Planned comparison of coalition versus combined non-coalition coverage revealed significant differences by the measures of “Right,” “Saddam,” “Threat,” “Dictator,” “Liberate,” “Good,” “Heroes,” and “ProAct.” Only the “Support” measure was insignificant.
Table 2: Planned comparison
(coalition vs. combined non-coalition European and Middle East)
This data confirms Hypothesis 1 and 2 by disproving the null and proving there is a media bias favoring the publication’s host country position for these measures.
Post-hoc tests comparing coalition and non-coalition European coverage revealed significant differences by the measures of “Peace,” “Inspect,” “Protest,” and “Oil.”
Table 3: Post-hoc test*
(Coalition vs. non-coalition European)
*Post-hoc test on the measures of ProUSA, WMD and Civilian were not significant.
The third prediction involved time and posited that over time, differences would occur in the tenor and tone of the news stories in all regions. Note omnibus Wilks’ Lamda results (main effect) F()=3.417, p<.05. Subsequent univariant for those dependent measures with an N of 10 or more indicated significant differences (one time) favoring the coalition position. Difference more significant on the measures of “Liberate,” “Good,” “Right,” “Heroes,” “ProAct,” “Dictator,” “Threat,” “Saddam,” “Support.”
Table 4: Wilks’ Lamba test
(Differences in tone and tenor)
This MANOVA test confirmed Hypothesis 3 by disproving the null. The population means differences statistically proved media’s tone and tenor changed as the war progressed.
Dependent measures cannot statistically reveal the spiral of silence effect, however when melded with relevant polling data, the researchers were able to answer the two research questions with a small degree of certainty.
Table 1: Polling data