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

Discriminant Validity

Another critical issue was to investigate the dimensionality of the identified sub-constructs; that is, to evaluate whether the single sub-dimensions are distinguishable from one another. Campbell and Fiske (1959, p. 84) emphasize this issue by noting that “one cannot define without implying distinctions, and the verification of these distinctions is an important part of the validation process”. This is called discriminant validity. In the research at hand, the discriminant validity of the five sub-dimensions of eWOM trust was assessed in three different ways: first, by testing the confidence interval around the correlation between factors; second, more rigorous evidence for discriminant validity is obtained by comparing the shared variance (i.e., squared correlation) between each pair of factors against the AVEs of the two factors (Fornell & Larcker, 1981); third, by comparing the results of chi square difference tests and fit statistics of competing measurement models.

In order to argue for discriminant validity, one has to examine the relationship of factors to ensure that they are not perfectly correlated; that is, the correlations equal to one (Bagozzi & Yi, 1991; Smith et al., 1996). Hence, evidence for discriminant validity in the first case is given if the correlations are two or more standard errors below 1.0 or, in other words, the confidence interval (+/-2 standard errors) around the correlation of two sub-dimensions does not contain

223

the value of 1.0 (Anderson & Gerbing, 1988; Schmitt & Stults, 1996). In the second case, discriminant validity is given if the shared variance estimate is smaller than the AVE estimates of the involved constructs (Farrell, 2010; Hair et al., 2010). In the third case, discriminant validity is assumed if a more constrained model, that is, a model that assumes the existence of multiple factors, exhibits a significantly better model fit compared to a less constrained model (Anderson & Gerbing, 1988; Joreskog, 1971) - that is one that assumes that all indicators load on a single factor. Literature regularly provides examples for the application of these rules of thumb (e.g., Netemeyer et al., 1996).

Table 22 reports the AVE of the individual sub-dimensions, the (completely standardized) correlations and squared correlations of the factor pairings. The table also includes the upper bound of the 95% confidence interval of the correlations between every possible factor pairing for the two samples. For instance, the upper bound of the confidence interval was .82 (.68) for the correlation between the ability and the integrity/honesty sub-dimension, .56 (.46) for the benevolence and integrity/honesty correlation, and .52 (.47) for the willingness to depend and benevolence correlation. The highest upper bound of the confidence interval (.82) could be found between ability and integrity/honesty in the development sample. However, a comparable result was not achieved in sample 3b. Here, in contrast, the highest upper bound was identified for the correlation between willingness to rely and ability (.70). As none of the confidence intervals included 1.0, support for discriminant validity is given.

A slightly different picture is provided by assessing the discriminative power of the factors by using the second criterion. Here, the squared correlation between ability and integrity/honesty (.61) was slightly above the AVEs of the integrity/honesty (.56), as well as the ability (.59) construct. Therefore, discriminant validity might not be given. However, as this pattern cannot be detected in the second sample (here the squared correlation among the factors was much lower: .40 compared to the two involved AVE (.50 for integrity/honesty and .60 for ability)), this threat might be eased. All remaining factors showed higher AVE than the squared correlation of all possible factor pairings, supporting discriminant validity among the five subdimensions of eWOM trust.

Integrity/

Honesty

Ability Benevolence Willingyss

to rely

Willingness to depend

3a

3b

3a

3b

3a

3b

3a

3b

3a

3b

Integrity/

Honesty

.56a

.50 a

Ability

.78b

.82c

.61d

.63b

.68c

.40d

.59a

.60a

Benevolence

.49b

.56c

.24d

.39b

.46c

.15d

.53

.59c

.28d

.42b

.49c

.18d

.43a

.43a

Willingness to rely

.68b

.73c

.46d

.60b

.66c

.36d

.73b

.77c

.53d

.65b

.70c

.42d

.46b

.53c

.21d

.40b

.48c

.16d

.70a

.70a

Willingness to depend

.66b

.71c

.44d

.58b

.64c

.34d

.71b

.76c

.51d

.63b

.69c

.40d

.45b

.52c

.12d

.39b

.47c

.15d

.62b

.67c

.38d

.60b

.66c

.36d

.67a

.64a

Notes: a = Average variance extracted (AVE); b = Pearson correlation coefficient (r); c = Upper bound of 95% confidence interval of r; d = Squared correlation (r2); n = 425 (425).

