Opinion Climates & Expression in the Electronic Forum
18
in terms of whether they held the national majority or minority position-- on both
measures of opinion consistency were found for school vouchers and the Electoral
College, even after group conditions and group size were also controlled. Main effects of
group conditions tended to be significant across issues. In comparison with the baseline
unanimous majority groups, participants in the ‘typical’ majority and minority groups
tended to express a smaller number of consistent statements. But one coefficient for the
Electoral College was significant. With regard to the proportion of consistent statements,
however, coefficients of typical condition were significant for the three issues (death
penalty: b=-.13, p<.01, school vouchers: b=-.18, p<.1, Electoral College: b=-.35, p<.001).
Participants in the ‘inflated’ groups where those holding the national
majority position were minorities within the group also tended to generate a smaller
number of consistent statements than those who in the unanimous condition, and to show
lesser degree of consistency in terms of the proportion (school vouchers: b=-.27, p<.05,
Electoral College: b=-.29, p<.05). There was a marginally significant interaction between
national issue position and the group opinion climate with regard to the proportion of
consistent statements (civil unions: b=-.28, p<.1). Group size generally was not a
significant predictor of consistency.
[Table 3 about here]
Figure 1 displays the estimated means of the proportion of consistent statements
according to issue position and the three group opinion climates on the basis of Equation Is.
As expected, in general, the estimated consistency scores of those holding the national
majority position, tended to decrease, as the group opinion climate became less favorable,
whereas national minorities consistency tended to remain relatively constant.
[Figure 1 about here]
condition’ when ‘minority*typical condition’ was already employed as an interaction term. Thus, the
estimated mean of each group in Figure 1 and 2 was calculated by following procedures; national majority in
the unanimous group
0
bˆ
Y
=
, majority in typical
11
0
bˆ
bˆ
Y
+
=
, majority in inflated
X
bˆ
bˆ
Y
12
0
+
=
minority
in inflated
12
10
0
bˆ
bˆ
bˆ
Y
+
+
=
minority in typical
}
bˆ
b
bˆ
bˆ
Y
13
11
10
0
+
+
+
=
Convention