EUGene Changes

EUGene Changes

(Cumulative)

Version 3.203

  • Ease-of-use improvement: Changed program interface so entry of subrange years automatically clicks the subrange button so user doesn’t have to (and subrange is output if user forgets). Clarification: Improved two error messages that report particular errors if certain elements of a user data set are not specified correctly.

Version 3.202

  • Bug Fix: Fixed an error where EUGene should have been (but was not) requiring the selection of complementary variables under some circumstances with non-directed data. This led to an error when certain subsets of user variables were selected from non-directed data user sets. This has been corrected, and EUGene now forces the selection of complementary variables when necessary.

Version 3.201

  • Bug Fix: Fixed an error introduced in v3.2 that prevented access to user data. Users should no longer get an error message when openting the “User Data” dropdown menu.

Version 3.2

Major additions include:

    • Two new units of analysis for output data sets based on non-directed disputes are now available.  Users may now select as output units one case per non-directed dispute dyad onset, or one case per non-directed dispute dyad year.  Since these units of analysis are based on dyadic MID interactions rather than the dyad-year, all instances of MID onset may be captured using this unit.  These non-directed units (where dyads are coded without referent to directed “initiation” which many consider problematic) are not based on the dyad-year or country-year as the core creation unit, but rather, start with all dyadic MID interactions.
    • Stata command (.do) files generated with EUGene now set variable labels.
    • Latest version (v3.1) of the COW MID data included, and is the basis for dyadic MID calculations.

Bug Fixes include:

  • Corrections to the computation of dyadic MIDs were made. EUGene now generates correct dyadic dispute data using the latest version of the COW MID data. The most important fix includes reporting the correct hostility level (especially war) when multiple incidents “add up” to a higher hostility level. Those hostility levels are reported only in the MIDB file, not the MID IP file, and are now correctly reported. In addition, the creation of disputes is now correctly handled for the circumstance when states exit and re-enter MIDs. Complete details on these changes are available in the document “Codebook for Dyadic MID Data v3.10.doc” in the “inputdat” subdirectory under your EUGene installation path.

Version 3.1

Major additions include:

    • New Variable: Countries’ ISO codes are now available as an output variable, as well as COW country codes. This should facilitate merging with other data sets that employ these codes.
    • Expanded Risk Attitude Data: Risk data computed with unweighted S scores as the similarity measure are now available for 1816-2000 (the period for which alliance data is available). Data from 1993-2000 were computed using the COW alliance data v3.0; risk data to 1992 are identical to prior risk attitude data distributed with EUGene, and use the earlier COW alliance data. [Risk data based on tau-b scores are available to 1992; data based on weighted S scores are not available.]
    • New data: Hypothetical alliance pattern output now available. Risk attitudes are computed on the basis of where a state’s actual alliances fall relative to the alliances it could form that would maximize its security, or minimize its security. The hypothetical best and worst alliance patterns may now be output for analysis.
    • Data Browser available. Under the file menu is an option for “Dataset Browser” which allows users to look through input data files without/before generating data sets.

Minor changes include:

    • Peace Days variable now available (in addition to peace years). For any dyad year, this is the number of days that passed since the termination of the last dispute between ccode1 and ccode2, through January 1 of that year.

Bug Fixes include:

  • Fixed an error with selecting sampling; when sampling is selected, a dispute exclusion set must also be specified. EUGene now sets an exclusion set if the user didn’t.
  • Fixed a bug in computing Pi in EU calculations when the denominator of the balance of forces expression was 0 (this situation was not being taken into account, there is now an appropriate check that catches 0-capability dyads.).


Version 3.04

  • Bug Fix to Maoz data set. A user discovered that creating dyadic output data when using the Maoz version of the dyadic dispute data in EUGene v3.03 was generating several duplicate dyad years. The problem was traced to some typos in the Maoz DyadMID6.0 data file, and a problem in EUGene processing this file. This version of EUGene corrects the internal problem, and also releases a new version of the Maoz DyadMID (file dyadmid601.csv) data which corrects these typos (with Maoz’s approval). A file which is installed in the “docs” subdirectory titled “Maoz data set MID errors corrected in dyadmid601.doc” details the 7 typos corrected. Note that none of the typos were in substantive variable values, but largely involved the ordering of multiple annual records within some MIDs.
  • Incorporated new COW capabilities data v3.01 (which corrected an important typo in Russian capabilities for 2000). The data is available through CINC scores and component capability series, and expected utility data have been recomputed using the updated data.
  • New documentation available: For any user wishing to trace what versions of component data files were used in each version of EUGene, the “docs” subdirectory now includes the file “EUGene dataset tracking.xls” which lists the file names and dates of each input and intermediate data set distributed with each version of EUGene.