To provide greater confidence in the proposed factor structure and the discriminant validity among the first-order variables of eWOM trust, alternative measurement models were estimated. These were compared using chi-square difference tests, as suggested by Anderson and Gerbing (1988) and a selection of fit indices. This approach follows common practices in scale research (e.g., Delgado-Ballester, 2004). The number of models as well as the factors that were combined were determined by the degree of (average) correlation among the different factors observable in the two samples. Accordingly, one model was proposed that understands eWOM trust as a second-order construct with three first-order factors (Model 3) and two others with four first-order factors (Models 4a-4b). For Model 3, the dimensions of ability and integrity/honesty as well as willingness to rely and willingness to depend were combined in two constructs because of their considerable interdependence; that is, a correlation of .78 (.63) between former and of .62 (.60) between the latter constructs. Benevolence was here regarded as a separate factor. Ability and willingness to rely (r = .73/.65; average correlation (ravg) = .69) were joined for a four-factor model (Model 4a) and an alternative four-factor model (Model 4b) originated by the combination of ability and willingness to depend (r = .71/.63; ravg = .67). Additionally, the hypothesized second-order factor model (Model 5), where responses to each item are reflective of the five proposed factors, and a one-factor model (Model 1), conzeptualizing eWOM trust as a uni-dimensional construct where the covariance among the items being accounted for by a single factor, were specified. Finally, a null model (Model N), for which the assumption that no systematic relationship among the scale items existed, was also included.

Table 23 and 24 present the results for the competing models. It can be shown that the second- order model, with its five first-order dimensions, fits the data best. This model had the lowest chi-square values across the two samples (x2 = 668.12 (589.12), df = 270)) and also the additional model fit indices indicate that it’s the most suitable model (RMSEA = .06 (.05), NFI = .90 (.90), NNFI = .93 (.94), CFI = .94 (.94), RMR = .10 (.11), SRMR = .06 (.07), GFI = .90 (.90), AGFI = .87 (.88)). Additionally, the improvement in chi-square was significant when comparing this model to all of its less complex counterparts. For instance, the chi-square difference between the one-factor model and the second-order model with its five first-order constructs was 1,235.41 (1,536.45) and highly significant (p < .001), and there was also a significant improvement from the simple three-factor to the five sub-dimensions model (x2Diff = 586.09 (784.15), dfDiff = 3, p < .001). Chi-square as well as the diverse model fit indices could also be enhanced when comparing the two four-factor models to the model mirroring the hypothesized structure of the construct. According to these results, it was concluded that the five-sub-dimension model is qualified to represent the data inherent in the two samples. Accordingly, merging any of the proposed dimensions would lead to an impairment of the measurement model. By presuming that these factors should be treated separately out of empirical considerations, further evidence for discriminant validity of eWOM trust’s subdimensions was provided.

For establishing discriminant validity, the second-order construct with its five-dimensional first-order constructs (Model 5) was additionally compared to a five-correlated-factors model (Model 5b). According to this, conceptualization covariance among the observable indictors is accounted for by a set of first-order constructs, each construct representing a distinct aspect of eWOM trust and each item being the effect of only a single factor. All of the Model-5-fit indices were close to those of this first-order model across the two separate samples (RMSEA = .05 (.05), NFI = .91 (.92), NNFI = .94 (.96), CFI = .95 (.96), RMR = .08 (.08), SRMR = .05 (.05), GFI = .90 (.91), AGFI = .88 (.89)). These results suggested that a second-order factor structure of type 1 was acceptable, as it validly captured the construct's true structure embedded within the sample data.

227

Model Chi Square df

P

Chi Square Difference

Model Fit Indices

Competing Chi Square df Models Difference Difference

Sign.

Chi

Square/ RMSEA NFI NNFI CFI RMR SRMR GFI AGFI

df

Null Model (N) 6,618.07 300

n.a. n.a. n.a.

n.a.

22.06 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

One Factor Model (1) 1,903.54 275

.001

6.92 .15 .71 .72 .74 .14 .08 .66 .60

(N-l) 4,714.53 25

***

Three Factor Model (3) 1,254.22 273 (A/I; B; WR/WD)

.001

4.59 .11 .81 .83 .85 .12 .07 .75 .71

(N-3) 5,363.86 27

***

(1-3) 649.32 2

***

Four Factor Model (4a) 977.05 271 (A/WR; I; B; WD)

.001

3.61 .09 .85 .88 .89 .11 .06 .82 .78

(N~4a) 5,641.02 29

***

(l~4a) 926.49 4

***

(3-4 a) 277.17 2

***

Four Factor Model 986.37 271 (4b) (A/WD; I; B;

.001

3.64 .08 .85 .88 .89 .12 .07 .83 .80

WR)

(N-4b) 5631.70 29

***

(l-4b) 917.16 4

***

(3-4b) 267.84 2

***

Five Factor Model (5) 668.12 270

.001

2.47 .06 .90 .93 .94 .10 .06 .90 .87

(N-5) 5,949.95 30

***

(1-5) 1,235.41 10

***

(3-5) 586.09 3

***

(4a-5) 308.92 1

***

(4b-5) 318.25 1

***

Table 23: Competing Models (Sample 3a)

Notes: A = Ability; I = Integrity/Honesty; В = Benevolence; WR = Willingness to rely; WD = Willingness to depend; *** = p <.001.

228

Model

Chi Square Difference

Model Fit Indices

Chi Square

4f

P

Competing

Models

Chi Square Difference

if

Difference

Sign.

Chi

Square /

if

RMSEA

NFI

NNFI

CFI

RMR

SRMR

GFI

A GFI

Null Model (N)

5,998.5 6

300

n.a.

n.a.

n.a.

n.a.

20.00

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

2,126.16

275

.001

7.73

.17

.65

.65

.68

.16

.16

.61

.53

(N-l)

3,872.40

25

***

Three Factor Model

1,373.86

273

.001

5.03

.13

.77

.79

.81

.13

.08

.70

.64

  • (3)
  • (АЛ; B; WRJWD)

(N-3)

4,624.70

27

***

(1-3)

752.30

2

***

Four Factor Model (4a) (A/WR; I; B; WD)

990.81

271

.001

(N-4a)

5,007.75

29

3.66

.10

.84

.86

.87

.11

.07

.80

.76

(l-4a)

1,135.35

4

***

(3-4a)

383.05

2

***

Four Factor Model

952.93

271

.001

3.52

.08

.84

.87

.14

.14

.08

.83

.80

(4b) (A/WD; I; B; WR)

(N-4b)

5,045.63

29

***

(l-4b)

1,173.23

4

***

(3-4b)

420.93

2

***

Five Factor Model (5)

589.71

270

.001

(N-5,

5,408.84

30

2.18

.05

.90

.94

.94

.11

.07

.90

.88

(1-5,

1,536.45

10

***

(3-5,

784.15

3

***

(4a-5,

401.10

1

***

(4b-5,

363.22

1

Table 24: Competing Models (Sample 3b)

Notes: A = Ability; I = Integrity/Honesty; В = Benevolence; WR = Willingness to rely; WD = Willingness to depend; *** = p <.001.

In order to confidently argue for the construct’s dimensionality, all factors were subjected to pairwise testing as proposed by Anderson and Gerbing (1988). Here, models for every possible pairing of the diverse factors were evaluated for their representativeness for the underlying data. For each pairing, three different models were assessed: (a) a null model which supposed that among the manifest indicators and the factors, no systematic relationship does exist, (b) a one- factor model in which all observable items were specified to a single factor; and (c) a two-factor model (i.e. two correlated first-order factors) representing the hypothesized separate factor structure, in which the manifest indicators are permitted to load only on their hypothesized factor. Evidence for discriminant validity among the factors exists if the chi-square fit of the two-factor model is better than the fit of the one-factor model.

Table 25 and 26 present the fit statistics for the competing models. Throughout the various factor pairings and samples, fit statistics suggest adequate fit for the two-factor model. To illustrate, the favourable chi-square measure and diverse fit indices can both be interpreted as evidence for the discriminability between the dimensions of integrity/honesty and ability. CFA resulted in a chi-square fit for the one-factor solution of 451.72 (402.65) (df = 77), while the chi-square for the two-factor model was 167.35 (162.47) (df = 76). The chi-square difference (X2Diff = 284.37 (342.71); df Diff = 1) was significant (p < .001), indicating a better model fit of the two-factor model. Simultaneously, other fit indices of the two-factor model were considerably enhanced (RMSEA = .06 (.05), NFI = .95 (.95), NNFI = .97 (.97), CFI = .97 (.97), RMR = .05 (.06), SRMR = .04 (.04), GFI = .94 (.95), AGFI = .92 (.92)), supporting the validity of the hypothesized model. As in the case of integrity/honesty and ability, all other pairings showed a significant improvement in chi-square and better fit indices of the two-factor over the single-factor conceptualizations. Hence, strong empirical arguments for the dimensionality of the eWOM trust construct and the discriminability of its components lead to the acceptance of the hypothesized construct structure.

230

Sub

dimension

Pairing

Model

Chi

Square

if

P

Chi Square Difference

Model Fit Indices

„ Chi Competing e

Models

Difference

if

Difference

Sign.

Chi

Square/

if

RMSEA

NFI

NNFI

CFI

RMR

SRMR

GFI

A GFI

Integrity / Ability

Null Model (N)

3,452.40

91

37.94

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

451.72

77

.001

(N-l)

3,000.68

14

***

5.87

.13

.87

.87

.89

.08

.06

.82

.75

Two Factor Model (2)

167.35

76

.001

(N-2)

3,285.05

15

***

2.20

.06

.95

.97

.97

.05

.04

.94

.92

(1-2)

284.37

1

***

Integrity / Benevolence

Null Model (N)

2,317.07

66

35.11

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

320.92

54

.001

(N-l)

1,996.15

12

***

5.94

.12

.86

.86

.88

.14

.07

.87

.82

Two Factor Model (2)

149.31

53

.001

(N-2)

2,167.76

13

***

2.82

.07

.94

.95

.96

.08

.05

.94

.91

(1-2)

171.61

1

***

Integrity / Willingness to rely

Null Model (N)

2,595.94

55

47.20

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

555.30

44

.001

(N-l)

2,040.64

11

***

12.62

.18

.79

.75

.80

.16

.10

.78

.68

Two Factor Model (2)

89.47

43

.001

(N-2)

2,506.47

12

***

2.08

.05

.97

.98

.98

.05

.03

.96

.94

(1-2)

465.83

1

***

Integrity / Willingness to depend

Null Model (N)

2,633.58

55

47.88

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

457.63

44

.001

(N-l)

2,175.95

11

***

10.40

.16

.83

.80

.84

.07

.07

.82

.74

Two Factor Model (2)

128.59

43

.001

(N-2)

2,504.99

12

***

2.99

.07

.95

.96

.97

.07

.04

.95

.92

(1-2)

329.05

1

***

Ability / Benevolence

Null Model (N)

1,832.20

45

40.72

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

320.82

35

.001

(N-l)

1,511.38

10

***

9.17

.15

.83

.79

.84

.19

.10

.85

.76

Two Factor Model (2)

93.38

34

.001

(N-2)

1,738.82

11

***

2.75

.07

.95

.96

.97

.09

.06

.96

.93

(1-2)

227.45

1

***

Ability / Willingness to rely

Null Model (N)

2703,56

45

60.08

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

399,33

35

.001

(N-l)

2304,23

10

***

11.41

.19

.85

.82

.86

.10

.07

.78

.66

Two Factor Model (2)

85,32

34

.001

(N-2)

2618,24

11

***

2.51

.06

.97

.97

.98

.05

.03

.96

.94

_(N2)

314,01

1

***

Table 25: Pairings of Sub-dimensions (Sample 3a)

Chi Square Difference

Model Fit Indices

Sub-dimension

Pairing

Model

Chi

Square

4f

P

Competing

Models

Chi

Square

Difference

4f

Difference

Sign.

Chi

Square / df

RMSEA

NFI

NNFI

CFI

RMR

SRMR

GFI

A GFI

Ability / Willingness to depend

Null Model (N)

2,182.27

36

60.62

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

402.74

27

.001

(N-l)

1779,53

9

***

14.92

.18

.82

.77

.83

.18

.08

.83

.71

Two Factor Model (2)

60.03

26

.001

(N-2)

2122,24

10

***

2.31

.05

.97

.98

.98

.05

.03

.97

.95

(1-2)

342,71

1

***

Benevolence / Willingness to rely

Null Model (N)

1,488.92

28

53.18

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

341.79

20

.001

(N-l)

1147,13

8

***

17.09

.22

.77

.69

.78

.28

.15

.80

.63

Two Factor Model (2)

43.93

19

.001

_(N-2)

1444,99

9

***

2.31

.06

.97

.98

.98

.10

.05

.97

.95

(1-2)

297,86

1

***

Benevolence / Willingness to depend

Null Model (N)

1,104.34

21

52.59

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

295.73

14

.001

(N-l)

808,61

7

***

21.12

.25

.73

.61

.74

.28

.15

.80

.60

Two Factor Model (2)

42.21

13

.001

_(N-2)

1062,13

8

***

3.25

.07

.96

.96

.97

.12

.06

.97

.94

_0-2)

253,52

1

...

Willingness to rely /

Willingness to depend

Null Model (N)

1,966.32

21

93.63

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

371.99

14

.001

(N-l)

1594,33

7

***

26.57

.26

.81

.72

.82

.20

.09

.78

.57

Two Factor Model (2)

75.80

13

.001

_(N-2)

1890,52

8

...

5.83

.10

.96

.95

.97

.10

.05

.96

.91

_d-2)

296,19

1

***

Table 25 (Cont.): Pairings of Sub-dimensions (Sample 3a)

Notes: *** = p <.001; n.a. = not available.

231

232

Sub-

Chi Square Difference

Model Fit Indices

Model

Chi

Square

if

P

Competing

Models

Chi

Square

Difference

if

Difference

Sign.

Chi

Square

/if

RMSEA

NFI

NNFI

CFI

RMR

SRMR

GFI

A GFI

dimension

Pairing

Integrity / Ability

Null Model (N)

3,122.84

91

34.32

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

685.94

77

.001

(N-l)

2,436.90

14

***

8.91

.19

.78

.76

.80

.12

.09

.71

.60

Two Factor Model (2)

162.47

76

.001

(N-l)

2,960.37

15

***

2.14

.05

.95

.97

.97

.06

.04

.95

.92

(1-2)

523.47

1

***

Integrity / Benevolence

Null Model (N)

2,009.13

66

30.44

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

320.18

54

.001

(N-l)

1,688.95

12

***

5.93

.12

.84

.83

.86

.13

.08

.87

.81

Two Factor Model (2)

123.52

53

.001

(N-l)

1,885.61

13

***

2.33

.06

.94

.95

.96

.06

.04

.95

.93

(1-2)

196.66

1

***

Integrity / Willingness to rely

Null Model (N)

2,245.10

55

40.82

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

602.75

44

.001

(N-l)

1.642.36

11

***

13.70

.18

.73

.68

.75

.18

.11

.78

.67

Two Factor Model (2)

82.80

43

.001

(N-l)

2,162.31

12

***

1.93

.05

.96

.98

.98

.04

.03

.96

.94

(1-2)

519.95

1

***

Integrity / Willingness to depend

Null Model (N)

2,187.78

55

39.78

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

413.29

44

.001

(N-l)

1,774.49

11

***

9.39

.14

.81

.78

.83

.18

.08

.85

.77

Two Factor Model (2)

99.80

43

.001

(N-l)

2,087.98

12

***

2.32

.06

.95

.97

.97

.07

.04

.96

.94

_d-2)_

313.49

1

***

Ability / Benevolence

Null Model (N)

1,864.92

45

41.44

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

381.59

35

.001

(N-l)

1,483.33

10

***

10.90

.18

.80

.76

.81

.23

.13

.81

.71

Two Factor Model (2)

87.94

34

.001

_(NUJ_

1,776.98

11

***

2.59

.06

.95

.96

.97

.09

.06

.96

.94

(1-2)

293,65

1

***

Ability / Willingness to rely

Null Model (N)

2,681.00

45

59.58

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

468.05

35

.001

(N-l)

2,212.95

10

***

13.37

.22

.83

.79

.84

.12

.08

.74

.59

Two Factor Model (2)

38.21

34

.284

(N-l)

2,642.79

11

***

1.12

.02

.99

1.00

1.00

.03

.02

.98

.97

_Ik?]_

429.84

1

***

Table 26: Pairings of Sub-dimensions (Sample 3b)

Chi Square Difference

Model Fit Indices

Sub-dimension

Pairing

Model

Chi

Square

if

P

Competing

Models

Chi

Square

Difference

if

Difference

Sign.

Chi

Square

/if

RMSEA

NFI

NNFI

CFI

RMR

SRMR

GFI

A GFI

Ability / Willingness to depend

Null Model (N)

2,100.59

36

58.35

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

402.65

27

.001

(N-l)

1,697.94

9

***

14.91

.18

.81

.76

.82

.24

.10

.83

.71

Two Factor Model (2)

60.13

26

.001

(N-l)

2,040.46

10

***

2.31

.05

.97

.98

.98

.09

.05

.97

.95

(1-2)

342.52

1

***

Benevolence / Willingness to rely

Null Model (N)

1,481.34

28

52.90

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

365.79

20

.001

(N-l)

1,115.54

8

***

18.29

.23

.75

.67

.76

.29

.16

.79

.62

Two Factor Model (2)

49.27

19

.001

(N-l)

1,432.06

9

***

2.59

.06

.97

.97

.98

.08

.05

.97

.95

(1-2)

316.52

1

***

Benevolence / Willingness to depend

Null Model (N)

981.21

21

46.72

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

290.44

14

.001

(N-l)

690.77

7

***

20.75

.25

.70

.57

.71

.27

.15

.80

.59

Two Factor Model (2)

15.99

13

.001

(N-l)

965.22

8

***

1.23

.02

.98

1.00

1.00

.07

.03

.99

.98

(1-2,

274.45

1

***

Willingness to rely /

Willingness to depend

Null Model (N)

1,813.34

21

86.35

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

One Factor Model (1)

309.39

14

.001

(N-l)

1,503.95

7

***

22.10

.23

.83

.75

.84

.22

.09

.83

.65

Two Factor Model (2)

44.66

13

.001

(N-l)

1,768.68

8

***

3.44

.07

.98

.97

.98

.09

.04

.97

.94

_(N2J_

264.73

1

***

Table 26 (Cont.): Pairings of Sub-dimensions (Sample 3b)

Notes: *** = p <.001; n.a. = not available.

to

u>

u>

 
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