from: Wormer, H. (2007) Figures, statistics and the journalist: an affair between love and fear. Some perspectives of statistical consulting in journalism, section 4 

>> read full text, pdf file

 

 

The first steps of CAR date back to the second part of the 1970s with the foundation of the society of Investigative Reporters and Editors (IRE) in the US. Now there is even a National Institute of Computer Assisted Reporting (NICAR) at the Missouri School of Journalism which stores and distributes databases to journalists systematically. Since 1989 there has been at least one Pulitzer Prize story every year generated by CAR techniques (Redelfs 2001). One example was the story of fraud in the election of the mayor of Miami: Journalists compared the voting register database with databases of estate agents to find the addresses, where five or more votes by mail came from one household or even from estates where no house existed (Wessel 2003).

However, the total number of database driven stories cannot accurately be estimated. According to the Acting Director of IRE in the past four years, more than 700 investigations that involved significant CAR work have been posted on the IRE website that tracks ongoing investigations. These represent a small percentage of the total number of stories that involve CAR in some way for a fact, context or graphic (personal communication, Houston 2007). Examples of database driven articles can be found in an overview of general investigative articles at IRE/extraextra or IRE/resourcecenter (search: ‘database’).

Getting started: integrity check of database

As in statistics, in database driven CAR the quality of the results can only be as good as the original database. At the same time it may already be an interesting story if a journalist discovers that an important official database which is distributed by government agencies has serious deficits and faults. CAR-experts like Houston (2007) point out that it has always to be ‘assumed that every database is in some way incomplete or imperfect and it is important to determine what its weaknesses are. ( . . . ) Integrity checks include checking for typographical errors, inconsistent entry of place names, amounts or numbers outside the normal range expected, and recalculating calculated fields or totals.’ A collection of databases for journalists can be found at IRE/datalibrary. Beside collecting and analyzing official databases from governmental and similar institutions some journalists also create databases by themselves for their own investigations. Of course, in both cases assistance from professional statisticians could be very helpful – maybe in the sense of further developing CAR into statistics assisted reporting (SAR), as proposed by one of the reviewers of this article.

4.2 CAR in Germany compared to the US

 

Although there is no evidence that the average US journalist is in general better qualified for dealing with numbers and statistics than his colleagues in Europe, at least among highly qualified journalists CAR seems to be more popular in the US than in many European countries. One reason may be a generally higher attraction to computer-based tools in the US; another one may be the Freedom of Information Act which traditionally forces US government organizations to transparency with facts and figures. A third reason probably is the much higher degree of specialization of US journalists, making it easier to establish a subspecies of journalists, ‘number crunchers’ who analyze databases to generate stories, whereas such behaviour is more or less unpopular among German journalists.

CAR under way: favourable legislation and special education 

Nevertheless the association of investigative reporters in Germany (Netzwerk Recherche) has already started some efforts to make the field more popular in Germany. Moreover since 2006 the preconditions for CAR have improved also in Germany by the nationwide Informationsfreiheitsgesetz which – however, not as strict as the US Freedom of Information Act – will provide better access to government databases for statistical (and journalistic) analysis. One opportunity to get more media interested in the possibilities of such a systematical analysis of masses of the feared figures (see full text, section 2) are tailored study courses combining skills of journalism and statistics (e.g., the science journalism course at the University of Dortmund). As science journalists also in general have by training the closest relationship to figures and formulas they may play a leading role here in the mass media. Another opportunity could be a systematic cooperation between journalists and statistical consultants.

Recently, there have been some encouraging developments: In June 2007 the biggest German news agency (Deutsche Presse Agentur, DPA) announced its intention to build a subdepartment (RegioData) in autumn 2007 which will specialize in generating and enforcing database driven articles. This is even more remarkable considering the fact that the most important costumers of DPA are regional newspapers.