Research methodology and method selection 1.1 Philosophical Foundation of the Research
As mentioned earlier the aim of this research is to gain knowledge of how virtual community members form impression about a company that is using those platforms for corporate communication activities. In the knowledge generation process there are four key concepts, called the ‘ologies’, which describe how we think about reality (Lee and Lings, 2008):
Ontology: Ontology is about how the researcher regards reality. Is reality objective that means it exists apart from our experiences or is it constructed through our experiences (Lee and Lings, 2008)? Ontology comprises the perspective of realist and constructivist. The realist perspective assumes that reality is objective and an external reality does exist (May, 1997) while the constructivist perspective is based on the assumption that reality is constructed through our experiences (Corbetta, 2003).
Epistemology: The epistemology follows the ontology and is the assumption about how we construct knowledge of the reality (Bernard, 2000) and what we can know about it (Lee and Lings, 2008). In social research there exist two main epistemological assumption: postitivism and interpretivism (e.g. Bernard, 2000; Corbetta, 2003). Positivism refers to the assumption that there is an objective truth that exists independent of people’s perceptions and thus phenomena can be observed (Corbetta, 2003). Interpretivism is about interpreting what we can see and thus the researcher constructs theory that explains the observed phenomena (Payne and Payne, 2004).
Axiology: It also follows from ontology and is about the aim of the research. In Lee and Lings’s (2008) words, it is about what we are trying to do: “for example, do you try to explain and predict the world, or are you only seeking to understand it? Can you even do one without the other?” (p. 11).
Methodology: Is the strategy of action that govern the choice and use of methods. It asked ‘how’ we are going to do this research (e.g. by using qualitative or quantitative methods). Hence, methodology and methods are two distinct terms. While the methodology outlines the philosophical assumptions taken that justify the use of the chosen methods, the methods refer to the techniques and procedures we propose to use for data collection and analysis (Payne and Payne, 2004).
The researcher’s stance of this research is that of a critical realist. This philosophical view is based on positivism but criticises that the “positivist conceptions of science only consider things to exist if they are directly observable. And any proposition which cannot be directly empirically tested is nonsense” (Lee and Lings, 2008 p. 31). The realist philosophy assumes that there are terms such as ‘motivation’ that cannot be directly observed but still exist and have independent existence (Lee and Lings, 2008). Thus, the underlying logic of this research is based on the assumption that the world exist independently of our experiences and that social phenomena exist regardless of a researcher’s interpretation of them. The researcher will not only investigate the causal link between corporate communication activities in virtual communities and the virtual corporate image and its underlying structures but also try to make sense of it.
Prior to the main survey study testing the causal links, a preliminary fieldwork with an inductive approach will be conducted. Therefore a pragmatic view is taken (Robey, 1996). The pragmatic view does not ascertain that the “truth is absolute and objective; but that it is co-created by us and the reality we are working within. In other words, only when a theory proves useful does it become true. So, theory and practice are not independent, they are inextricably interlinked” (Lee and Lings, 2008 p. 33). This view is important for this study as virtual corporate image in virtual communities has not been measured by now and is supposed to be a construct that is co-created by the individual.
Churchill’s (1979) paradigm will be employed by using a quantitative approach with the use of multi-methods in the early stage of the research. The reason for this is that even if virtual corporate image might to some extent be construed socially as the community members influence each other, the fact that people form images about companies is considered to be an external reality. This study aims to understand the relationship between virtual corporate image and its antecedents and consequences as well as the generative systems underlying such relationships. Through exploratory fieldwork the researcher aims to gain insights into how virtual corporate image is formed and then theory will be verified and thus the hypothetico-deductive approach (Burrell and Morgan, 1979) will be followed.
1.2 Three approaches to research
The previous section has discussed the research philosophies, which determine how a researcher sees the world. Next, the research approach that means, the specific methods used, should be discussed further. Social research often makes a clear distinction between qualitative and quantitative approaches (Bernard, 2000, Payne and Payne, 2004), while qualitative approaches are often referred to as being based on interpretevist assumptions and quantitative methods on positivist assumptions (Creswell, 2003). However, qualitative and quantitative methods do not have to be contrasting and can be used complementary. Campbell and Fiske (1959) are said to be the first who have used mixed methods (Creswell, 2003). The use of mixed methods in the same study is termed triangulation (Denzin, 1978). Triangulation is the use of different data collection methods within on study in order to ensure data are telling what we think they are telling (Churchill and Iacobucci, 2005).
There exist four types of triangulations (Neuman, 2003): (i) triangulation of measures (more than one measure is used to investigate the same phenomena), (ii) triangulation of observer (several observers are collecting the data), (iii) triangulation of theory (multiple theoretical perspectives are used) and (iv) triangulation of methods (mixing quantitative and qualitative methods). This research uses the triangulation of methods approach which means that data will be collected with qualitative and quantitative methods and at some point in time will be integrated, related or mixed (Creswell, 2003). By using the triangulation of methods approach the results from one method can inform or develop the other method (Creswell, 2003). By using the mixed-method approach “the researcher tends to base knowledge claims on pragmatic grounds… It employs strategies of inquiry that involve collecting data either simultaneously or sequentially to best understand research problems” (Creswell, 2003, p.18). This research will apply the sequential procedures, where qualitative methods, namely netnography and interviews will be used for exploratory purposes, followed by quantitative methods. Qualitative methods will be used to explore community member’s view about how they form a virtual corporate image in a virtual community, with the intention of using these views to develop and test an instrument with a sample from a population. The purpose of the quantitative self-administered survey research is to generalise from the sample to the population so that inferences can be made about antecedents and consequences of virtual corporate image formation of this population (Babbie, 1990).
