• RSS feed
  • Tweets
  • Changes




Final country selection

We have completed the country selection and include 37 EU and OECD democracies with a parliamentary or semi-presidential system. We removed countries from the experimental version that we have not fully covered (alb, chl, gdr, kor, lie, mex, nir, sco, usa, wal).

Mon 22 December 2014 — change data
ParlGov 2014 clean-up

We are in the process of cleaning up the data in ParlGov. Our goal is to provide reliable data for parties, elections and cabinets in "all EU and OECD democracies". We are almost there with only some OECD countries and a few election results (EU members) of parties that won no seats missing. However, the documentation still leaves room for improvement and there need to be some more routines that check for data consistency.

Currently, we are completing a new version of the webpage and are cleaning up the code that runs the page and calculates the data. There has been a lot of cruft over the years and often new tools provide more straight forward solutions. ParlGov needs some streamlining to be maintainable for the years to come.

We trim down on our earlier aspirations and are removing all countries that are neither EU nor OECD members. This affects data in the experimental version only. There will also be some changes to the linking of political parties.

We will provide news entries once we have completed the respective clean-ups.

Again, none of this should effect information in the development version of ParlGov but only additional data in the experimental version (guest login required).

Sat 8 February 2014 — change data
Recoding party families and imputation of missing party positions

We have completed recoding our party families. All parties are now classified into one of the main categories first and additional party families are recorded in the table 'party_family' only. Our main categories are primarily based on a two dimensional perspective (state/market and liberty/authority) of the political space and follow other typologies of party families. These party families are now also used to impute missing party positions by the mean value for the respective party family.

Mon 1 July 2013 — change data
Germany CDU/CSU party group

In Germany, CDU and CSU run independently in elections without competing at the state level. Both parties form a joint parliamentary group in the German parliament after the election. This fact has now also been included into ParlGov. CDU and CSU are kept distinct in our election results but the joint party group is recorded in our parliament change data. As a consequence most of the cabinets including the CDU/CSU are no surplus coalitions anymore.

Thu 14 February 2013 — change data · Germany
Party clean up

Following the recent update of coding rules for our election results, we have also cleaned up the parties included into ParlGov. Fur this purpose, we removed minor parties without election results according to our new criteria (1.0% vote share or 2 seats). A more detailed summary of coding rules for parties, election results and cabinets will follow over the next month.

Tue 2 October 2012 — change data
Updated coding rules election

ParlGov includes information about the number of seats and the share of votes for all parties in parliament and elections we cover. For some countries and new elections, we also include the number of votes and some of the parties that won no seats. We used to include all parties that won more than 0.5% vote share but have now increased the threshold to 1.0%. We have learned that many of the parties below the 1.0% threshold run only in one election and it was difficult to gather reliable information about these parties. As a consequence, we increased the threshold for including election results and removed observations below the threshold. We cover only those parties with less than 1.0% vote share that won seats in parliament or the largest party winning no seats in the election.

Wed 13 June 2012 — change data
Update calculation party positions

ParlGov party positions are based on various party expert surveys (Castles/Mair 1983, Huber/Inglehart 1995, Ray 1999, Benoit/Laver 2006 and CHESS 2010). These positions are just the mean value of all observations available from these sources in a particular dimension (left_right, state_market, liberty_authority, eu_anti_pro).

There was a bug in the scripts that converted the data from Benoit/Laver and not all positions from the data set were included into the calculation. We became aware of the issue, when someone pointed out the extremely moderate left/right position of the German far-right NPD in ParlGov. We have now solved the issue and include data on all relevant dimensions from Benoit/Laver into the calculation. This also solves the issue of having information in the left/right dimension for some minor parties instead of using the state/market dimension.

While updating the position data we have also renamed the variables for the EU dimension from ‘eu_pro_contra’ into ‘eu_anti_pro’ since lower values on the scale present eurosceptic positions.

Tue 1 May 2012 — change data
ParlGov II -- database refactoring

The second major change for the next ParlGov version will be a major refactoring of our database design. The causes for the need to refactor the structure of our database are similar to the previously described plans to relaunch the web design. Some of the fundamental design decisions were made at the early stages of the project and limit its growth now. Over time, our experience and technical expertise has grown and we are convinced that the benefits of a redesign outweigh the (transformation) costs.

Most fundamental will be the overhaul of all IDs. In the next version or ParlGov, there will be newly defined identifiers for all observations, most importantly for countries, parties, elections and cabinets. Previously, we had unique identifiers per country only and some of the ID variables were based on other observations and contained meaningful information. We will change that to a system of IDs were all identifiers are unique for every variable and are just numerical identifiers without containing other information. In database terminology, we are moving from a system of composite (natural) primary keys to (surrogate) primary keys.

There will also be a clean-up for some table and variable names and the new names are often more explicit than the previously used shortened versions. In addition, we are changing from camel case VariableNames to underscore variable_names. Finally, there are some updates on the table structure.

These modifications require users of ParlGov to update the scripts they have written to work with ParlGov data. We are convinced that it will be for the better. The new database design is easier to understand and allows us to integrate innovative aspects into future versions more easily.

Tue 28 September 2010 — change data