Running Head: Invisible Leverage in Adoption of Online Social Support Community

19

the major contributor, and network activeness (See Table 1)

10

. The number of registered

members accounted for 5.5% of the total variance (p< .01) in terms of the frequency of postings

per adopter, for 8.6% (p< .01) in terms of the frequency of postings per adopter excluding the

major contributor, for 4.6% (p< .01) in terms of the average duration excluding the major

contributor, and for 41.6% (p< .01) in terms of network activeness (all linear).

Hypothesis 2-2 was partially supported. The number of posters showed positive linear as

well as curvilinear relationships with the frequency of postings per adopter excluding the major

contributor and network activeness. Total variances in the two measures were better explained

by curvilinear regression than by linear regression. Positive linear relationships were found with

the frequency of postings per adopter, the average duration of adopter excluding the major

contributor, and the proportion of adopters among posters (See Table 2).

The number of posters accounted for 22.3% of the total variance (p<.01) in terms of the

frequency of postings per adopter excluding the major contributor, for 58.4% (p<.05) in terms of

network activeness (all curvilinear), for 7.4% (p<.01) in terms of the frequency of postings per

adopter, for 2.6% (p<.01) in terms of the average duration of adopter excluding the major

contributor, and for 3.8% (p<.01) in terms of the proportion of adopters among posters (all

linear).

Hypothesis 2-3 was also partially supported. The number of adopters showed positive

linear as well as curvilinear relationships with the frequency of postings per adopter excluding

the major contributor and the proportion of adopters among posters. Total variances in the two

measures were better explained by curvilinear regression than by linear regression. Positive

linear relationships were found with the frequency of postings per adopter, the average duration

of adopter excluding the major contributor, and network activeness (See Table 3).

The number of adopters accounted for 27.5 % of the total variance (p<.01) in terms of the

frequency of postings per adopter excluding the major contributor, for 11.1% (p<.05) in terms of

the proportion of adopters among posters (all curvilinear), for 11.1% (p<.01) in terms of the

frequency of postings per adopter, for 3.8% (p<.01) in terms of the average duration of adopter

excluding the major contributor, and for 66.2% (p<.05) in terms of network activeness (all

linear).

10

A cubic relationship found with network activeness was considered inappropriate because the tolerance level is

close to ‘0’. In all later analyses, cubic relationships with low tolerances were considered inappropriate.