Version 3.03

  • Bug Fix to component contiguity data: Direct contiguity values for states were being given incorrectly for 2001 and 2002 because the direct contiguity data actually only went up to 2000, although EUGene allows output after that. The data file actually showed all listed direct contiguity relationships as ending in 2000; all relationships after this were listed as non-contiguous. The component input data file for contiguity has been updated to cover through 2002.
  • Bug Fix: The Maoz MID dyadmid data set codes the “highest action” variable on a scale of 1-23, following the MID v2.1 data coding. The COW MID 3.0 data set codes the “highest action” variable on a 0-21 scale (two categories were collapsed, and the base is 0). EUGene did not make any adjustment for this, and so data from 1816-1992 had a different range on this variable than 1993+ data when the Maoz MID data was selected. The “highest action” variable values from the Maoz dyadmid data are now converted to the new scale so that the scale now matches across the full range of the data.
  • Bug Fix: Germany and Yemen changed their country code numbers when the formerly separate E/W Germany and N/S Yemen unified in 1990. This resulted in four incorrect dyads being generated between 255-260 and 255-265 in 1990, and 678-680 and 679-680 in 1990. These are invalid dyads because they are in fact the same state. These dyads are now no longer included in output unless the user selects the “Include dyads i vs. i” output option, which creates a dyad of all states vs. themselves.
  • Bug Fix: corrected a bug in version 3.0 that resulted in an incorrect number of variable names being listed in output data sets when the “Revision Type” variable within the Maoz dyadic conflict data within non-directed dyads was selected.

Version 3.02

  • Bug Fix: corrected a bug in version 3.0 that affected the coding of the MID onset variable in nondirected dyad-year output data sets. Due to a programming error, a significant number of MIDs were not correctly being coded as new MID onsets in v3.0 (in the program, a search loop for new MID initiations in a given year was being terminated prematurely, stopping after identifying a MID in an earlier year without continuing the search to the year in question). The output correctly reported the MID numbers of the MIDs in these years, but the MID onset variable sometimes took the value “0” instead of the correct “1.”
  • Clarification: corrected the variable label in the nondirected dyad-year unit of analysis which was labeled “MID initiation” rather than “MID Onset,” the appropriate label for the nondirected output.
  • Clarification: eliminated the box allowing the user to choose a different treatment for target vs. initiator dyads for the directed-dyadic-initiation unit of analysis, because this choice was not relevant to that unit of analysis and made no difference in the output.

Version 3.0

Major additions include:

  • New Variables: Several new variables are available concerning system structure, including the number of states in the system at any time, number of major powers, and the CON and MOVE indices of system power concentration. The MOVE index allows options to compute annual change or change over 5 years, and change involving all states or only major powers.
  • Redesigned and updated interface
  • New integrated conflict data sets, including ICB dyadic crisis data and MID v3.0 dyadic dispute data, along with the Maoz version of the MID 2.1 alliance data.
  • New component data sets, including COW alliance data v3.0, state membership data to 2002, COW MID v3.0 data, and contiguity data (colonial and direct).
  • S and Tau have been recomputed using the COW alliance data v3.0 set. Expected Utility data have been recomputed with new S and Tau values.
  • Seamless integration between new COW MID v3.0 data and pre-1992 COW or Maoz dyadic MID data.
  • In years with multiple MIDs in a dyad, provides the option to output information on either the first MID, or the most serious MID.
  • EUGene now outputs separately the MID numbers of disputes that are ongoing, disputes that are newly initiated, and MIDs with the highest level of hostility in any given dyad-year, allowing easier identification of dyad-years with particular key MIDs.
  • New variables, including MID names, number of MIDs states engage in within a year, state role in MIDs, system characteristics (CON and MOVE).
  • Output may now include all conflict variables (e.g. Maoz dyadic MIDs, COW MID set, ICB data) rather than just 1.
  • Peace Years available and computed separately for each conflict set, e.g. based on ICB crisis occurrence, or COW MID occurrence, or based on Maoz dyadic MID set.
  • Any of the conflict data sets (Maoz dyadic MIDs, COW MID set, ICB data) may be used to drive case exclusions for ongoing MIDs and related settings.
  • Incorporation of Suzanne Werner’s adjustment to 1816 “peace years” values.
  • Many more user data sets are now available on our server for download to use within EUGene.
  • Expected utility data have been recomputed using the latest (v3.03) alliance and capability (v3.0) data, but using existing risk scores. Because risk scores have not yet been recomputed, expected utility estimates only run to 1993. Now that we have the new COW capability data, we will begin reestimation of risk scores, a process expected to take significant computer time, as it involves 1) reestimation of EU-War Trap with new alliance data and new capability data, 2) reestimation of risk attitude scores with EU-WarTrap values, new alliance data, and new capability data, and then 3) reestimation of EU-War and Reason data with new risk, new alliance, and new capability data.Minor changes include:
  • In computation of “dem” variable based on democ and autoc values, values of -66, -77, and -88 are not automatically treated as missing, but rather if both values are -66 (e.g.) then the output dem variable will have a value of -66.
  • Menus now provide direct access to documentation on the program and on user data set creation.
    A web link to our site is now provided under the Help | About menu.
  • Options to use old versions of the alliance data have been removed with the integration of the COW alliance data v3.0.Bug Fixes include:
  • Fixed a problem reading some user data sets, and accessing some dyadic user data sets through the directed dispute-dyad output unit.

Cautions:

  • The location of some critical data set specification settings have moved (in particular, settings about how to treat ongoing disputes, and other dispute initiation settings have moved under the “Conflict data” tab under “Variables”).
  • Several key variable names on conflict variables have changed in this version in order to accommodate the multiple conflict data sets now included in EUGene. For instance, the dyadic variable marking the dyadic occurrence/initiation of a MID was formerly named “initiate” in directed dyad output, and “mid” in nondirected dyadic output. Now, dyadic conflict variables are named “cwinit” (COW data set based conflict initiation), “mzinit” (Maoz dyadmid data set based conflict initiation), “cwmid” (COW data set based nondirected conflict occurrence), and “mzmid” (Maoz dyadmid data set based nondirected conflict occurrence). Variable name changes are listed here; the full list of variable names is contained in the full EUGene documentation file.

 

Version 2.40

  • New COW Alliance data incorporated. Distribution of Version 2.40 now includes the new Correlates of War alliance data set (v3.0) as its default alliance data set. This involved converting the COW list of alliances to dyadic format. Note that we have not yet recomputed tau-b scores, S scores, risk attitudes, expected utility values, or any other variables that rely on alliance patterns. [Prior to EUGene version 2.4, selecting “dyadic alliance file” as the input alliance data used data as corrected in Bennett (1997 – AJPS), along with alliances coded from 1984-1992 by Dan Reiter. You may still access the earlier version of the alliance data by editing (manually) the file “eugene.ini” and changing the entry for “dyadic_alliance_data” in that file. If the entry is changed to “inputdat\AllianceData_Oct2001.txt”, then the prior version of the data including Reiter’s updates will be used.]
  • New Variables: Several new variables are available concerning system structure, including the number of states in the system at any time, number of major powers, and the CON and MOVE indices of system power concentration. The MOVE index allows options to compute annual change or change over 5 years, and change involving all states or only major powers.
  • New case selection options: 2 new subset options are available: 1) all dyads within a specified maximum distance (distance between capitols), and 2) all dyads within a region. Note that currently, generating dyads within a maximum distance is quite slow. We hope to improve this speed in a future version.
  • Command file enhancements: In all command files generated by EUGene, there now appears a list of citations for the component data sets employed in creating program output. Users should include these citations in any publication related to the generated data set. Also, each command file now includes full program specification information in terms of years selected, subset of cases, and important options on case selection such as the inclusion or exclusion of ongoing dispute years. This should allow for easier replication.
  • Distance data fix: In the course of doing this update, we discovered that the latitude and longitude of 6 cities (Dhaka, Bishkek, Vanuatu, Cape Town, St. Petersburg, and Berlin) had been misentered in the original “latlong.raw” data file. The original file has been replaced with “latlong2.raw”. The new cities will affect distance between certain states. However, we have not yet recomputed expected utility values and other variables that rely on distance discounting.
  • Bug fix: v2.25 previously corrected a problem with the coding of Germany’s alliances, corrected in the alliance data file “AllianceData_Oct2001.txt”. However, in some distributions/installations of v2.30, the fix was broken by an incorrect parameter in the initialization file “eugene.ini” which led to an old data file being used rather than the new, corrected file. With the release of the COW alliance data v3.0, the parameter for alliance data uses the newest data, and so this problem has been corrected. However, if you were using alliances heavily in prior research, we recommend investigating the old eugene.ini file to determine whether the correct (Oct2001) or incorrect (July2001) alliance data file was used in your research.
  • LIMDEP message: The software program LIMDEP only allows 200 variables. An error message is now displayed if the user selects too many variables and has specified that a LIMDEP command file is to be generated.

