Quantitative approach was conducted in this study in order to measure the effect of independent variables towards dependent variables. This chapter focused on the research methodology which comprised of several sections, research framework, research procedure, research method, participant, population, and the approach for the instrumentation
development.
Research Framework
Follow the logic of literature review, a model was proposed. The model was based on Trust Model by Shockley-Zalabak, P., Ellis, K., & Cesaria, R. (2003), Organizational Learning Model by Pilar Jerez-Gómez, José Céspedes-Lorente, Ramón Valle-Cabrera (2005), Successful factors towards innovation Model by Van der Panne et al. (2003); Innovation Model by Cheng-Ping Shih and Deh-Lun Tzeng (2009), and Business Performance Model by Emden et al. (2005). In Figure 3.1, TSLIP model was developed by Cheng-Ping Shih and Yi-Ping Tsai. TSLIP model served as the research framework for this study (see figure 3.1).
22
Figure33.1 TSLIP model, developed by C.P. Shih and Y.P. Tsai.
Innovation Process innovation Technological innovation Organizational innovation Successful factors towards
innovation
Organization-related factors Project-related factors Human-related factors Market-related factors Demographics-related factors
Organizational learning Management commitment
System perspective Openness and experimentation Knowledge transfer and
integrate Trust
Concern for employee Openness and honesty
Identification Reliability Competence
Business performance Partnership performance
Market performance Financial performance
H1
H2
H4 H3
H5
23
Research Hypotheses
Based on research questions, literature review, and research framework, the following null-hypotheses were formulated as follows:
H1: Trust have no effect on Innovation.
H2: Trust has no effect on organizational learning.
H3: Organizational learning has no effect on Innovation.
H4: Successful factors towards innovation has no effect on Innovation.
H5: Innovation has no effect on business performance.
Research Procedure
After researcher reviewed some literature on innovation, there are several key topics commonly related to organizational learning and business performance. Besides, there are some articles show the importance of trust. It resulted a huge interested for researcher to study it. The literature also showed many studies related to innovation has been conducted on different fields and different countries, but seldom of them did the research in ICT industry, this has prompted this research to focus on ICT industry in Taiwan. Moreover, even the topic was previously well researched, not all of constructs are investigated together and described as a comprehensive theory model for both theoretic and practical field. This study will provide a more comprehensive perspective for innovation in order to help HR field. This study was conducted by following the subsequent research process (see figure 3.2):
24 Figure43.2 Research Process.
Conduct Pilot Study
Review of Instrument
Data Collection
Data Analysis Data Coding
Conclusion and Suggestions Proposal Meeting
Identify Research Questions and Hypothesis Review of Literature
Develop Theoretical Framework Research Motivation
Develop Instrument Identify Problems
Develop Research Method
Translation and Expert Review of Instrument
25
Research Method
The purpose of this study is to investigate the relationship and the effects of trust on organizational learning and the relationship and the effects of successful factors towards innovation on innovation as well as business performance. The relationship and the effects of organizational learning on innovation as well as business performance in ICT industry, a quantitative approach was perceived as the appropriate methodology. There are several reasons for using quantitative approach in this research. Firstly, quantitative approach can help research to classify, count and construct a statistic model, and help researcher explain the finding of this study (Neill, 2007). Secondly, Quantitative approach provided a more objective way for the research. Within the questionnaire and surveys, this research can make the study more precise and correct. The more objective this research be, the results will be more accurate (Neill, 2007).
There are several reasons for using a survey. First, surveys can be analyzed and interpreted quickly, because they are built with the specific topics and key words. Second, it made respondents feel more comfortable, and they can be able to honestly provide their perspectives without any pressures. Since the purpose was investigating what factors may affect employees’ behavior and thinking, this study used self-reported as a measurement.
According to Renrsch (1990), instead of individual’s perception of objective reality, individual’s perception of reality can have more affect to human’s behaviors and attitudes.
Self-reporting questionnaire surveys allowed respondents to report their perceptions of reality, it was considered as the most suitable measurement for the research of individual’s perception (Howard, 1994; Spector, 2011). This research will focus on Taiwanese ICT industry, even though the results can’t be generalized, the results can provide as a deep data for the new theory development.
Population
The population for this study consist full-time employees and managers who were working for H Multinational Corporation in Taiwan. The total employees in H Multinational Corporation in Taiwan were 100.
Respondents were selected randomly. Paper questionnaire was used to gather data from staffs. The researcher gave the hard copied to the assistant of H Multinational Corporation in Taiwan, and the assistant gave the hard copies randomly. Employees with self-report can provide a more realistic perceptions to the study. The sample size chosen for the pilot study
26
was 37 randomly chosen employees. The size of sample may be less, but 35 to 40 sample would be preferable if estimating test-retest reliability (Hertzog, von Oertzen, Ghisletta, &
Lindenverger, 2008). For the main study, 80 paper questionnaires were sent to the H Multinational Corporation in Taiwan.
Instrumentation
The instrument consisted of 5 variables with total amount of 78 questions. After the pilot test, a statistical analysis was initially conducted to see the validity of the instrument. Some items were dropped as either the items could not pass the minimum requirement of reliability test or factor analysis test.
There were six parts included in the questions; which were: part I) basic information;
part II) trust; part III) organizational learning; part IV) successful factors towards innovation;
part V) innovation; and part VI) business performance. The design of questions was adopted from 5-point Likert Scale. For part I, respondents were asked to choose one of different available options. For part II to part VI, respondents were asked to rate each item with scale anchors ranging from “strongly agree” (5) to “strongly disagree” (1).
Part II trust, consist of 5 factors, in total 20 questions were adopted from Shockley-Zalabak, P., Ellis, K., & Cesaria, R. (2003), because the research separated trust into a comprehensive view. The questionnaire were modified and adopted for this research.
The validity and reliability of the items were then tested in the pilot test.
Part III organizational learning, consist of 4 items, in total 15 questions were adopted from Pilar Jerez-Gómez, José Céspedes-Lorente, Ramón Valle-Cabrera (2005). There are several literature investigate organization learning, the reason adopted from previous research was that research provided a systemic view of organization learning; besides, the reliability and validity are really significant. The questionnaire were modified and adopted for this research. The validity and reliability of the 15 items were also being tested in the pilot test.
Part IV and Part V successful factors towards innovation, and innovation, consist of 8 items, the 31 questions were adopted from Van der Panne et al. (2003). The questionnaire were modified and adopted for this research. The validity and reliability of the items were also being tested in the pilot test.
Part VI business performance, the 12 questions were adopted from Emden et al.
(2005), there’s no previous literature added partnership performance into its’ research model,
27
this research provided a more comprehensive view to business performance field. The validity and reliability of the 12 items were also being tested in the pilot test.
Moreover, peer reviews and expert review were utilized to preserve the validity of the instrument. Pilot test was initially conducted to ensure the validity of each item, before gathering the whole data for the study.
Validity and Reliability of Instrument Face Validity
The reliability of questionnaires should be reexamined again because items are from different studies and each study had different samples. In this study, the reliability of these items were reexamined in a pilot test. Reliability of instrument is the external and internal consistency of measurement. Besides, validity of instrument is the degree to which scales measure what researchers claim they measure (Williams & Monge, 2001).
In order to make respondents have a better understanding about questions, the questionnaire was available in both English and Chinese Language. Originally the instrument was gathered in English. Researcher translated English into Chinese. The translation was then revised by two bilingual experts, English – Chinese language translators. The questionnaire was also conducted a back translation to ensure the meaning of each items were not changed.
The questionnaire was again reviewed by a Chinese academic professor from international human resource department.
Construct Validity
Before testing these factor analyses, there are two analyses that needed to be performed in advance. First is the Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy value.
This value indicates the appropriateness of using factor analysis on data. KMO value has to be greater than .80, it can be considered excellent, as it is an indication that component or factor analysis will be useful for these variables. The closer the KMO value is to 1.00, the better the factors extracted from the data will account for the variance in the data (Friel, 2010). However, if the value is above .70, it can also be considered as acceptable. When KMO less than .50, there is a need for remedial action, either by deleting the offending variables or by including other variables related to the offenders. KMO value was tested by using SPSS 20.0 software.Second is the Bartlett’s test for Sphericity. The value tells whether or not the correlations between variables happened by chance. In Bartlett’s test, the small
28
p-value, which is less than .05, confirmed that a factor analysis may be useful with the data gathered.
After the data can pass requirement for KMO test and Bartlett’s test for Sphericity, the data can then undergo the construct validity or factor analyses tests. The convergent validity is established by conducting EFA (factor loadings). EFA is used to test if the factors of each scale are consistent with those of previous studies. The correlations between item’s score with the total score of the dimension are used. If the item-to-total correlations are less than .40, the item is then dropped from further analysis (Kerlinger, 1986). This is because the item may not be correlated with other items that measure the same construct. EFA value was tested by using SPSS 20.0 software.
