This chapter constructs the methodological framework and the research method of the empirical study. The methodological framework is the instruction and guide for the choices of research methods and consists of philosophical assumptions, logical reasoning paths, and criteria of evaluating research results (Scotland, 2012). On the basis of positivism, deductive approach, and quantitative method, this chapter designs a questionnaire survey. The measurement scales are designed based on previous empirical studies; the procedures of data collection and analysis are discussed; the data quality is evaluated; ethical issues are fully addressed. This chapter closes with a reflection of methodological limitations.
In social research, researchers’ choices of research methods, criteria, and procedures are substantially affected by their basic assumptions, including epistemology, ontology, and axiology. The divergences in these assumptions lead to different research philosophies. In 2012, Saunders et al created the “Research Onion” which is one of the general guidelines for designing research processes and strategies (Figure 6). Follow the ‘Research Onion’, the first step of constructing a methodological framework is to choose the research philosophy according to the nature of research questions (Saunders et al, 2012).
In consumer behavior research, two research philosophies occupy the dominant roles, i.e., positivism and interpretivism. The two philosophies vary greatly in terms of their epistemological, ontological, and axiological assumptions (Hunt, 1991). Positivism holds a “positive” assumption of the existence of the external world, i.e., the world exists independently and objectively from people’s thoughts (epistemology). Therefore, the positivism advocates an objective perspective to probe into the external features of social phenomena. Meanwhile, the positivism believes that only observable facts can be used to construct accredited knowledge (ontology). Moreover, because of the objectivity, positivism adopts the principle of value-free, i.e., the researcher is separated from its research and serve as a complete outsider (Saunders, 2011). As a result, positivism is applicable if the researcher views its research objects as objective existences and seeks to identify the objective laws in social phenomena (Collis & Hussey, 2013).
By contrast, interpretivism suggests that the existence of the world is interpreted by people’s thoughts and is hence characterized by both subjectivity and objectivity (epistemology). Meanwhile, knowledge can be constructed by subjective meanings because the nature of social phenomena is subjective, multiple, and changeable (ontology). In view of this, it is important for researchers to be involved in their studies in order to interpret social phenomena and construct knowledge. In this way, researchers are parts of their studies (Saunders, 2011). Therefore, interpretivism is applicable if the researcher views its research objects as subjective existences and intend to understand the rich and in-depth connotations of social phenomena (Collis & Hussey, 2013).
In this study, the positivism philosophy was applied. The reason is that the positivism philosophy refers to applying existing theories and depends on quantifiable observations which lead to statistical analysis (Saunders, 2011). The UTAUT2 model was applied to develop hypothesis in this research, which limited researcher to do interpretation in an objective way. In view of this, positivism is adopted by this study.
At the first stage of research design, the research method choice is the primary task. This part will conclude three classification ways, exploratory, descriptive or causal research, quantitative or qualitative research, deductive or inductive research (Table 4). The detailed selection process will be presented as below.
Exploratory, Descriptive or Causal Research
According to Babin & Zikmund (2016), the research method includes exploratory, descriptive or causal research. The exploratory research always be used to define problem and clarify research in initial stage without conclusive evidence which needs additional research (Babin & Zikmund, 2016). The descriptive study is to provide an accurate and valid presentation of the variables pertaining to the research question (Jackson, 2011). It describes characteristics of people, objects, groups or environments and is more structured than the exploratory research. Accuracy is critically important in descriptive research. However, such research does not provide causality and direct evidence which need further research. The causal study is employed when causal inferences are needed to be made (Babin & Zikmund, 2016). It could be a continuation of exploratory or descriptive research (Williams, 2007). In this research, since this research intended to identify the causal relations between the choosing factors from UTAUT2 model and user acceptance, causal research was used.
Qualitative & quantitative research
Research methods in social research can be generally classified into three types, i.e., the quantitative method, the qualitative method, and the mixed method. The distinctions in these methods are not simply the expression of research data, i.e., numbers or texts, but the different philosophical assumptions (Östlund et al., 2011). Positivism advocates the use of the quantitative method because numbers have single, concrete, objective, and accurate meanings, which is consistent with the epistemology and ontology of positivism. Meanwhile, the meanings of numbers cannot be discretionarily interpreted by researchers and hence using numbers ensures the value-free principle insisted by positivism (Saunders, 2011). The use of numbers can accurately describe the external characteristics of social phenomena, such as extent, level, size, and strength. As a result, the quantitative method is conducive for the identification of the relationship between different things because the relationship is an external characteristic (McCusker & Gunaydin, 2015). Evidently, the seven research hypotheses describe the relationship between users’ technological acceptance and seven factors. Hence, the quantitative method is applicable to test these hypotheses.
