Comparative Extension Projects: Belgium

1
Comparative Extension Projects Belgium
Project History 2001 2003 Joint project of 4
universities (KUL, UCL, UA, VUB). During this
period the data collection took place. The final
report was delivered in May 2004 (Walgrave, S.,
De Winter, L. and Nuytemans, M. (Eds.) (2005)
Politieke agendasetting in België (1991-2000).
De moeilijke dialoog tussen publieke opinie,
media en het politieke systeem, Academia Press,
Gent). 2005 2007 no new data collection, but
the project received further funding to be
continued. Current funding ends in 2007, however
we submitted a new grant proposal to the Belgian
Fund for scientific research.
Codebook As we were unaware of the existence of
the American Policy Agendas Project when we
started coding in 2001, we did not use the US
topic coding system. We drew upon another
self-developed system based on the so-called
EUROVOC thesaurus. We devised a coding scheme
with 143 topic codes, and 36 main categories. An
exploratory conversion table translating the
American in the Belgian coding scheme has been
developed.

• Media Agenda
• Largest agenda (more than 160.000 news items
  coded).
• 4 TV channels, 5 newspapers.
• Both French-speaking and Flemish speaking media
  were coded.
• Newspapers
• In Flanders, a tabloid (Het Laatste Nieuws), and
  two quality newspapers were selected, De
  Standaard being the largest in this category.
• In the French-speaking community, the largest
  quality newspaper (La Libre Belgique) was
  coded, as well as the largest newspaper (Le
  Soir).
• Four days per week (Mon Wen Fri Sat)
• The front page of the newspapers was coded.
• Core information
• Smallest time unit one day
• Unit of Analysis newspaper article (1 code /
  article)
• Time span coded 1991 2000
• N 66607
• Saliency size of article (large / medium /
  small)
• TV-news broadcasts
• In Flanders, VTM and VRT are the only two
  companies who where continuously showing
  news-broadcasts for a large public during the
  research period.

• Government Agenda
• Ideally every item on the agenda of the Council
  of Ministers between januari 1st, 1991 and
  December 31st, 2000.
• Obtaining original data was impossible.
• Solution coding of weekly magazine Feiten,
  which contains all press releases following the
  meeting of the Council of Ministers.
• Issues with the data
• Not a perfect match with the actual agenda (80
  similarity). This is the biggest problem with
  this datafile.
• A few extra subcodes were added to code 71 for
  this agenda.
• Fourth state reform in 1993 cuts the research
  period in two, since the division of powers
  changed (e.g. Environmental policy was
  transferred to the regions).
• Core information
• Smallest time unit one week
• Unit of analysis one press release
• Time span coded 1991 2000
• N 6296
• Measuring saliency not available

• Possibilities for automatic coding
• Source of the coded data are problematic almost
  no electronic versions available, or in an
  inappropriate format. Almost everything was done
  manually, even the enormous media database was
  compiled by skimming through paper versions of
  the newspapers.
• We decided to focus on the media agenda
• Largest N of all the agendas
• Therefore most diverse
• Best possibilities to (somewhat) automate the
  recoding (see below)
• Preparing data for the automatic coding
• Full text of the newspaper articles in the years
  1999 2000 is placed in a database. Then a match
  is made based on the title of the article, which
  was originally coded.

• Civil Society Agenda
• Based on data of previous work on Protest and
  Movements, which was adjusted and recoded
  following the EUROVOC codebook.
• Data was gathered using three sources the policy
  archives (main source), and the protest marches
  mentioned in the newspapers De Morgen and Le Soir
  (secondary sources).
• Core information
• Smallest time unit one day (day of the protest
  march)
• Unit of analysis one protest march
• Time span coded 1991 2000
• N 4000
• Measuring saliency using the amount of
  protesters. However, different sources report
  different amounts. Therefore police count as a
  baseline figure, if this is not available, media
  count multiplied by a coefficient

• Belgian Parliament Agenda
• Core information
• Smallest time unit one day
• Unit of analysis one activity (written / oral
  question, interpellation, initiative or law)
• Time span coded 1991 2000
• N 39240
• Measuring saliency not available

• Government Agenda
• Ideally every item on the agenda of the Council
  of Ministers between Januari 1st, 1991 and
  December 31st, 2000.
• Obtaining original data was impossible.
• Solution coding of weekly magazine Feiten,
  which contains all press releases following the
  meeting of the Council of Ministers.
• Issues with the data
• Not a perfect match with the actual agenda (80
  similarity). This is the biggest problem with
  this data file.
• A few extra subcodes were added to code 71 for
  this agenda.
• Fourth state reform in 1993 cuts the research
  period in two, since the division of powers
  changed (e.g. Environmental policy was
  transferred to the regions).
• Core information
• Smallest time unit one week
• Unit of analysis one press release
• Time span coded 1991 2000
• N 6296
• Measuring saliency not available

• Party Agenda
• Based on the original party manifestoes, as
  published in the year of the Federal elections
  held within the research period (1991, 1995 and
  1999).
• Coded using a slightly modified version of the
  EUROVOC codebook.
• Core information
• Smallest time unit every 4 years (each manifesto
  has a date, but theyre only made every 4 years)
• Unit of analysis one sentence
• Time span coded every manifesto published during
  the research period was coded (1991, 1995, 1999)
• N 36030
• Measuring saliency not available, since each
  manifesto is treated as being equal