and add new distance-weighted reach measure

29 Sept 2002

• New help files added

• The Tools|Dissimilarities procedure was fixed. It had been ignoring one cell in all computations, resulting a dissimilarity score that was only approximate. Thanks to Jim Moody for pointing this out.

• Continuous Core/Periphery routine altered so that its recommendation on where to cutoff the core is now based on the correlation criterion

• Expected values from the continuous core/periphery routine have been fixed -- Previously, the rows and columns had been sorted according to coreness, but the row and column labels were not changed, resulting in incomprehensible numbers. Now, the matrix is presented unsorted.

• Centrality routines now sort output automatically in order of decreasing centrality

• Netdraw graph drawing routine added.

• All routines that used to output a partition-by-actor vector or matrix now output an actor-by-partition matrix. This way, what are normally considered variables are always columns.

• Help files updated to include new procedures. NOTE: Netdraw still has no documentation. It's coming.

• Edge Betweenness program fixed to eliminate the "can't open scratch file" error.

• Now able to launch both Mage and Pajek programs from within UCINET.

• Now able to import Pajek Net files

• Experimental degree group centrality procedures added. One procedure allows you to find which individuals should be included in a proposed group such that they have ties to as many different individuals as possible. Another procedure accepts a priori grouping (such as a set of managers) and measures their degree group centrality along and provides a statistical test

• Improved method of separating core from periphery when using the continuous core/periphery model, along with improved concentration measure.

• Changed current directory display to a drop down menu so you can easily go back to previous folder

• Fixed bug in Attribute procedure which lost the labels of nodes.

• Improved optimization algorithms for structural equivalence and regular equivalence

• Bi-component routine now outputs a vector indicating for each node whether it is a cutpoint or not

• Components routine now outputs a vector indicating whether a node is in the main component or not

• Fixed bug in E-I index procedure which caused crash on some computers

• Improved display of the "loglinear" autocorrelation procedure (which basically does crosstabs of ties)

• Includes new routine for exporting to the METIS program, which can cluster graphs with millions of nodes extremely quickly.

• Also includes a much faster betweenness centrality routine (relative to the one in versions 5.64 and earlier), using Ulrik Brandes' algorithm.

• Includes slightly improved routine for point-connectivity (# of node-independent paths between pairs of points). It is still very slow, though.

• Fixes bug in DESCRIBE that prevented imported labels from being saved.

• Adds betweenness decomposition routine which iteratively removes nodes with zero betweenness, yielding a core group of nodes that are mutually interdependent.

• Edge betweenness routine, ala Anthonisse has been added. Even newer version of betweenness implemented, using Ulrik Brandes' (Ulrik.Brandes@uni-konstanz.de) algorithm. It works with both symmetric and non-symmetric data.

• New closeness routine added that calculates average length of all paths between two nodes, not just shortest paths. To be used with small graphs (< 100 nodes)

• Bug fixed in MRQAP causing near infinite standardized coefficient when there were more than 32767 ordered pairs of nodes.

• Import>RAW routine improved to guess the number of rows and columns better when the entire dataset is in one row.

• Even newer version of betweenness implemented, using Ulrik Brandes' (Ulrik.Brandes@uni-konstanz.de) algorithm. It works with both symmetric and non-symmetric data.

• Closeness and geodesic distance routines have also been upgraded to run faster with sparse data (switched from dijkstra to bfs)

• Mage export routine can now export 2-mode, rectangular datasets (as long as the number of points in the coordinates dataset equals the number of rows plus the number of columns in the 2-mode matrix

• Scatter plot routines (include MDS and Correspondence analysis) now include menu option for increasing or decreasing size of label font

• Mage export routine now allows you to skip the adjacency matrix and just input coordinates.

• New version of betweenness implemented that typically runs around 3 times faster for typical graphs of about 1000 nodes. Algorithm courtesy of Mark Newman (mark@santafe.edu). However, it only runs on symmetric data (undirected graphs).

• Bonacich Power routine modified to allow non-symmetric data.

• Right eigenvector routine added to matrix algebra procedure. To use, type REIGEN <filename>. Note this routine uses the power method and will only converge on a subset of matrices.

• Instability in the mrqap regression routine was eliminated. Again.

• Ability to export files to Pajek and Mage was added

• Bug fixed in SORT: it wouldn't recognize lower case letters in the dimension prompt of the dialog box.

• Instability in the qap regression routine was eliminated.

Two major bugs fixed in this version.

• There was an error in the egonet>density routine in some versions of 5.4 (I think earlier versions are fine), producing random numbers instead of the calculated values.

• There was a bug in the REGE routine so that all data containing decimal places were truncated. Numbers less than one but greater than zero were converted to 1s, and numbers greater than 1 were rounded to nearest integer. This bug goes back many years.

• In addition, a new routine was added in the statistics menu, enabling users with krackhardt-style css data to test for significant differences in different aggregations of the data (for example, comparing the aggregate female matrix with the aggregate male matrix). No documentation is available yet.

• Several minor fixes, including the default output name of the dichotomize procedure.

• This version fixes a major design error in the density or anova-based categorical autocorrelation routine. In previous incarnations, the routine calculated r-square correctly, but the significance test was biased, yielding significance too often. The routine has been completely replaced and now calculates significance correctly. In addition, it gives additional informative output.

• Adds Gould and Fernandez brokerage measures.