Validation of CWB-T

As for the initial CWB-T items are generated by means of qualitative

72　Higher Education Evaluation and Development 9:1 (June 2015)

expert opinions, therefore, in order to have a psychometric validated instrument, the CWB-T is then subjected to a quantitative Confirmatory Factor Analysis (CFA). Descriptive statistics and correlation estimated was computed using the SPSS version 21, while the Composite Reliability (CR) and Average Variance Extracted (AVE) were used to prove the reliability and validity of measurement model. Structure model was used to explain the relationship and effect among latent variables. Structure Equation Modeling (SEM) was estimated using the maximum-likelihood method in the AMOS 20 program (Arbuckle, 2011).

Anderson and Gerbing (1988) proposed the use of a CFA to examine whether the measurement model provides an acceptable fit to the data. Since the conceptual design (as shown in Figure 1) of the CWB-T is of a second order CFA; meaning that within the CWB-T there are various different dimensionality or factors, while within these factors there are items (description of attitudes) that describe them. To remedy this, the two step method was used (DLittle, Cunningham, Shahar, & Widaman, 2002). First, based on the focus group session results, the CR and AVE for each of the factors are computed (see Table 2 for the CR and AVE values, all of which are within the accepted values). This then followed by examining the first and second order fit of the 46 items.

To determine the goodness of fit of the CWB-T, five indices were used (Byrne, 2001; Tucker & Lewis, 1973). SEM method with the use of AMOS 20 program was used to compute for various fit indices. Results show that the test for the second order CFA resulted in a relatively good fit to the data with χ² = 115.03, df = 20, GFI = .94 (Goodness of Fit Index; values > 0.90 which indicate good fit), CFI = .96 (Comparative Fit Index; values > 0.90 which indicate good fit), TLI = .95 (Tucker-Lewis Index; values > 0.90 which indicate good fit), NFI = .95 (Non-normed Fit Index; values > 0.90 which indicate good fit), and RMSEA = .071 (Root-Mean-Square Error of Approximation; values < 0.08 which indicate good fit); while all of the standardized loadings of the measured variables on the latent variables were greater than .63 and statistically significant at p < .001. All in all denoting a relatively good fit (Arbuckle, 2011; Byrne, 2001; Tucker & Lewis, 1973).

For the factor analysis of the CWB-T, Table 2 shows the various items together with their corresponding factor loadings and CR with values ranging from .73 to .90 denoting quite reliable results. In addition, factor loadings are above .6, while the AVE ranges from 46.69% to 63.25% denoting appropriate factorability (Anderson & Gerbing, 1988; Fornell & Larcker, 1981).

Hu, Hung and Ching: Examining the Counterproductive Work Behaviors within Taiwan Academic Setting: A Pilot Study 73 Table 2. Factor Loadings of CWB-T (N = 217)

Item TT IUR ISR IPR

Lying about being sick .74

Leaving without asking for leave .70

Coming to school late and/or going home early .60 Asking for leave regardless of the work situation .63

Doing personal stuff while on duty .70

Being online (personal internet surfing; FB) while on duty .73

Chatting while on duty .67

Waste of school’s resources .65

Occupying school’s resources as if one’s own property .63

Stealing school resources .84

Destruction of school’s resources .85

Favoritism or discriminating specific students .71

Improper student punishment .69

Mocking students .75

Discrimination against students .64

Deliberate singling out of specific students .76

Focusing only on students with good grades and ignoring others .81

Separated and cold towards students’ problems .74

Deliberate concealment or providing misleading information .77

Improper behavior in front of parents .81

Encouraging parents to go against the school .80

Conniving with parents .78

Ignoring or unwilling to communicate with parents .73

AVE 46.69 55.96 53.68 60.34

CR .86 .83 .89 .88

Item LOP AP PT RAD

Inadequate teacher preparation .78

Not following proper curriculum .79

Saying improper things during class .79

Too few or too much assignments/class activities .76

Casual checking of students’ assignments .79

Improper use of teaching pedagogy (such as too much movie time) .79

Unwilling to undergo tutoring .64

Lacks teaching enthusiasm .76

Wrong use of educational resources .79

Lacks professional content knowledge .69

Unwilling to participate in professional development workshops .80

74　Higher Education Evaluation and Development 9:1 (June 2015)

Lacks the motivation to join professional development programs .84

Gossiping .72

Spreading wrong/bad information .83

Improver verbal conduct .77

Deliberate neglect or ignoring others .80

Deliberate singling out others .85

Forming small groups/alliances to go against others .81

Convincing others to go against the school .77

Unwilling to cooperate with school administration .86

Going against all educational reforms .78

Unwilling to undertake administrative responsibilities .75

Miscommunication between teachers and administrators .78

AVE 61.16 57.11 63.02 63.25

CR .90 .89 .92 .87

Source: This study.

Table 3 shows the various correlations and mean values of the CWB-T factors. Lowest mean values are IPR with .36 denoting that faculty and staff are either not having much interactions (or the opportunity to interact) with parents or are quite cautious in dealing with their students’ parents. This is followed by IUR with .45, which is actually nearing the boundary of 50% chance of taking advantage of the schools’ resources. For the remaining six CWB-T factors, the mean values ranges from .52 to as high as .76 for RAD. The reason for faculty to shy away from taking on administrative responsibilities might be due to the financial incentives for faculty who have administrative responsibilities. It is noted that for some individuals the additional income might not be comparable to the additional time spent (having to come to work every day) as compared to the more flexible teaching work schedules. Analysis shows that the CWB-T factors are quite correlated with each other, denoting the occurrence of one CWB might lead to other deviant behaviors.

Table 2. Factor Loadings of CWB-T (N = 217) (continued)

Hu, Hung and Ching: Examining the Counterproductive Work Behaviors within Taiwan Academic Setting: A Pilot Study 75 Table 3. Correlations among the CWB-T Factors

factors Mean SD Skew 1 2 3 4 5 6 7 8

Note: All values of correlation are significant (p < .001).

Lastly, analysis was done on the various background demographics of the participants. Table 4 shows the results of the various tests of differences (T-test and ANOVA). Results actually show that there are no significant differences among the participants perceived CWB within their institutions. This means that no matter what gender, job position, or school location, both faculty and staff tends to have similar opinions towards CWB. In other words, CWB is non-selective and does exist in all levels of academic institutions.

Table 4. Test of Difference Against the Various Demographical Backgrounds

Gender Position Location

Note: All values of correlation are significant (p > .05).