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

Scale Finalization and Development of Norms

A concluding consideration in the scale development process targets scale length optimization (DeVellis, 2012; Netemeyer et al., 2003). The final eWOM trust scale comprised 22 (English version: 21) items, each one mirroring one of the five sub-dimensions: ability: 5 items; integrity/honesty: 7 items; benevolence: 3 items; willingness to rely: 4 items; and willingness to depend: 3 items. In this thesis’ introduction, it was mentioned that the new scale should also be a source for the further development of sub-scales, such as for the measurement of eWOM trustworthiness. Hence, it was advisable to approach scale finalization by considering scale length of the individual factors. Three of five factors were measured by four or more items, allowing (individual) one-factor CFA models to be over-identified. It followed that all these constructs were measured in accordance to general scaling recommendations put forward, for example, by Netemeyer et al. (2003). Benevolence as well as willingness to depend dimensions included three items each, implying that one-factor CFA models would be perfectly identified. Three items represent the lower end of the number of indicators for valid and reliable construct measurement. All constructs achieved considerable internal consistency. The three-item constructs achieved a Cronbach alpha greater than .80, while the other constructs were in the .90s range. This suggests that the later constructs had (theoretically) some “reliability to sparse” for a possible trade-off between scale length and internal consistency (DeVellis, 2012; Worthington & Whittaker, 2006). However, this research refrained from doing so due to several reasons. First, all items performed considerably well across multiple samples, giving no reason to exclude them without support from additional sample data. Second, the total eWOM trust scale was not considered as too lengthy but placing a burden on the respondents which is still adequate. Hence, the new scale possesses practicability. Respondents’ convenience was not regarded to be a valued compensator of scale reliability. Third, Bagozzi (according to Netemeyer et al., 2003) suggests that, for scales that target the measurement of personality traits or similar constructs, each dimension should be ideally measured by 8 to 10 items. As eWOM trust represents such a construct, a higher number of items was preferred to less. In sum, total and factors’ scale length were considered adequate.

Finally, this research recognized the development of norms as a valuable final step of the scale development process. Scholars such as Spector (1992) and more recently MacKenzie et al. (2011) advance the notion that researchers should provide a frame of reference which aids in the interpretation of the observed scores on a scale. In social sciences, the distribution of scores represents such a frame of reference to which the score from an individual is regularly compared. Investigating a population’s distribution of scores means administering the scale to a representative, as well as large enough, sample of respondents of the population of interest. The kind of data collection used in this research to obtain responses from sample 4 (Austrian/German consumers) and 6 (US consumers) suggested that these two samples are, to a considerable degree, representative in respect to age and gender of the two target populations

(i.e., online consumers/Intemet users). Additionally, as each sample includes more than 500 respondents, both samples are assumed to be of sufficient size to provide scores that are quite stable. Table 50 provides the means and standard deviations of the scores for these two samples. Additionally, the table also includes information concerning the shape of the distribution (i.e., skewness and kurtosis).

Table 50: Construct and Item Descriptives (Sample 4 and 6)

Construct

Item

Mean

SD

Skewness

Kurtosis

eWOM Trust

3.73 (4.30)

.89 (.95)

-.41 (-.70)

.80 (1.11)

Ability

Ab7

Ab8

Ab9

Ab10

Ab11

4.31 (4.59) 4.41 (4.81) 4.19 (4.06) 4.46 (4.79) 4.16 (4.52) 4.33 (4.79)

  • 1.01 (.97)
  • 1.18 (1.14)
  • 1.18 (1.28) 1.17 (1.13) 1.16 (1.13) 1.11 (1.12)

-.69 (-.93) -.66 (-1.21) -.58 (-.64) -.76 (-1.06) -.53 (-.74) -.56 (-.93)

.16 (1.59) .40 (2.26) .36 (.65) .74 (1.53) .46 (1.13) .46 (.83)

Integrity/

Honesty

In2

In3

In4

In5

In6

In9

In10

  • 3.39 (4.12) 3.50 (4.18) 3.37 (4.28) 3.33 (4.04) 3.54 (4.13)
  • 3.36 (4.14) 3.24 (4.14)
  • 3.36 (3.96)
  • 1.01 (1.08) 1.24 (1.26) 1.19 (1.16)
  • 1.18 (1.25)
  • 1.14 (1.30) 1.23 (1.29)
  • 1.18 (1.25)
  • 1.14 (1.26)

-.33 (-.52) -.21 (-.57) -.28 (-.39) -.35 (-.45) -.30 (-58) -.37 (-.57) -.13 (-.65) -.20 (-.33)

.34 (.53) .16 (.37) -.07 (-.02) .38 (.40) .13 (.43) .24 (.34) .20 (.80) .47 (-.03)

Benevolence

Be1

Be21

Be3

2.89 (3.57) 2.68 (3.63) 3.19 (3.74) 2.81 (3.50)

1.16 (1.27) 1.44 (1.41) 1.30 (1.54) 1.34 (1.34)

-.08 (-.06) -.03 (-.22) -.25 (-.44) -.76 (.09)

.16 (-.13) -.35 (-.27) .25 (-.12) .02 (-.16)

Willingness to rely

Wi1

Wi4

Wi5

Wi8

  • 4.58 (4.89)
  • 4.58 (4.87)
  • 4.58 (4.96)
  • 4.58 (4.78) 4.60 (4.95)
  • 1.10 (1.08) 1.25 (1.20) 1.20 (1.21)
  • 1.22 (1.30)
  • 1.23 (1.16)

-.71 (-1.31) -.79 (-1.28) -.66 (-1.50) -.71 (-1.38) -.82 (-1.34)

.44 (2.24) .27 (1.97) .33 (2.83) .21 (2.26) .52 (2.33)

Willingness to depend

Wi2

Wi6

Wi7

  • 3.45 (4.33) 3.27 (4.19)
  • 3.50 (4.47)
  • 3.50 (4.34)

1.28 (1.17) 1.56 (1.47) 1.49 (1.24) 1.41 (1.30)

-.38 (-.75) -.45 (-.77) -.43 (-.99) -.43 (-.81)

-.10 (.83) -.36 (.32) -.33 (1.44) -.12 (.57)

Notes: Results for US respondents (sample 6) in parentheses; 1 = item not included in final US scale.

 
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