In examining the results of the data collected from coding of the newspaper articles, several trends and statistical differences became apparent. Team members analyzed the information statistically using SPSS software to examine the relationships and differences yielded within the data. The team examined three independent variables: reporter status, conflict, and focus of story. Initially, a 3 (embeds, unilateral, staff compilation) x 3 (1991 Operation Desert Storm, 1999 Kosovo Conflict, 2003 Operation Iraqi Freedom) MANOVA was computed in the dependent variable of attitude and feeling.
The omnibus MANOVA did not indicate an interaction effect involving reporter and war, as was predicted in H1, or a main effect for reporter, probably due to insufficient power (Power for the interaction test was .712). However, because theory predicted such an interaction, planned comparisons were computed comparing coverage of the 2003 Operation Iraqi Freedom to combined coverage for the 1991 Operation Desert Storm and the 1999 Kosovo Conflict, first for embeds and then for unilateral and staff compilation reporters. Results for embeds indicated that coverage of 2003 Operation Iraqi Freedom was more positive than with the other conflicts on the dependent variable of attitude, F(1, 115) =11.09, p<.01, and feeling, F(1, 115) = 13.185, p<.01. Results for unilateral and for staff compilation were not significant. These means are displayed in Table 1.
The omnibus results revealed a main effect for war, F(4, 228) = 3.61 p<.01, eta2 = .06. Subsequent unilateral tests indicated significance for the dependent variables of attitude, F(2, 115) =3.43 p>.05, eta2 = .06, and feeling, F(2, 115) = 6.97, p<.01, eta2 = .11. Subsequent Scheffe post-hoc tests revealed that for the dependent variable of attitude: coverage of 2003 Operation Iraqi Freedom was more positive than coverage of 1999 Kosovo Conflict (p<.05); for the dependent variable of feelings, coverage of both 2003 Operation Iraqi Freedom (p<.01) and the 1991 Operation Desert Storm (p<.01) were more positive than the 1999 Kosovo Conflict. These means are displayed in Table 2. Thus, H2 was partially supported.
The team examined the focus of the stories using a one-way ANOVA. The results revealed a significant on dependent variable of feeling, F(2, 137) = 7.21, p<.01. Post-hoc tests indicated that troop-focused stories generated more positive feelings toward military members than command-oriented (p<.01) or neither (p<.05). See Table 3.