Page 110 - 33-2

Service Innovation in the IT Service Industry: Social Influence and Relationship Exchange Perspectives

innovative ideas to deal with team tasks. We used four items based on Zhang and Bartol
(2010) to measure EML. We asked respondents about the extent to which they perceived
individual empowerment based on the four dimensions of leadership: (1) enhancing the
meaningfulness of work; (2) fostering participation in decision making; (3) expressing
confidence in high performance; and (4) providing autonomy by removing hindrances. To
measure TMX, we used four items developed by Seers (1989). We asked respondents to
rate the strength of TMX with their team.

To account for potential influences on individual innovation performance, we
controlled for employees’ education level, company tenure, and age. For example,
employees with advanced degrees may exhibit a wider array of ideas at work, those
with longer tenures at the same company may be more adept at generating novel ideas
(Zhang and Bartol, 2010), and older employees may bring a wealth of work- and life-
related experiences that could be advantageous in their tasks (Scott and Bruce, 1994).
Education level was categorized as follows: middle school or below, senior high school,

junior college, university, and postgraduate studies. Company tenure was measured by the
number of months an employee had worked at the company.

4. Research Analysis

According to the two-step approach of Anderson and Gerbing (1988), we perform

both measurement model evaluation and structural model evaluation. We first assess the
measurement properties of the latent constructs by running a Confirmatory Factor Analysis
(CFA); we then test all hypotheses of the conceptual model by conducting a structural
equation analysis.
We use LISREL 8.80 to run our research models, and evaluate the goodness-of-fit
of the models using chi-square tests, the Non-normed Fit Index (NNFI), the Root Mean
Square Error of Approximation (RMSEA), and the Comparative Fit Index (CFI). The

results also show good model fit, requiring nonsignificant chi-square tests, the values of
RMSEA less than or equal to .08, and the values of NNFI and CFI greater than or equal to
2
.90 (Bagozzi and Yi, 1988). The present study also conducts χ -difference tests to verify
the effects of mediation and moderation.

4.1 Construct Validity
We adopt LISREL 8.80 to perform a CFA and determine the validity and reliability

102