This investigation featured two components. A content analysis was conduced in order to describe the content of milblogs: specifically, their credibility, emotional content, and their depiction of the U.S. military and its personnel.  Then, an experiment was conducted to compare the effects of milblog exposure versus military and civilian on-line news venue exposure in terms of their impact on people’s perceptions of the credibility of news stories, the emotional impact of news stories, and the influence of news stories on user’s involvement and overall attitude about the U.S. military.
Content Analysis
            Research Questions 1, 2, and 4 simply probed the content of milblogs.  The results of the content analysis address these questions.  Since the questions simply ask the nature of the content of milblogs, the analysis was limited to descriptive statistics.
            Research Question 1 asked about the overall tone and depiction of the military portrayed in milblogs. To answer this, researchers examined the overall means of two measures: one that assessed the overall tone of coverage toward the military and the other that measured the trust in U.S. military personnel. Both measures were 7-interval scales with 4.0 as the absolute midpoint.
            The overall tone of milblog coverage was 4.37 (n=528), indicating a very weak positive valence. With 4.0 as the mid-point, 4.37 suggests a near-neutral overall depiction of the military in milblogs. The results were similar on the measure of trust in military personnel. The mean trust score was 4.42 (n=528), suggesting a weak positive depiction. Of the individual milblogs examined in the content analysis, some were more positive than the overall mean, including
Maj. K (M=5.51, n=22); 365 and a Wake Up (M=5.48, n=13); and Sgt. Hook (M=5.05, n=10).
All of the other milblogs were neutral to slightly positive in tone, including A Line in the Sand (M=4.59, n=11), Ma Deuce Gunner (M=4.46, n=14), Mudville Gazette (M=4.38, n=202), Blackfive (M=4.35, n=103); Lt. Smash (M=4.33, n=108), 2Slick (M=4.19, n=8); and Froggy Ruminations (M=3.99, n=37). We are confident in the overall results. However, the results of individual milblogs are unstable due to differences in the number of stories coded across sites.
Descriptive statistics for the tone, trust, and credibility of the individual milblog sites are shown in Table 1.
            Research Question 2 asked whether milblogs communicate a credible message. To assess this question, researchers examined the overall mean of the news credibility measure. It was a 7-interval set of scales with a 4.0 midpoint. The overall credibility of milblogs was 3.96 (n=528), which is very close to the midpoint.
            Finally, Research Question 4 asked whether milblogs communicate an affective message and, if so, which emotions are more prevalent. To evaluate this question, researchers used mean scores across nine emotions. The emotion measure consisted of 0-6-point scales with 3.0 as the midpoint. The 528 cases revealed very little emotional content. Emotional content, in order from greatest to least emotion, were: pride (M=0.33), anger (M=0.25), happiness (M=0.18), surprise (M=0.13), sadness (M=0.13), puzzlement (M=0.11), fear (M=0.06); contentment (M=0.04), and humor (M=0.04). The results indicate that milblogs did not elicit much emotion (see Table 2) and, what little emotion was revealed consisted of more pride and anger than others.
The emotional content of the individual milblog sites are shown in Table 2.
Experimental Results
Research questions 3, 5 and 6 addressed differences between milblogs and military on-line news or civilian on-line news stories. To assess these questions, a 3 (on-line site: milblog, military news, and civilian news) x 3 (topic: Iraqi constitution, IEDs, and Iraqi military forces) MANCOVA was computed on the dependent variables of: attitude toward military, involvement, credibility of news source, and emotional response (surprise, puzzlement, anger, sadness, fear, pride, humor, happiness, and contentment). Covariates included gender and use of blogs for international news.
The omnibus results indicated significant differences for the covariate of gender, Wilks’ λ F(12, 293) = 5.00, p < .001, partial eta2 = .17. Subsequent univariate tests revealed significant differences for the covariate gender on the dependent variables of attitude toward the military F(1, 315) = 2.92, p < .10, partial eta2 = .01; and the emotions of surprise F(1, 315) = 13.37, p < .001, partial eta2 = .04, puzzlement F(1, 315) = 33.92, p < .001, partial eta2 = .10, anger  F(1, 315) = 4.81, p < .05, partial eta2 = .02, sadness F(1, 315) = 26.38, p < .001, partial eta2 = .08, and fear F(41, 315) = 30.02,  p < .001, partial eta2 = .09. All valences were positive, thus indicating that females were more positive toward the military and manifested greater emotional response to the news stories. There were no main effects for the covariate use of blogs for international news, but subsequent univariate tests for the covariate revealed significant differences for the dependent variables of surprise F(1, 315) = 8.08, p < .05, partial eta2 = .03, puzzlement F(1, 315) = 5.41, p < .05, partial eta2 = .02, and sadness F(1, 315) = 6.92, p < .01, partial eta2 = .02. Those who used blogs to stay informed about international news were more likely to feel emotion.
The omnibus MANCOVA also revealed significant results for the independent variable of experimental condition, Wilks’ λ F(24,586) = 2.04, p < .01, partial eta2 = .08; and topic Wilks’ λ F(24, 586) = 3.99, p < .001, partial eta2 = .14; and an interaction of topic and experimental condition Wilks’ λ F(48, 1130) = 2.01, p < .001, partial eta2 = .08.
Subsequent univariate tests revealed significant differences for the independent variable experimental condition on the dependent variables of surprise F(2, 315) = 3.05, p < .05, partial eta2 = .02, anger F(2, 315) = 8.57, p < .001, partial eta2 = .05, sadness F(2, 315) = 6.43, p < .01, partial eta2 = .04, fear F(2, 315) = 3.62, p < .05, partial eta2 = .02, and happiness F(2, 315) = 2.60, p < .10, partial eta2 = .02. The source of the news stories impacted emotional response to them.