The divergent validity is established by conducting CFA. CFA is also used to test the hypothesis. Factor loadings of the items are observed. Structural equation modeling software is typically used for performing CFA. In this study, CFA value was tested by using SmartPLS 2.0 software.
Table 3.1
Reliability of Instrument Reliability
Type Theoretical Meaning Reliability of Instrument
Reliability Analysis
The degree to which a scale consistently reflects the construct (Field, 2007)
Cronbach’s Alpha should be higher than .70, then the result is considered acceptable
Individual item reliability is also computed. If the factor loading are below 0.5, then the questions should be dropped, because it means these questions do not belong to a particular factors (Hair, Anderson & Tatham, 2006).
29 scientific community (Neuman, 2011).
The questionnaire should be examined and translated by native speakers. It should also be examined by two bilingual experts and one reviews and expert reviews.
Criterion
The questionnaire was also initially undergone pilot tests.
Construct Validity
It is a type of measurement validity that uses multiple indicators and has two subtypes:
how well the indicators of one construct converge or how well the indicators of different constructs diverge (Neuman, 2011).
Convergent validity is established by using Explanatory Factor Analysis (EFA). Before the EFA test, KMO and Bartlett’s test for Sphericity are conducted. If the items are confirmed then EFA can be conducted.
Divergent validity is established by using Confirmatory Factor Analysis (CFA).
Descriptive Statistics
Descriptive statistics are used to describe the main features of a collection of data quantitatively without employing a probabilistic formulation (Mann, 1995). This holistic overview provides a simple summary about the sample in order to have a better understanding of where the data comes from. This study used frequencies and percentage to compare the sample and population according to categorical variables gender, age, education background and working experience. There are two purposes in using this descriptive statistics. First, descriptive statistics helped to this study to identify whether the sample can represent the case study. Second, it helped this study to identify the main aspects of each variable that seems to be most influential within population under study. Mean and standard deviation are used in pilot test and main study. Descriptive statistics are tested by using SPSS 20.0 software.
30
Statistical Analysis
Validity and reliability test was conducted to ensure the accuracy of the measurements.
Reliability tests were conducted to see the consistency of participants’ answers of each items in the questionnaire. Reliability tests included Cronbach’s alpha item if deleted, Cronbach’s alpha coefficient, and Corrected item-total correlation. The data were then further analyzed by evaluating the coefficient of determination (R2), path coefficients, and t-value (t). These analyses were conducted by using SmartPLS 2.0 software.
Correlation Analysis
A correlation analysis is conducted in order to investigate the direction and strength of linear relationship between the independent variables and dependent variables. In the finding, researcher can conclude the relationship between dependent variable is positively or negatively related to independent variable. A high correlation coefficient may not necessarily imply multicollinearity existed between the variables. Since the data has been confirmed by CFA, the items investigated in the study are also confirmed. If CFA is confirmed and the correlation coefficient value is still high, it means that the variables are highly related, and the value of regression can still be accepted. Correlations analysis is tested by using SPSS 20.0 software.
Coefficient of Determination (R
2)
Coefficient of determination or regression is used to explain the endogenous latent variables to the total variance. This is to show the percentage of how much the data gathered can explain the real situation. The higher the percentage, the data is closer to explain and to predict in reality. Regression value is tested by using SmartPLS 2.0 software.
Structural Equation Modeling (SEM)
SEM is used to determine the relationship between set of interconnected variables. In order to test the model can be applied to explain a set of data and to test the model is applicable to explain the situation, this research will use SEM method. Because SEM is suitable for testing proposed research framework and test the validity. There are two core SEM techniques that applicable in this study. First is to perform CFA. Second is to perform causal modeling, or path analysis. Path coefficients values are used to judge the relationship between variables, determine the direction of the relationships and its significance. Partial
31
Least Square (PLS) is used to measure the path coefficients. The two values of SEM are used to offer an evidence to support or reject the hypothesis proposed in the beginning.
Partial Least Square (PLS).
PLS path-modeling algorithm is a type of SEM technique. It allows the simultaneous modeling of relationships among multiple constructs. It was developed by Herman Wold in 1975. PLS estimates path models using latent variables. It’s a tool to confirm both theory and predictions. There are several reasons why this research used PLS. Firstly, it is adequate for small sample analysis and complex models. Secondly, it has the ability to handle a complicated model which based on exploratory research.
32
33