CSES Module 3 Data Set Errata
Posted: Mar 27, 2013
Updates made to the Full Release of March 27, 2013, since the Advance Release of May 31, 2012
The Full Release of March 27, 2013 contain additions, updates, error corrections, and improvements to the third Advance Release of May 31, 2012. Please do not use files from the previous release with files from the current release. Variable names and attributes, as well as the data, have sometimes been changed.
Thirteen new election studies have been added to the file, bringing the number of election studies appearing in CSES Module 3 to 50 election studies in total. The studies added as of the March 27, 2013 Full Release are: Canada (2008), Chile (2009), Czech Republic (2010), Finland (2011), Latvia (2010), Netherlands (2010), Norway (2009), Peru (2011), Philippines (2010), Romania (2009), South Africa (2009), Turkey (2011), and the United States (2008).
Some of the changes made to election studies which appeared in previous releases are listed here, though this is not intended to be comprehensive...
Micro section improvements:
- In the Austria (2008) election study, the code for "Don't Know" was separated out from "Missing" for several variables.
- In the "degree of closeness to this party" variable (C3020_4) for the Brazil (2006) study, 96 previously valid answers have been set to missing. For these answers the like-dislike questions were previously being used as a proxy for the 96 cases.
- In the Czech Republic (2006) study, corrections were made to the election date (C1016-C1021), and codes for the party leaders were adjusted (C3008_2).
- Vote variables for the previous presidential election for the France (2007) study had been collapsed by accident, and now appear their original, detailed format.
- The distribution for the education variable (C2003) in the Germany (2005) study included a recoding error and has been corrected. Please note that the education variable (C2003) distribution for the Germany (2009) study was verified to be correct as provided, and there is a special election study note in the codebook describing the circumstances.
- The "date questionnaire administered" variables (C1026-C1028) were deleted from Portugal (2009) due a techincal problem in their collection.
- The distribution for the household income variable (C2020) in the Switzerland (2009) election study included a recoding error and has been corrected.
Variables newly available for a election study:
- In the Brazil (2006) study, party affiliations previously missing for candidates were identified, and information added to the C3023-C3030 series of variables as a result.
- Data are newly available in the Ireland (2007) study for these variables: C3001 series and C3002 series.
- Data are newly available in the Poland (2005) study for variable C1023.
- Data are newly available in the Poland (2007) study for these variables: C1023, C2031, C3001-C3002 series, C3007 series, C3008 series, C3020 series, and C3023 series.
- Data are newly available in the Slovenia (2008) study for these variables: C1024, C1025, C2031, and the district-level data section.
- Data are newly available in the Spain (2008) study for variable C3036_1.
- Data are newly available in the Thailand (2007) study for variable C2031.
- In the Taiwan (2008) study, data are newly available for detailed age (C2001) and additional district-level data are newly available for the presidential election.
Data added back in to the dataset:
- Added back in information for variables C2012-C2014 and C2017-C2019 for election studies where respondents and spouses outside of the labor force were consistently eligible to answer the follow-up questions.
- Added back in information for variables C2015-C2019 for election studies where respondents who did not indicate living together or married (C2004 > 1) were consistently eligible to answer the follow-up questions.
- Added back in information for variables C3029_LH and the C3030 series for election studies where non-voters (C3021_1 > 1) were consistently eligible to answer the follow-up questions.
- In the "most important problem" questions (C3001 and C3002 series), previously non-missing (substantive) answers that were mentioned twice by a respondent had the second mention deleted in the dataset, but the second (duplicate) mention has been added back in as of this release. Likewise, answers for C3003 was added back in for respondents even where the answer seems not consistent with their answer in C3002.
Macro section improvements:
- The macro variables for "government composition before election" (C5009 series) for Estonia (2011) were coded incorrectly and have been corrected.
- In the "geographic concentration of violence" variable (C5027), a coding error in Netherlands (2006) was recoded to "no election violence".
- Improvements were made to the macro variables for cabinets for the following election studies, to make the coding more consistent cross-nationally: Australia (2007), Estonia (2011), Finland (2007), Germany (2009), Greece (2009), Israel (2006), Slovakia (2010), Slovenia (2008), South Korea (2008), Spain (2008), Sweden (2006), and Thailand (2007). At the same time, two mistakes were corrected in the variables for "party of the prime minister before the election" and "after the election" as well - the correct party was listed in the codebook, but the wrong numeric code was in use in the data.
- The electoral tiers for Greece (2009), Estonia (2011), and Slovenia (2008) were changed.
- Macro variables were updated from third-party sources (Freedom House, World Bank, etc.) for election studies where the information was previously missing.
- The method by which election date irregularities (C5025) are coded has been streamlined, and dates corrected in the C5024 series for Greece (2009) and Estonia (2011).
- In the macro section, the counts for the number of of overhang seats has been improved in the Germany (2005), Germany (2009) and New Zealand (2008) studies to count the size of the lower house, and a mistake was corrected in the Finland (2007) study.
- In the macro section, some changes were made in electoral formulas (the codebook was deterined to be correct, but not the data, for a handful of countries).
If you have questions or concerns, e-mail us at firstname.lastname@example.org and we would be pleased to help.