Version 2.30

  • Bug Fix: Due to a server change at our website, there was a problem with ftp access to add-on user data sets in V2.25. We have updated the ftp code inside the program using a new ftpcomponent package, and access to add-on user data sets should now be possible again. We also updated the server access information in the program, and have updated how identification of EUGene ftp servers is handled to make future server changes easier to handle.
  • Bug Fix: If users ran a dyadic data set creation, then ran a monadic data set creation in succession, errors would occur if dyadic variables had been selected that are not available in monadic data sets. The problem was that EUGene did not clear the dyadic variables from its list of variables to put in the output file. However, when monads were selected as the unit of analysis, it made no logical sense to attempt to output dyadic variables, the program did not read the dyadic data sets, and crashed. Now, dyadic variables are appropriately cleared when the user switches units of analysis.

Version 2.25

  • New Variable: Alliance portfolio counts (the tables used for computation of tau or S similarity scores).
  • New Variables: Expected utility and IIG equilibrium data computed with weighted S available (in addition to data computed with unweighted S).
  • New Variable: COW country abbreviations as well as country code numbers.
  • Joiner Clarification: Clarified the nature of the Joiner variables in the documentation. EUGene reports that some states are both “originator” and “joiner” dyads, which seems impossible. However, this isn’t actually what the “joiner” variable tells you, a subtle but important difference that has now been clarified in the documentation. First, recall that the joining variables tell you only about the behavior of ccode1, and whether/how it was involved in joining in this dyad. Intuitively, we would think that joiner_i and joiner_t would tell you whether ccode1 was a joiner in this dyadic dispute on the initiating or target side, respectively. BUT, in fact the variable doesn’t really tell you that. It tells you whether, IF this dyad is a joining dyad (so at least one of the two states was a joiner), whether ccode1 was on side i or side t. In some cases, ccode1 could be listed as an originator and a joiner, if the TARGET in this directed dyad was a joining state (became involved late). In this case the dyad involves at least one joiner, but ccode1 was also in fact an originator. A revised phrasing of joining on the initiating side is: In this directed dyad, one of the states was a latecomer, and ccode1 was on the initiating side of the MID. A revised phrasing of joining on the target side is: In this directed dyad, one of the states was a latecomer, and ccode1 was on the target side of the MID. More details are available in the documentation, or this file (a Word document): Originator-JoinerAnomalies.doc
  • Joiner Clarification 2 and new joiner variables: An additional clarification on joining has resulted in new names for the joining variables, and two variants on those variables. The joiner_i and joiner_t variables coded through v2.0 of EUGene coded whether the state was involved in joining in any MID in that year. Frequently, though, this joining did not occur in the first MID of the year, which is the MID on which information is reported by EUGene when there are multiple MIDs. So, even though the joiner variables reported joining behavior, the MID # reported might not have involved any joining, and so cases of valid initiation were being dropped accidentally. As a result, EUGene now reports two sets of joining variables. The first, joinanni and joinannt, report on whether ccode1 joined the initiating side or target side of ANY MID in the year in question. The second variables, joinmidi and joinmidt, report whether ccode1 joined the initiating side or target side of the MIDNUM reported for the year in question. This will better separate out these two situations of joining. In most cases users will want to use joinmidi and joinmidt; the annual variants are left mainly for legacy coding reasons.
  • Bug Fix / data update. Alliance data has been updated. Previously, there was a problem with Germany in 1990-1991-1992, because the COW country code switched back from 260 to 255, but the alliance data did not reflect the resulting NATO alliance involving 255, nor did it reflect a neutrality pact with Russia involving 255. As a result, data output from EUGene would show no alliances between Germany and other states after 1990. New alliance data is now contained in the data files AllianceList_oct2001.txt and AllianceData_Oct2001.txt. Note that expected utility data have NOT yet been recomputed using the revised/corrected data.
  • Bug Fix: Fixed a bug in the Maoz dyadic dispute data which would incorrectly code joining in subsequent years of a dispute (joining in the first year of a MID was OK; joining coded using original COW data was also OK).
  • Bug Fix: Fixed a bug in using the COW sequenced alliance data which would report some (nonexistent) errors having to do with mismatched country codes. [Essentially, when COW state abbreviations changed, the coding file for the sequenced alliance data did not change. This has been updated.]
  • Bug Fix: Fixes a bug whereby contiguity after 1993 was reported as “6” (non-contiguous) rather than “-9” for missing. (Contiguity data ends in 1993).
  • Bug Fix: Fixed a problem in entering years for a data set in the User Data Set upload/creation section, where it would report problem with each character a user entered for a date, rather than checking once after everything was entered.

