Table of Contents Table of Contents
Previous Page  55 / 372 Next Page
Information
Show Menu
Previous Page 55 / 372 Next Page
Page Background

臺大管理論叢

26

卷第

2

55

Construct

Measurement item

Factor

loading

Cronbach’s

α

Reference

Knowledge

sharing

behavior

I frequently participate in knowledge sharing

activities and share my knowledge with my

friends on Facebook.

.76

.79

Lin et al.

(2009)

I usually spend a lot of time conducting

knowledge sharing activities on Facebook.

.84

When discussing a complicated issue, I am

usually involved in the subsequent interactions.

.63

Table 2 Results of CFA

Construct

Number

of items

Composite

reliability

AVE

Correlation between latent constructs

1.

2.

3.

4.

5.

6.

1. Structural social capital

2

.80

.80 .90

2. Cognitive social capital

2

.79

.79 .76 .89

3. Relational social capital

3

.86

.69 .60 .76 .83

4. Psychological safety

3

.85

.65 .44 .50 .46 .81

5. Knowledge sharing

self-efficacy

3

.88

.71 .45 .58 .49 .38 .84

6. Knowledge sharing

behavior

3

.79

.56 .46 .66 .48 .41 .68 .75

4.3 Model Fit and Hypothesis Testing

Regarding the hypothesized structural equation model, the resulting chi-square value

divided by degrees of freedom does not exceed 3 (

χ

2

(90) = 233.56), and all the fit indices

(GFI = .94, AGFI = .91, RMR = .05, RMSEA = .06, NFI = .97, NNFI = .98, CFI = .98, PNFI

= .73) were within the acceptable range, indicating a relatively good model fit to the data.

Table 3 provides the estimated parameters and results of the hypotheses testing. The

standardized path coefficients indicate that H1, H3, H4, H5, H8, and H11 are supported. For

H1, psychological safety has a significant, positive effect (

β

= .11,

p

< .01) on knowledge

sharing self-efficacy. As Table 3 shows, of the three facets of social capital, cognitive social

capital has the greatest impact. Cognitive social capital has a significant influence on all

three endogenous latent variables: psychological safety (H3;

γ

= .25,

p

< .05), knowledge

sharing self-efficacy (H5;

γ

= .46,

p

< .001) and knowledge sharing behavior (H8;

γ

= .54,

p

< .001). In contrast, relational social capital has only limited influences on the endogenous

latent variables because it only significantly affects psychological safety (H4;

γ

= .19,

p

<