Results

To begin, we analyze personal attitudes and perceived attitudes with standard statistical approaches. There were significant associations between personal MAI and perceived MAI (schoolboys, r = 0.80, p < .001; footballers, r = 0.54, p < .001). Paired sample t-tests showed that the perceived MAI was significantly higher than personal MAI (schoolboys, t(71) = 5.70, p < .001; footballers, t(37) = 2.58, p < .05). Thus, the students and the footballers perceived that masculine attitudes were more

196 Exponential Random Graph Models for Social Networks

Table 14.2. Two models for aggression relations among the footballers

parameter

model

A

B

structural (higher-order) and personal attitudes

structural (higher-order), personal attitudes, and perceived attitudes

Purely structural effects (endogenous)

Arc

-5.49 (0.81)*

-6.45 (0.91)*

Reciprocity

0.25 (0.72)

0.17 (0.67)

Simple connectivity [Mix2star]

-0.09 (0.34)

-0.08 (0.36)

Popularity spread [AinS]

0.65 (0.25)*

0.62 (0.29)*

Activity spread [AoutS]

1.05 (0.25)*

1.02 (0.26)*

Path closure [AT-T]

-0.18 (0.29)

-0.21 (0.29)

Multiple 2-paths [A2P-T]

0.12 (0.39)

0.10 (0.41)

Actor-relation effects (exogenous)

Heterophily - personal attitudes

0.19 (0.19)

0.15 (0.21)

Heterophily - perceived attitudes

-0.10 (0.18)

Sender - personal attitudes

0.03 (0.14)

-0.17 (0.17)

Sender - perceived attitudes

0.36 (0.18)*

Receiver - personal attitudes

-0.02 (0.16)

-0.29 (0.21)

Receiver - perceived attitudes

0.44 (0.21)*

Sender - relationship

0.02 (0.31)

-0.01 (0.34)

Receiver - relationship^

-1.32 (0.52)*

-1.68 (0.62)*

Heterophily - experience

0.003 (0.002)

0.005 (0.002)*

Sender - experience

0.000 (0.002)

-0.002 (0.002)

Receiver - experience

0.004 (0.003)

0.004 (0.003)

Covariate network effects (exogenous) Covariate arc (“best player” network)

-0.21 (0.40)

-0.45 (0.40)

* Significant effect

a There is no homophily effect for “relationship” because there are no such ties in the network. Trying to fit a model with an observed value of zero will result in a degenerate model (see Section 13.2).

dominative than the actual average score of personal attitudes. The pattern of these findings suggests evidence for pluralistic ignorance. We now present exponential random graph model (ERGM) results for our two networks separately.

 
Source
< Prev   CONTENTS   Source   Next >