Version 2.10

  • Bug Fix: There was a problem creating the MID onset variable using the Maoz MID data. Sometimes an onset was not marked when there really was a new dispute onset. This was due to an internal programming error in looping through all possible disputes to find new disputes in a given year. If you previously used EUGene to create non-directed dyad data using the Maoz MID data, we recommend that you recreate your data file using version 2.10 and rerun any analysis.
  • Save and Load settings: under the “File” menu there are now options for “Save Settings” and “Load Settings.” Selecting “Save Settings” will save all of the current specifications you have entered in your EUGene run in terms of population of cases selected, output files and format, variables selected, sampling proportions, and MID and case exclusions. Selecting “Load Settings” will load previously saved settings.

Version 2.013

  • New Data: With this version of EUGene we are releasing several new component data sets. These data all run to 1992, as we have used Dan Reiter’s suggested updates to extend alliance data to 1992 (the latest COW alliance data set runs to 1984). Newly available data include:
    • risk, utility, and IIG equilibrium data generated using the Signorino and Ritter “s” score (these were generated using the unweighted version of s only)
    • IIG equilibrium data using tau-b to 1992, and corrected for the minor problem listed below.
    • s (weighted and unweighted) scores to 1992
    • tau-b scores to 1992.
  • Updated Data: this release includes version 1.1 of Zeev Maoz’s Dyadic dispute data.
  • Stata command file change: Stata command files for reading EUGene data are now created with explicit variable data types (float, int, byte) on the variables. This will save memory. Users may still want to manually increase the memory size (“set memory = xxM”) in the command file created by EUGene if they plan to create additional variables or do sophisticated analysis procedures.
  • Bug Fix: Internally, in the program, an initial probability value for B’s probability of winning was not being initialized to exactly 0.0 in the computation for expected utility following the War and Reason methods. Normally the unitialized variable had a default value very close to 0, so fixing the code to set the variable to 0 led to very few changes. All new values correlated with old values at >0.99; the EU changes resulting from the procedure change led to a change in the expected utility equilibrium prediction in .24% of cases (so a proportion of 0.0024). This would only affect users who used the equilibria as independent variables in other work.
  • Variable Name Fix: Corrected the renaming of user data set variables when users select variables at one level of analysis to appear in output data sets at a different level. When user variables were already 8 characters long, they were not being renamed as described in the documentation. Now variables such as “gdppercp” will be correctly named “gdpperc1” and “gdpperc2”.
  • Updated the internal procedure to read string values from input data files to properly handle strings that are marked off by either single or double quotes and also contain commas, and also double quote delimited strings that contain a single quote (e.g. Yemen People’s Republic).
  • Minor interface changes: Made some minor corrections in highlighting of buttons and user data setup procedures.

