• 沒有找到結果。

Quantitative approach is chosen for this research. This chapter seeks to elucidate the research method which comprises conceptual framework, research hypotheses, and research procedure. Data collection methods and instruments are also included.

Research Framework

The research framework is based on the research purposes and the literature reviewed in chapter two. This research seeks to analyze the impacts of KM strategies on Innovation and business performance. As theoretical and empirical studies suggested, KM strategies are assumed to have effects on Innovation and Business performance (Darroch, 2005; Edvinsson &

Sullivan, 1997; Vaccaro et al., 2010).

Figure 3.1. The Research Framework

H1

KM strategies

- Human-oriented - System-oriented

H2

Innovation

- Product - Process

Business performance

- Customer performance

- Financial performance H3

Research Hypotheses

As being discussed in Chapter 2, according to Botha, Kourie, and Snyman (2008), there are two reasons for one company to outperform the other in the same industry with similar access to knowledge and information. The first one is embedding in its human, structural, and relational capital, not in its tangible assets. The second one is depending largely on how the company exploits or manages these intangible assets. Considering these two reasons, it can be proposed that KM strategies positively affect companies’ business performance. Also, innovation process greatly relies upon knowledge (Gloet & Terziovski, 2004), and firms embrace innovations to attain first and early advantages that will produce excellent performance (Damanpour, Walker, &

Avellaneda, 2009), or to bridge a performance gap resulted from uncertainties in the external surroundings (Damanpour & Evan, 1984). Thus, the aim of this study is to investigate the interrelationships amongst the variables: KM strategies, Innovation, and Business performance.

All the variables are defined and integrated in the literature review. Based on the established framework, three alternative hypotheses are assumed and detailed as follows:

H1: Knowledge Management Strategies positively affect Business Performance.

H2: Knowledge Management Strategies positively affect Innovation.

H3a: Innovation positively affects Business performance.

H3b: Innovation mediates the effects of KM Strategies to Business Performance.

Research Procedure

The literature review on KM revealed several key topics commonly revolved around innovation and BP. Some articles concerning the significance of innovation has initiated an interest for the researcher to pursue studying. The literature also contained many studies related to KM and innovation that have been carried out in various fields and countries, but very few in the IT industry. This has prompted the researcher to focus on the IT industry. Although the topic was somewhat researched before, not all of the variables are investigated together and integrated into a theoretical model. This study will provide a more comprehensive perspective for KM strategies and innovation in order to help the business management field. This study was conducted by following the subsequent research process (see Figure 3.2).

As illustrated in Figure 3.2, first, the researcher identified problems that were worth studying. Next, research subject and variables were generally established. After reviewing relevant literature, research questions, hypotheses, and research framework were constituted;

methodology was chosen accordingly; measurement scales were searched for and later adapted from preliminary studies. Next, the measurements chosen underwent a strict review and translation (English to Chinese) from two separated translators and other two professors in the field. After the theoretical foundation of the study was well-researched, the first three chapters were completed for the proposal meeting. Pilot study was later conducted to detect necessary adjustments to the instruments. As the instruments were finalized, data for the main study was collected by online questionnaire. After that, data was coded and analyzed; results were produced, reported, and concluded. Thesis draft was completed for the final defense. Finally, feedback was provided for revision after the defense, and the thesis will be submitted after revision.

Figure 3.2. The Research Procedure Establish Research Questions and

Hypotheses

Developing Research Framework Identify research subject

Finding Measurements Identify Problems

Developing Methodology

Translation and Review of measurements

Proposal Meeting

Conducting Pilot Study

Review of Instrument

Data Collection

Data Analysis Data Coding

Thesis draft Completion Report Results and Conclusions

Final Defense

Thesis Submission Revision Review literature

Research Method

The study adopts quantitative approach. It is selected because it provides basis for the researcher to construct a theoretical model that explains what is being observed (Neill, 2007).

Also, a quantitative method approach facilitates the description, illustration and exploration of a given phenomenon. To explain the process through which variables interact with one another, as well as for the purpose of research validity, precise measurement is required. Quantitative approach allows researchers seek precise measurement objectively and analyze data gathered from questionnaires and surveys.

Survey is to be used for two reasons. First, as surveys are structured around key items and topics, they are easily interpreted and analyzed. Second, as respondents’ confidentiality is ensured, respondents are more likely to provide honest feedback (Debowski, 2007). This study will be conducted in IT industry context, whilst the results cannot be generalized, they can provide directions for future research.

Data Collection

The targeted population of the study is full-time employees working at IT companies. IT industry is a suitable choice for this research because of its fast-pacing nature, thus, innovation is considered a critical factor within the industry.

Employees were randomly selected and contacted through email, personal contact, and social media sites. Employees with self-evaluation can provide a more realistic account of the organization they belong to. To gather data, an online questionnaire was utilized. The instruments were first tested on a small group of randomly selected employees. The sample size chosen for the pilot study is 50. Convenient sampling method was employed for both pilot and main studies.