By contrast, the qualitative method is underpinned by interpretivism which emphasizes the multiplicity, subjectivity, complexity, and dynamics of social phenomena. Texts have multiple, subjective, complex, and flexible meanings and are hence applicable to describe social phenomena from an interpretivist perspective (McCusker & Gunaydin, 2015). This study does not adopt the qualitative method because it does not allow the researcher to test its research hypotheses in an objective, accurate, and rigorous manner.
Deductive or inductive research
There are two opposite logical thinking approaches, by which researchers connect theories with observations, including deduction and induction. As a result, empirical research can be generally divided into deductive research and inductive research (Heit & Rotello, 2010). Deductive research is applicable to test the previously established theories in a specific context. Commonly, research hypotheses are developed based on theories and subsequently, the researcher collects empirical evidence to verify whether these hypotheses are true in a specific context. In this way, the generally applicable laws (i.e., theories) are reduced to specific laws (disconfirmation or confirmation of hypotheses) (Yvonne Feilzer, 2010). By contrast, an inductive approach is applicable to explore new phenomena or construct new theories. The empirical observation is commonly not constrained by research hypotheses. Instead, it is guided by specific research questions. In this way, the researcher extracts the commonalities from a set of observations and develops new hypotheses or theories (Yvonne Feilzer, 2010).
In the present study, it uses the UTAUT2 theory to explain Chinese users’ technology acceptance of mHealth Apps and hence develops 7 research hypotheses, moves from the general to specific. Testing these research hypotheses will fully address the research questions raised in introduction chapter. In view of this, a deductive approach is workable to this study. This study does not adopt an inductive approach because of two reasons. First, developing new theories or hypotheses falls outside the scope of this study. Second, the conclusions drawn from an inductive research lack of logical rigor and objectivity (Heit & Rotello, 2010).
As shown in Table 3, all the 8 empirical studies pertaining to users’ technological acceptance of mHealth used a questionnaire survey to collect their primary data. The questionnaire survey refers to a structured and standardized method of interviewing a large group of individuals in order to understand their thoughts, beliefs, attitudes, experiences, backgrounds (Collis & Hussey, 2013). Hence, it is an appropriate strategy to measure the variables involved in the seven research hypotheses on a large scale.
Using a questionnaire survey brings the following benefits to this study. First, it best maximizes the objectivity of the data. In the data collection process, researchers are separated from research subjects, which avoids the interferences from researchers. Besides, unlike an in-depth interview, the data collection process was less likely to be influenced by researchers because the survey questions are highly structured and standardized (Fricker & Schonlau, 2002). Second, because the survey data are in a standardized and structured form, data analysis is convenient and the application of statistical tools further mitigates the interference from researchers (Melkert & Vos, 2010). Third, the questionnaire survey allows researchers to interview a large number of research subjects simultaneously and collect a large sample with relatively cheap costs (Collis & Hussey, 2013).
It should be noted that a questionnaire survey also has the following disadvantages. First, the standardized data collection process provides very limited flexibility and it is hence difficult for researchers to collect in-depth information. Research subjects only passively answer a set of fixed questions and hence they are not free to express their true thoughts (Couper, 2008). Second, the response rate, validity, and reliability of the survey data are difficult to ensure. It is difficult to use a set of structured questions to reflect the essence of social phenomena. Moreover, respondents may not take the questionnaire seriously or inaccurately understand the meanings of questions (Hoonakker & Carayon, 2009). In view of this, it is important to carefully and prudently design the measurement scales in order to enhance the data quality.
According to the overview of previous empirical studies presented in Table 3, all the variables included in the theoretical model of this study have been covered by previous studies. Hence, the measurements used in these studies can be borrowed by this thesis after slight modifications. The design of the draft questionnaire followed the following procedures. First, the author reviewed the questionnaire scales used in previous empirical studies. Second, the author extracted relevant questionnaire scales from these studies. Third, the author carefully compared the different measurements of the same variables and chose the most appropriate ones. Finally, the author made modifications to these measurements according to the research context of this study.
As shown in Table 5, the questionnaire involves 11 variables, including 1 dependent variable (i.e., user’s acceptance towards mHealth App), 6 independent variables (i.e., factors that influence technological acceptance), and 4 control variables (gender, age, education, and experience). Each dependent and independent variable was measured by multiple Likert-7-Point items, i.e., respondents were asked to self-report their degrees of agreements from 1 (totally disagree) to 7 (totally agree) towards a set of statements, which describe their attitudes and experiences towards mHealth apps. In this way, these 7 variables were transferred into numbers ranging between 1 and 7. The 4 control variables were measured by single choice questions.
After the draft questionnaire was accomplished, the author translated it into standardized Chinese and asked two professional bilinguals to scrutinize and polish the Chinese translation.