Version 2.0

  • Major feature: Version 2.0 adds the ability for users to add their own variables and data to the program, and to upload and download datasets to the EUGene data server. We hope that this will expand EUGene’s usefulness to the research community by providing an easy way for users to share their own important datasets with others and merge new variables into output datasets.
    • Under the “variables” output tab, there is now a check box for “User Supplied Data” and a button labeled “User Variable Selection”. Clicking the button opens a new window that will display available user datasets and the variables in them. Selected variables will be included in the output files created by EUGene. EUGene is aware of the nature of the nature of the variables (whether the variable is country-year data [e.g. GDP], annual data [e.g. system power concentration], nondirected-dyadic data [e.g. total dyadic trade], or directed-dyadic data [e.g. exports from country A to country B]). EUGene will make appropriate conversions given the input data type and output selected by the user: for example, if the user selects the variable “GDP” while requesting a dyadic data output file, EUGene will create variables containing GDP information for both countries in the dyad; if the user attempts to select a directed-dyadic variable while requesting non-directed output, EUGene will output both directions of the variable; if the user attempts to select dyadic variable when requesting monadic output, the user will not be able to select that variable.
    • Users may make their datasets available for others to use in EUGene. Once they have created the appropriate “EUGene configuration file” (see below), user can upload their data to the EUGene data server. Datasets that have been uploaded will be available for other users to download, so that variables from these uploaded datasets will appear for selection within EUGene. .
    • User dataset requirements: User datasets must be in a format that EUGene can process (data files must be comma-separated variable text files, and must include variables identifying country-codes, year, and a dataset version). Also, in order to properly handle user datasets, EUGene must be aware of key characteristics of user datasets such as the unit of analysis, variable names, and certain relationships among variables. EUGene understands and incorporates user data by looking for a “EUGene data configuration file”, a file that includes these key specifications. This file can be created by users for their own datasets using a data configuration program included within EUGene and accessed via the “User Data | Create User Data Configuration File” menu selection.
  • Bug fix:  corrects a discrepancy that was introduced in v1.94 and 1.95 concerning treatment of joiners if “include all joiners” was selected in directed dyad or dyadic-dispute data sets. Even if users selected “include all joiners” in specifying MIDs exclusions, one subset of joiner-dyads was still excluded, namely dyads where the initiatior was involved on day 1 of the MID but the target became involved later. Such cases were included in output from EUGene through version 1.94, but were excluded in v1.94 and 1.95. They are again included in output using v2.0. Click here to see further details of this problem. NOTE:       This discrepancy was only a problem for users who 1) were selecting “directed dyads” or “directed dispute dyad data” for output, 2) were specifiying “Side A as initiators” for coding initiators,” and 3) WERE INCLUDING JOINERS IN DATASET CREATION AND ANALYSIS. This is not an issue for nondirected dyads, nor is it an issue when analyzing dispute initiation involving only dispute originators.       It was also not an issue if you selected “revisionist states” as the initiators, rather than side A. In these cases the “Include All Joiner Dyads” selection worked as it had in v1.19.
  • Bug fix: corrects a problem with using the Maoz dyadic mid data given a nondirected MID output selection. Some output disputes were incorrect; MID variable values were being repeated in years after the MID ended. In years after a MID had ended, MIDs were still being output for the previous MID. The program was searching for information on MIDs in prior years, rather than checking only the current year. So, for instance, US-Germany in 1990 reported information on a MID and reported variable values from the previous MID (which actually ended in 1945) rather than reporting no MID.

Version 1.95

  • Bug fix:  corrects a small problem with the alliance variable, which was reporting values of “-1” for missing values (from 1993 forward) rather than “-9.” Values of “-1” in version 1.94 on this variable should be treated as missing values.
  • Modification: The resizing of some windows has been modified with the goal of better readability on multiple systems in mind.

Version 1.94

  • Addition:  non-directed dyads as an option for creating data sets.
  • Addition:  several new variables may be selected for output, including s scores, alliance membership, components of COW capability data, and additional MID variables.
  • Added Maoz dyadic MID data set for dependent variable output.
  • Internal changes:  Code for I/O procedures rewritten to allow for user data sets and variables (to be available in version 2.0).  Procedures for expected utility and risk calculations modified to allow s or tau-b to be used as input.

