Menu
Home
Log in / Register
 
Home arrow Engineering arrow Measuring Electronic Word-of-Mouth Effectiveness: Developing and Applying the eWOM Trust Scale
Source

Reliability

Table 21 summarizes the psychometric properties of the scale for the development as well as holdout sample. In this research, reliability of the items was evaluated by assessing the path coefficients and examining the squared multiple correlation (R2) (Bollen, 1989). All the items turned out to load highly on their respective sub-dimension and t-values associated with the loadings exceeded critical values (2.75) for significance on the .01 level. This verifies the hypothesized relationships among the observable indicators and constructs. Similar results were achieved in the holdout sample. In respect to the squared multiple correlation, values better than .50 (Fornell & Larcker, 1981) are typically preferred. This implies that the majority of the variance in the indicator is attributable to the latent (sub-)construct. The R2 for most items were well above this threshold. However, in the developmental as well as the holdout sample, seven items were not able to achieve the required reliability level and had slightly lower values.

Cronbach’s alpha, Fomell and Larcker’s (1981) index of construct (or composite) reliability, the average inter-item-correlation (discussed earlier), as well as the average variance extracted (AVE), which provides an index for the amount of variance that is captured by the latent construct in relation to the amount of variance due to measurement error, were used to assess the reliability of each latent sub-dimension. In both samples, Cronbach’s alphas range from .75 to .90, while construct reliabilities varied from .75 to .90/.91, which implies that the minimum threshold of .70 was exceeded in all cases. In respect to the AVE, four of the five sub-constructs also met the recommended level of 50% (Hair et al., 2010). Only the benevolence factor performance was inferior to the other components of eWOM trust (mean AVE = .43).

A further investigation demonstrated that all five first-order constructs had positive and significant loadings on the second-order construct. The squared multiple correlations also achieved desirable levels above the recommended threshold in all but one relationship (benevolence) (R2 = .33/.26). On the second-order level, the reliability of the first-order construct was also assessed by the index of construct (or composite) reliability and the AVE for both samples. The internal consistency measures exceeded the accepted standard of .70 (Nunnally & Berstein, 1994) (.89/.85) and .50 (Fornell & Larcker, 1981) respectively (.63/.54). In general, the model provides reliable measurement of the eWOM trust construct.

222

Sub

dimension

У

Item

X

Squared

Multiple

Correlation

Corrected

Item-to-Total

Correlation

Average

Inter-Item

Correlation

Cronbach ’s Alpha

Construct

Reliability

AVE

Construct

Reliability

AVE

Ability

.92 (.83)

Ab6

Ab7

Ab8

Ab9

AblO

Abll

.68 (.67) .78 (.85) .71 (.72) .83 (.83) .82 (.74) .78 (.82)

.46 (.45) .60 (.72) .50 (.51) .68 (.69) .60 (.55) .61 (.67)

.63 (.62) .74 (.81) .66 (.68) .79 (.78) .75 (.69) .71 (.77)

.59 (.59)

.89 (.90)

.89 (.90)

.59

(.60)

.89 (.85)

.63

(.54)

Integrity/

Honesty

.85 (.76)

In2

In3

In4

In5

In6

In7

In9

InlO

.76 (.71) .70 (.65) .77 (.79) .82 (.77) .76 (.72) .65 (.63) .69 (.67) .75 (.70)

.58 (.50) .49 (.43) .60 (.62) .67 (.59) .59 (.51) .42 (.40) .47 (.44) .56 (.50)

.73 (.67) .67 (.62) .74 (.74) .77 (.72) .70 (.66) .61 (.59) .63 (.61) .70 (.65)

.54 (.50)

.90 (.89)

.91 (.89)

.56

(.50)

Benevolence

.57 (.51)

Bel

Be2

Be3

Веб

.72 (.65) .67 (.63) .67 (.71) .55 (.64)

.52 (.43) .45 (.40) .49 (.51) .30 (.41)

.60 (.55) .53 (.51) .68 (.60) .45 (.52)

.42 (.43)

.75 (.75)

.75 (.75)

.43

(.43)

Willingness to rely

.79 (.79)

Wil

Wi4

Wi5

Wi8

.82 (.83) .86 (.82) .79 (.84) .87 (.86)

.67 (.68) .74 (.66) .63 (.71) .76 (.74)

.75 (.77) .81 (.77) .75 (.78) .81 (.80)

.70 (.70)

.90 (.90)

.90 (.90)

.70

(.70)

Willingness to depend

.78 (.76)

Wi2

Wi6

Wi7

.63 (.63) .92 (.89) .88 (.86)

.40 (.39) .86 (.80) .77 (.74)

.58 (.56) .81 (.77) .74 (.72)

.64 (.61)

.84 (.82)

.86 (.84)

.67

(.64)

Notes: Results for the holdout sample 3b in parentheses; у = Completely standardized second-order loading; X = Completely standardized first-order loading; AVE = Average variance extracted.

Table 21: Psychometric Properties of the eWOM Trust Scale (Samples 3a and 3b)

 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel