Hypothesis 11: International policy interventions resulted in an improvement of the situation in South Africa
4 Research Design and Methods
4.3 Data Collection: The South Africa Interaction Data Set (SAID)
4.3.2 Source Selection
The Associated Press (AP) news wire serves as the main data source in this study. AP is a U.S.-domiciled news organization with 242 branch offices in 121 countries and is used as an information source by 1,700 newspapers and 5,000 television and radio outlets all over the world (figures as in 200630). Founded in 1846, the wire service supplies a steady stream of news around the clock to its domestic members, 8,500 international subscribers and commer-cial customers. In academia, the AP newswire is widely accessible via the electronic database LexisNexis.
As demonstrated in Figure 10 below, AP has the highest frequency of news stories on South Africa per year from all machine-readable news sources available from LexisNexis for the years 1977 to 1996. AP, Xinhua and The Washington Post cover the whole period, UPI and The New York Times are available from 1980, The Financial Times from 1982.
30 According to AP, http://www.ap.org/pages/about/about.html. Accessed 24 May 2006.
Figure 10: Number of South Africa News Stories per Year for Different Sources (LexisNexis)
The selected news stories include the key-word “South Africa!” either in title or lead sentence based on a search query on Lexis-Nexis.31
A comparison of the performance of wire services of AP, Xinhua and UIP on the one hand and the newspapers of The New York Times, The Washington Post and The Financial Times on the other reveals clearly that the wire services are preferable to newspapers in terms of story density. AP reports on average over 2,000 South Africa related stories per year be-tween 1980 and 1994, whereas The New York Times provides only about half as many South Africa stories per year over the same period. The Financial Times does slightly better than The New York Times (almost 1,300 South Africa stories per year), but is not available before 1982. UPI, available from September 1980, covers South Africa for the years 1981-1986 almost as intensively as AP, but the coverage decreases dramatically after 1986.32
Xinhua, the third English-language newswire that is available from LexisNexis back to the late 1970s needs some special consideration. The news agency, China’s central media agency and official press agency of the Chinese government, covered South Africa decently and increased its reporting significantly during the second half of the 1980s. For the years 1987 to 1992, the reporting was comparable to AP, in terms of number of stories per year.
31 The !-appendix indicates a wild-card for subsequent characters to catch stories including the term “South African”. I tested story frequencies also for additional key-words such as “apartheid”, “ANC”, or “Mandela”.
But it turned out that virtually every story containing one or several of these additional key-words was already picked up by the “South Africa!” search query.
32 United Press International (UPI) used to be one of the three biggest news agencies in the world, together with AP and Reuters, but has dwindled in size and importance over the last two decades. In the mid-1980s, the news agency was struggling to avoid bankruptcy and went through several ownership turnovers after 1986 (see, e.g., The New York Times, 10 November 1985, 21 February 1988, and 24 December 1989).
Comparing the Chinese and the U.S. news wire it seems interesting that reporting peaks in different years (AP in 1986, Xinhua in 1988). Different national interests in South Africa could be an explanation for these differences in reporting. Reeves, Shellman and Stewart (2004: 10) found in Cambodia, for example, that Xinhua reported intensively on the tance groups, whom the Chinese government actively supported. Such a focus on the resis-tance movement could be assumed for the case of South Africa as well since the Chinese government also supported oppositional forces in South Africa. Xinhua could therefore serve as a complementary source to an American news source like AP.
But, unfortunately, a use of Xinhua would involve many technical problems. From a formal perspective, Xinhua’s reporting is far from being consistently structured in the same manner as Anglo-Saxon media report. The validity of the machine coded data highly depends on such a structured and consistent report, as will be outlined in more detail below. Xinhua’s reporting style is not constant over time either but changed several times between 1977 and 1996. The linguistic quality of the news reports varies as well. Finally, the vocabulary used differs from news stories originating from natively English-speaking countries so that differ-ent actor and verb dictionaries would be necessary to code the source. Overall, Xinhua is a problematic source for automated coding and can therefore not be considered a useful source for the purpose of this study.
Figure 11: Number of South Africa News Stories per Year for Different Sources (Factiva)
The selected news stories include the key-word “South Africa*” either in title or lead sentence based on a search query on Factiva.
Figure 11 illustrates the coverage of South Africa by different sources obtained from Factiva. Reuters reported much more densely than the other available news sources for the period 1977 to 1996. Even AFP, the other big newswire service available from Factiva, pro-vides significantly less South Africa stories than Reuters. Unfortunately, Reuters is not acces-sible anymore for the time prior to May 1987.
Because Reuters used to be widely used in event data research in the 1980s and 1990s, scholars tested event data coded from Reuters extensively for its validity (Reeves et al.
2006; Gerner et al. 1994; Huxtable 1997; Huxtable and Pevehouse 1996; Schrodt and Gerner 1994; Schrodt et al. 2001). Event data coded from Reuters displayed the political reality fairly well and worked technically fine within the machine coding system. I will therefore use Reuters both as a complementary source to AP and as a validation of AP’s performance as a main data source for this study. But before I turn to this validation of the sources used in this study, I introduce the applied coding scheme and machine coding system in more detail.