Version 1.19

  • Bug Fix:  The dyadic duration variable counted years since two states first entered the international system.  This did not take into account that states could exit the system.  The corrected version modifies the “dyadic duration” variable so that it resets to 0 when a state drops out of the system by COW rules.  For instance, France-West Germany have a dyadic duration of 0 in 1955 when West Germany becomes a sovereign state again.
  • Clarification:  Setting contiguity through the button under the “Population of Cases” tab of the output window changed the set of cases considered “politically relevant” (which incorporates contiguity) in the politically relevant variable in the “variables” tab.  A “change contiguity” button has been added under the politically relevant variable selection to allow the user to set contiguity either there or on the population of cases tab.
  • Documentation updated.

Version 1.18

  • Bug Fix:  “joiner” variables output incorrectly for years from 1993 forward.  After the MID data ends in 1992, EUGene was outputting “-9” codes for “joiner_i” and “joiner_t” but was not putting a separator (tab, comma, or space) between them.  This led to data from 1993 forward being read incorrectly by statistical analysis programs.  These variables are now separated by the appropriate separator character.
  • Bug Fix:  If the user chose to create a data set where the variables were separated by commas, the Stata command file generated by EUGene was incorrect.  The output command file is now correct.

Version 1.17

  • Notes:  Mismatch between COW and Polity III country codes.  The Polity data sets do not always use the same country code numbers as COW.  Earlier versions of EUGene corrected for some differences, but missed the mismatched coding for Serbia and Italy.   COW codes pre-WWI Serbia and Yugoslavia as the same country code, 345. Polity III codes Croatia as ccode 344 up to 1915, and then Yugoslavia (345) from 1921 forward. No 344 is coded after 1915 (until 1991); no 345 is coded before 1921. We now use the country code 345 for both of these up to 1915, so only ccode 345 is coded from 1800 to 1990.  Also, COW codes pre-unification Italy and Italy 1861+ as the same country code, 325. Polity III codes Sardinia as ccode 324 up to 1860, and then Italy from 1861 forward. No 324 is coded after 1861; no 325 is coded before 1861. We now use the country code 325 for both of these, pre and post 1861.
  • Bug Fix:  EUGene v1.15 contained an error in reading country names from the latest COW listing of state membership.  This resulted in country names appearing incorrectly in the window that allowed users to select a subset of countries for output.  More importantly, this bug resulted in the appearance that South Korea did not exist in output from v1.15!  [this bug may exist in v1.14 and 1.13, but was not reported until noticed in version 1.15.]

Version 1.15

  • Bug Fix:  “Peace Years” computed incorrectly.  Versions 1.13 and 1.14 computed the “peace years” variable incorrectly.  This variable, a count of the number of years since the last MID between two countries, is used for creating spline variables to account for serial autocorrelation within time-series cross-section dta sets on conflict following the methods of Beck, Katz and Tucker (American Journal of Political Science, October 1998).  In v1.13 and 1.14, EUGene reset the peace years variable to “0” in the first year of a MID, rather than the year after a MID (or in the 2nd year of the MID if it is a multi-year MID).  This would lead to problems estimating coefficients on the peace years variable and associated splines.  The new version resets the variable one year later.

Version 1.14

  • Modification: MID-based criteria for excluding cases only applied if MIDs selected for output. When outputting dyad-year data, EUGene through version 1.13 would drop cases based on MID criteria (joiners, ongoing disputes, etc.) even if the user had not selected MIDs for output. This was not an error, as EUGene was designed from scratch to output dyadic data with selections on the basis of MIDs in mind. To overcome cases being dropped, users had to select “Keep disputes with ongoing MID” “Keep Target vs. Initiator Dyads” or other choices that would keep both directions of all dyads in the data set. This might be done, for instance, if the user wanted to output a set of all directed dyad years without considering MIDs. This procedure was somewhat awkward and was not transparent to all users, and so we have modified the program. Now, if the user has not selected MIDs on the “Variables” output tab, no cases will be dropped because of ongoing disputes, target vs. initiator status, or joiner status. The option to select various exclusions is no longer enabled unless the user selects MIDs for output.  Note: this change does not affect procedures for outputting country-years or dispute-years; country-years by definition cannot include dyadic MID selection, and dispute-years always include those selections as defined in the documentation.
  • Bug Fix/File update:  A new version of the file of country/capital locations (latlong.raw) has been included with this release.  The version of this file included with EUGene v1.13 did not properly update country system membership dates to go along with the 1997 COW system membership list;  this resulted in an error when generating distances between certain countries.  In addition, the new state of Palau was omitted from the earlier file.
  • Bug Fix: Missing values for Polity Data set to -9. Version 1.13 introduced a bug along with the expansion of years that came from using the COW 1997 system membership, which runs beyond the data included in Polity III data files. When years were requested for output outside the date range of the Polity III file, a “-9” was output for the missing value. Normally a “-99” “-88” “-77” or “-66” are used to indicate missing data for the Polity data (-99 when data is out of range, or a –88, -77, -66 as those data are used in the Polity data set) because a –9 is an acceptable value for many polity variables. This has been corrected so that a “-99” is output when polity data is missing.
  • Bug Fix: Capability values set to –1. For several country code / year combinations, EUGene v1.13 mistakenly reported a capability CINC score of –1. This was a result of a discrepancy between the COW capabilities data set and 1997 system membership data that was introduced in version 1.13. These errors occurred in years when according to the 1997 system membership list, a state is a state, but there are no capability data in the most recent COW capabilities data set. EUGene assumed that data would be present in all cases; when new data were not read, EUGene reported its internal initialized value of “-1” instead of a value of “-9”, the missing value code. This has been corrected, so that a “-9” is now reported.  [This problem occurred in ccode 135 year 1881, 135 1882, 42 1924, 620 1951, 812 1953, 552 1965, 990 1976, 990 1977, 990 1978, 57 1979, 990 1979, 990 1980, 990 1981, 990 1982, 990 1983, 990 1984, 990 1985, 990 1986, 990 1987, 990 1988, 990 1989, 990 1990, 990 1991, 990 1992, 990 1993.]

Version 1.13

  • EUGene will now use and read the 1997 release of the COW system membership list.   Version 1.12 (and earlier) used a 1994 version of the system membership data.
  • An option has been added for coding the dispute initiation variable.   When outputting directed dyads, you may now select to code as dispute initiators both true originators and either 1) no joiners, 2) joiners on the initiating side, or 3) joiners on either side.
  • The option to code joiners as initiators when outputting directed-disputed-dyad data has been clarified to indicate that only joiners who join the initiating side can be coded as initiators.
  • You may now create a “peaceyrs” variable as part of your output data set, on the “variables” output tab.  This can be used to create splines as suggested by Beck and Tucker (1998).
  • A minor bug in the treatment of ongoing disputes in years when COW does not code the state as a state has been fixed. There are some occasional cases where the COW MID data set includes a dispute between 2 states in a year that the system membership list (at least, the 1994 list) did not have both parties being a state.  For example, the MID data reported an ongoing dispute in 1917 between the US and the Dominican Republic, even though the COW system membership list listed the Dominican Republic as a state only through 1916.  Normally, EUGene should not print any information about 1917, because by its own internal logic it prints only information on dyad-years when both parties are states.  However, when a dispute started in a year when both parties were states, and continued into a subsequent when one or both were not states, EUGene still output the subsequent year.  This is now fixed, with the system membership list taking precendence even in a case like the 1917 case listed above (it would correctly no longer print now).  NOTE:  The source of this problem was a discrepancy between MID data and the system membership list.  Given the update to the system membership list that EUGene is now using, this problem should not arise.  However, if the lists ever do not match again, EUGene is corrected to rely on the system membership list.
  • In version 1.12, EUGene modified the list of major powers to include Germany as a major power through the period 1919-1925, even though the COW system membership list did NOT consider Germany a major power.  With the update to 1997 system membership data, this has been changed, and EUGene will not treat Germany as a major power for this period.  So, EUGene now relies solely on the COW system membership list for its date codings.
  • The documentation has been updated and clarified in some areas.
EUGene Copyright © 1997-2017, D. Scott Bennett and Allan C. Stam. All Rights Reserved
© 2020 Penn State University