Instrument

The instrument has 22 questions in total that are divided into four parts: part I) KM strategies; part II) innovation; part III) business performance; and part IV) demographics. In part I, II, III, respondents were asked to rate every item in a 5-point Likert Scale (1 = strongly disagree, 2 = disagree, 3 = average, 4 = agree, 5 = strongly agree). For part IV, respondents were asked to provide some demographic information.

In sum, 3 research variables are included in the questionnaire – KM strategies, Innovation, and Business performance. The measurements are described in the following:

1. Knowledge Management Strategies (7 items): Adopted from (Choi & Lee, 2003), KM strategies consist of Human-oriented or KM_H and System-oriented or KM_S. Both are the approaches of knowledge management strategies based on whether knowledge is tacit or explicit.

2. Innovation (6 items): Innovation measurement is adopted from Bae and Lawler’s (2000) measurement of a firm’s product and service innovation.

3. Business performance (6 items): Adopted from Gonzalez-Padron, Chabowski, Hult, and Ketchen Jr. (2010), two dimensions of business performance are Customer performance (BP_C) and Financial performance (BP_F).

Part IV of the questionnaire is intended to gather information on respondents’ jobs, gender, and marital status. Participants are advised to fill out all information.

Reliability and Validity

Reliability represents the quality of the measurements. It is the degree to which a scale consistently measures what it is intended to measure. A Cronbach’s alpha of higher than 0.7 is the evidence of reliability (Nunnally & Bernstein, 1978). Validity of measurements is the extent to which scales measure what researchers claim they measure.

Construct validity comprises convergent validity and divergent validity. Before establishing construct validity, researchers need to run two analyses and obtain required values.

Convergent validity can be assessed using Composite Reliability and Average Variance Extracted values. Discriminant validity can be assessed by different criteria such as Fornell &

Larcker criterion or Heterotrait-monotrait (HTMT) ratio of correlation (Henseler, Ringle, Sarstedt, 2015). The basic difference between convergent and discriminant validity is that convergent validity tests whether items that should be related, are related, and discriminant validity tests whether proposed unrelated items are, in fact, unrelated.

The first test is the Kaiser-Meyer-Olkin (KMO). It measures Sampling Adequacy value.

This value reveals how suitable it is to perform factor analysis on measurement scales. KMO value of lower than 0.5 is unacceptable, higher than 0.6 indicating mediocre fit, higher than 0.7 indicating middling fit, and higher than 0.8 indicating excellent fit, which means factor analysis is useful in this case. The closer the KMO value is to 1.0, the better the indicators of the construct explain the variance of the data. To improve KMO value (especially when the value is under 0.5), deleting bad items or adding items is necessary. KMO test were run by SPSS 23.0.

The second test is the Bartlett’s test for Sphericity. The test declares whether or not the correlations among variables happen by chance. For Bartlett’s test, the small p-value, which is under .05, confirms that factor analysis is eligible to conduct. After those tests are run, if all the values reach the minimum standards, factor analysis can be performed. Bartlett’s test for Sphericity value will be tested using SPSS 23.0 software.

After the data passes requirements for KMO test and Bartlett’s test, it will undergo Confirmatory Factor Analysis (CFA) tests for factor loadings analysis, construct reliability, and construct validity. For factor loadings, if the item-to-total correlation score is below 0.4, the item is excluded from further analysis (Kerlinger, 1986). The reason is that the item may not be correlated with other items measuring the same variable. CFA value will be tested by using SPSS 23.0 software.

CFA is also used to test hypotheses. Items’ factor loadings are observed. Structural equation modeling software is typically used for performing CFA. In this study, CFA value will be tested by using SPSS 23.0 and Amos 23.0 software.

Data Analysis

Data collected for this study will be analyzed by three methods: Descriptive Statistics, Confirmatory Factor Analysis, and Structural Equation Modelling. Totally, 25 questions are coded by a 5- point Likert scale. Three demographic questions’ coding is given in Table 3.1 below:

Table 3.1.

Descriptive Statistics

Descriptive statistics are conducted to obtain general description of basic features of data, providing a summary of the sample. Descriptive statistics will help to identify which aspect of a variable that has the most impact within the population under study. The method also interprets data by calculating numerical indices such as Standard Deviation (SD) and Means (M).

Confirmatory Factor Analysis

A confirmatory factor analysis (CFA) will be conducted to validate measurement scales and initially detect correlations among constructs. CFA is a statistical technique used to verify the structure of a set of observed variables (or indicators). CFA allows researchers to test whether a relationship between observed variables and their underlying latent construct exits.

Multi-collinearity can also be identified through CFA. Multi-collinearity occurs when two variables are closely correlated to one another. This indicates that they are likely to measure the same construct. SPSS 23.0 and SmartPLS 3.0 will be used to perform this analysis.

Structural Equation Modelling

Structural Equation Modelling (SEM) technique will be used to analyze structural relationships among variables. This technique is a multivariate statistical analysis technique that combines factor analysis and multiple regression analysis. There are two types of variables in SEM: endogenous (or dependent) variable, and exogenous (or independent) variable.

CHAPTER IV FINDINGS AND DISCUSSION

相關文件