This independent variable was most central to this investigation, which sought to determine whether milblogs, compared to on-line military and civilian news venues, exert different effects in overall tone, credibility, or emotional response. The univariate results  indicated no main effect differences involving either tone or credibility.  However, results revealed main effects involving a number of emotions.
Scheffe post-hoc tests were used to find the where the specific differences were located. Civilian on-line news sources (t(217)=4.84, p<.01) and milblogs elicited more surprise than military on-line news sources (t(214)=4.27, p<.01). Civilian on-line news sources elicited more anger than military on-line news sources  (t(216)=8.17, p<.01) and milblogs  (t(218)=5.18, p<.01). Milblogs elicited more anger than military on-line news sources  (t(214)=3.00, p<.01). Civilian on-line news sources elicited more sadness than military on-line news sources  (t(217)=7.67, p<.01) and milblogs  (t(219)=5.22, p<.01). When looking at the dependent variable fear, civilian on-line news sources elicited more fear than military on-line news sources  (t(217)=5.60, p<.01); and milblogs elicited more fear than on-line military news sources  (t(214)=3.45, p<.01).
There were several significant differences depicted by subsequent univariate tests for the independent variable of topic on the dependent measures of attitude toward the military F (2, 315) = 3.58, p < .05, partial eta2 = .02, anger F(2, 315) = 5.25, p < .01, partial eta2 = .03, sadness F(2, 315) = 13.17, p < .001, partial eta2 = .08, fear F(2, 315) = 13.55, p < .001, partial eta2 = .08, humor F(2, 315) = 4.04, p < .05, partial eta2 = .03, and happiness F (2, 315) = 9.17, p < .001, partial eta2 = .06. Topic affected overall attitude and emotional response to the news stories.
Scheffe post hoc tests were also used to examine the specific differences in topic (IED, constitution, or Iraqi forces). Attitude toward the military was more positive for those in the constitution condition (t(213)=4.55, p<.01) when compared to those in the IED condition. Attiude toward the military was also more positive for those in the topic condition of Iraqi forces when compared to IEDs (t(219)=4.45, p<.01). The omnibus test has depicted significant differences on the emotions of anger, sad, fear, humor, and happiness. Scheffe post hoc tests revealed that participants in the IED topic condition experienced more anger than those in the Iraqi forces condition (t(218)=6.45, p<.01) as well as those in the constitution condition  (t(215)=2.90, p<.01). Participants in the IED condition also felt more sadness when compared to those in the Iraqi forces condition (t(219)=9.33, p<.01) and those in the constitution condition (t(216)=8.56, p<.01). The IED condition also elicited more fear when compared to the constitution condition  (t(216)=9.20, p<.01) and the Iraqi forces condition  (t(219)=9.10, p<.01).
Humor was elicited more when reading stories about the constitution when compared to the IED condition (t(216)=5.83, p<.01) as well as when reading stories about Iraqi forces compared to IED stories (t(219)=3.83, p<.01). Happiness was elicited more when reading stories about the constitution than IED stories (t(216)=7.33, p<.01), as well as when reading stories about Iraqi forces compared to IEDs (t(219)=7.22, p<.01).
 Subsequent univariate tests for the interaction effect revealed significant differences for the following dependent variables: credibility F(4, 315) = 1.99, p < .10, partial eta2 = .03, anger F(4, 315) = 3.33, p < .05, partial eta2 = .04, pride F(4, 315) = 2.49, p < .05, partial eta2 = .03,  and happiness F(4, 315) = 3.51, p < .05, partial eta2 = .04.
            The findings for anger and happiness override the main effect findings reported for experimental condition and topic. Scheffe post hoc tests revealed some significant differences in the pattern of means across experimental condition and topic. When reading milblog stories, the constitution stories elicited more anger than IEDs stories (t(69)=3.47, p<.01). Constitution stories also elicited more anger than stories about Iraqi forces  (t(70)=3.74, p<.01). Civilian news stories about IEDs elicited more emotion than those about Iraqi forces (t(70)=3.05, p<.01). Military on-line news sources found more anger depicted in IED stories than those about Iraqi forces (t(70)=3.05, p<.01). For the emotion pride, milblogs found more pride elicited when comparing the IED stories to the constitution stories (t(69)=5.05, p<.01) and stories about Iraqi forces (t(73)=2.84, p<.01). Civilian on-line news stories elicited more pride when reading stories about Iraqi forces when compared to IED stories (t(72)=3.68, p<.01); and when reading stories about the constitution compared to IEDs (t(74)=2.44, p<.01). When examining happiness, milblogs elicited more happiness when reading stories about Iraqi forces when compared to stories on the constitution (t(70)=2.31, p<.05). Military on-line news elicited more happiness in stories about the constitution (t(69)=7.44, p<.01) and Iraqi forces (t(70)=7.56, p<.01) when compared to stories about IEDs. Civilian on-line news stories about the constitution (t(74)=6.67, p<.01) and Iraqi forces (t(72)=3.80, p<.01)both elicited more happiness then stories about IEDs.
Regression Results

Finally, regression analysis examined the relative impact of uses of milblogs, TV news, and newspapers for securing information about international affairs, national affairs, and weather/sports. The results indicated that use of milblogs for national news was a positive indicator for discussing blogs with others(b=.36, p<.05). The use of blogs for international news was a positive predictor for the amount of knowledge students had about the war in Iraq (b=.36, p<.05) and involvement about the issue and the war (b=.34, p<.01). There were no significant effects for TV news or newspaper use.