As suggested by Couper (2008), the pilot test is critical for the success of a questionnaire survey. A pilot test helps the researcher to correct the mistakes and improve the quality of the questionnaire design. The researcher cannot clarify the concerns of every respondent face-to-face in the formal survey and respondents’ concerns may reduce data quality. Hence, the pilot survey can mitigate these concerns before the formal survey. Moreover, the questionnaire was designed based on English versions of scales and hence it is necessary to evaluate whether the Chinese version of the questionnaire would be workable in the formal survey with Chinese people. In view of this, the author recruited 14 initial respondents to take part in a pilot test.
During the pilot test, the following things have been done. First, the author preliminarily analyzed the data and found that the reliability was basically acceptable, and the results largely matched with expectations. Second, the author modified the expression and wording of the questionnaire content and confirmed that the content can be easily and accurately understood by respondents. Third, the author confirmed that respondents generally felt comfortable towards the questionnaire content and there was no sensitive information. After the pilot test was accomplished, the author carried out the formal data collection.
Primary data refers to the process of conducting researches by the researcher to meet specific and unique needs, while the secondary data is to use previously completed studies and apply the results to own situation. However, if the researcher wants to study a specific research question, the secondary data may not meet the needs of the researcher and cannot cover the precise and specific information that required to answer the research questions (Johnston, 2017). In this view, primary data will be applied in this study, so that the researcher has control over what their data set contains.
In this research, a primary data collection was conducted through a questionnaire. Suitable respondents were those individuals who needed healthcare services, i.e., the target users of mHealth apps. In order to identify and get in touch with these individuals, the author joined several social media health related communities, where members frequently discussed and interacted healthcare information. The questionnaire was uploaded on a Chinese survey website (www.wjx.cn) and the questionnaire URL and QR code were shared in these healthcare communities and other social media (QQ and Wechat) groups. Using smartphones or personal computers, members can either click the link or scan the QR code to get access to and answer the questionnaire.
Therefore, due to the budget and time limitation, this study used a convenience sampling technique, which is a non-random sampling method where research subjects are selected because of their convenient proximity and accessibility to the research (Fink, 2003). Although this technique cannot secure the equivalent probability by which each research subject is selected (Fink, 2003), it helped the author to accurately locate and get access to suitable respondents.
After the data were returned, the author organized the data in Excel and input them into SPSS software. First, in order to outline the characteristics and the distribution of the sample, frequency statistics were implemented on the four control variables (i.e., gender, age, education, and experience).Second, in order to evaluate and purify the quality of the Likert-7-Point scale data, a Reliability analysis (Cronbach’s alpha test) was implemented on the measurement of each variable. Also, factor analysis was used to access the factorability of the data (Bartlett’s) and make variables more suitable for managing number of factors. Items with insufficient reliability were dropped. Third, a descriptive analysis on the 7 variables was implemented in order to evaluate the levels of respondents’ intention to adopt and evaluations of mHealth apps. Fourth, the Pearson Correlation Analysis was employed to identify and check the bivariate correlations of 7 variables. In this way, the relationships between different variables and the potential collinearity can be evaluated. Finally, the hypotheses tested in this study was based on multivariable regression analysis. In previous studies, scholars used two different statistical techniques to test their research hypotheses, including multivariable regression analysis (Cilliers et al., 2018; Guo et al., 2015) and structural equation model (Chen & Lin, 2018; Choi et al., 2018; Dou et al., 2017; Dwivedi et al., 2016; Gao et al., 2015; Zhang et al., 2017). Although structural equation model is superior to multivariable regression analysis because it allows the use of multiple dependent variables, this study has only one dependent variable and hence multivariable regression analysis is sufficient to examine the research hypotheses.
Table of Contents
1.2 Problem Discussion
1.3 Research question
1.4 Research method
1.6 Expected Contribution
1.7 Research path
1.8 Key words .
2. Literature Review
2.1 Consumer behaviour in ICT
2.2 Technology acceptance models
2.3 mHealth technology
2.4 mHealth in China
2.5 Technology Acceptance of mHealth
2.6 The proposal research framework
2.7 Hypotheses Development
3. Research Methodology
3.1 Research Philosophy
3.2. Research approach
3.1 Research strategy
3.4. Data collection
3.5. Data Analysis
3.6. Data Quality
3.7. Ethical Issues
4.1. Profile of Respondents
4.2. Factor Analysis
4.3 Reliability analysis
4.4. Descriptive Analysis
4.5. Correlation Analysis
4.6. Regression Analysis
5.1. Hypotheses Testing Outcomes
5.2. Technological Acceptance of mHealth apps
5.3. Performance Expectancy
5.4. Effort Expectancy
5.5. Social Influence
5.6. Facilitating Conditions
5.7. Hedonic Motivation
5.8. Privacy and Security
6.1 Purpose and Research Questions
6.2 Implication for practice
6.4 Future research
GET THE COMPLETE PROJECT
User Acceptance in mHealth industry