During the content analysis, the researchers compiled data to disprove the null of three hypotheses:
The analysis assessed the dependent variable (tone/tenor/valence) against two independent variables (time and region of publication). Of the 18 dependent variable measures, 9 were determined to be of significance with a MANOVA test due in part to N of 10 or more. When Middle-Eastern newspaper data is combined with non-European newspaper data and compared to United States newspapers, using an ANOVA test, the null hypothesis is rejected in 4 dependent measures of the hypotheses. Two measures included the 3-day periods sampled, and the following regions: United States, non-coalition European, and non-coalition Middle-Eastern. The 9 measures determined to be significant with the MANOVA test contained at least 10 articles for comparison from each period.
Hypotheses 1 and 2 involved media bias of coverage with regard to region of publication. The omnibus test did not reveal a significant main or interaction effect involving publication, probably due to insufficient power. Further analyses were conducted and coalition coverage was compared to non-coalition coverage. Where the N of non-coalition Middle Eastern publications was insufficient, the two non-coalition groups were combined and a planned comparison was computed. Where it was sufficient, a one-way ANOVA (post-hoc test) was computed comparing coalition and con-coalition European coverage.