Page 288 - Comparing Political Communication Theories, Cases, and Challenge

P1: JZL/KDD/KAB P2: JZZ
0521828317c11.xml CY425/Esser 0521828317 June 8, 2004 22:9

Wolfgang Donsbach and Thomas Patterson

when numerous questions are asked, a faulty question will stand out
because the responses it yielded are markedly inconsistent with other
distributions. Nonetheless, the burden on survey research is greater in a
comparative study than in a single-country study.
For this reason, a comparative survey requires a substantial initial
investment of time. It took us roughly a year to develop our question-
naire. As a first step, the two authors conducted in-depth interviews
with a dozen U.S. and foreign correspondents in Washington, DC to
explore the various theoretical dimensions of the study. Original drafts
of the questionnaire were then prepared by the principal authors and
were reviewed by scholars in the countries where the survey would be
conducted. English-language versions of the country-specific question-
naires were then pilot tested. The U.S. version was tested on American
journalists while the other versions were tested on foreign correspon-
dents working in the United States. Final versions of the questionnaires
were then developed and translated. A double system of translation was
employed. After the questionnaire was translated into another language,
it was reverse translated to determine its fidelity to the original. Alto-
gether, the survey progressed through nearly two-dozen drafts before
the final version was settled upon. Anything less than this type of effort
would have subjected our study to substantial errors of inference. There
is no doubt we made some errors of this type anyway. But they did not
occur because of hasty execution of the survey.
Although we were careful, we were not tentative. We took advantage
of the fact that journalists are an elite population and could handle a
complex survey, which enabled us to treat it as a flexible instrument.
It would have been enormously expensive for us to conduct a content
analysis of news outputs in five countries of local and national broadcast
and newspaper outlets. Our quasi-experimental news decision questions
allowedusthroughsurveyresearchtoaddressissuesthatnormallywould
have required a content analysis component.
The study also indicates that a survey of this type can be a means
of obtaining system-level measurements. For some of our questions, we
used our survey respondents as “expert judges.” Although this chapter
does not provide an example, we had our respondents assess, for exam-
ple, the Left-Right positioning of major news organizations within each
country.
Finally, our study indicates that comparative surveys of journalists do
not require large foundation grants. Although we had some financial
support, the fact that we were able to employ a mail survey kept the costs

268