Research design
This study will be a two-phase sequential mixed-method study. The research will start with qualitative explorations. The reason for doing so is (i) gaining knowledge about the virtual community (ii) refine and revise the proposed conceptual framework and hypothesis, (iii) develop measurement scales for measuring virtual corporate image and (vi) purify measurements for the main questionnaire. The qualitative exploration will be conducted by using netnography methods. In order to refine and revise the proposed conceptual framework and hypothesis the researcher will also conduct some in-depth interviews with academics and practitioners. The virtual community member will also be involved in the design and testing to ensure, rather than assume, what is important.
In a second phase quantitative methods will be used in form of an online survey. The questionnaire will be developed based on the literature review, expert interviews and the other qualitative results. The results of both phases will be mixed at the end of the research.
The first phase – exploratory fieldwork
In a first phase an exploratory fieldwork will be conducted. This will allow gaining insights into and understanding the communication context in the virtual community. Furthermore, it will allow to develop measures for the construct of virtual corporate image and to purify measures for the other constructs of the conceptual framework.
1.3 Rational for the context
The exploratory fieldwork involves three stages, for which netnography methods will be used. Stage one, involves the continuous observation of the virtual community in order to understand the communication context of the virtual community under observation. The second stage will use expert interviews with industry experts and academics in order to verify the conceptual framework and purify the proposed measures. It is important to include industry experts as the field of study is expected to move very quickly and industry expert’s knowledge is at least as advanced if not more advanced as the knowledge of academics. Finally, in stage three measurements for the focal construct (virtual corporate image) will be developed. This stage is used to define the virtual corporate image from a virtual community member’s perspective, hypothesise about the potential dimensions of VCI, identity the language of virtual community member’s for writing the survey, and identity a pool of potential VCI attributes that characterises each dimension. The insights will be quantified and verified by conclusive research in phase two.
Because netnography methods will be used, a brief insight into netnography will be given below.
1.3.1 Netnography
Netnography is a qualitative interpretive online research method that extends the traditional techniques of ethnography. Ethnography refers to the description of people and their culture (Schwarz and Jacobs, 1979). Netnography is a “modern” version of ethnography that has been adopted to study online cultures or in Kozinets’ (1997) words it is a ”written account of on-line cyber culture, informed by the methods of cultural anthropology” (p. 3).
Netnography is not based on scientific realism but constructs meaning in online social groups in an open-ended manner and thus differs from the general scientific realism approach taken by the researcher. Its fieldwork and textual account is methodologically based on the “traditions and techniques of cultural anthropology” (Kozinets, 1998 p. 369). Kozinets (1998) states that “Observing the general guidelines and traditions of ethnography while adapting them to the unique circumstances of cyber cultures, netnography may be empowered and legitimated through building on anthropological tradition, adapting and drawing on its consensually-derived standards of evaluation where necessary” (p. 369)
Kozinets (1997, 1998, 1999, 2002) is the first to have recognised the potential of netnography, followed by many researchers studying virtual communities such as Star Wars fan community and Citroen brand community (Cova and Carrère, 2002), Napster (Giesler and Pohlmann, 2003), football fan cultures (Richardson, 2005) and wedding planning (Nelson and Otnes, 2005).
Netnography is pursued in computer-mediated settings and allows studying online cultures such as communities and behaviours taking place in those communities. This study is focussing on the behaviour of communication taking place in virtual communities. Netnography is “based primarily on the observation of textual discourse” (Kozinets, 2002 p. 64). Netnographers join an online social group, interact with group members in order to get a rich picture of the culture and context. Virtual communities have a real existence for their members and therefore influence many aspects of behaviour, including consumer behaviour (Bym 1995; Turkle, 1995). They may also be said to form manifest cultures, which have also been recognised in cultural anthropology (Escobar, 1994; Kozinets, 1998). As has been outlined in depth in the previous chapters computer-mediated environments have their peculiarities, which also explain the differences between ethnography and netnography. In this text-based environments non-verbal context cues cannot easily be transmitted and might need to be interpreted by the researcher. The interactions between community participants are automatically saved and thus create a permanent record that can easily be accessed and handled.
Kozinets (1997, 1998, 1999, 2002) has adopted the common ethnographic procedures to netnography and proposes some steps that can be followed. He states “although netnography, like ethnography, is inherently flexible and adaptable to the interests and skill set of the individual marketing researcher, these steps may act as a guide to researchers who are interested in rigorously applying the method to their own research” (Kozinets, 2002 p. 63)
Kozinets (2002) proposed the following methodological stages and procedures:
1) Cultural entrée: A researcher needs first to formulate a research question and then select an appropriate virtual community. The studied virtual community will be selected according to the most important criteria that are: (i) abundance of member-generated contributions; (ii) lively participation and high traffic; (iii) large number of members; (iv) enough variation among them (participation and characteristics). Netnographers are said to have an easier entrée as ethnographers because for instance no long distance journeys need to be undertaken. A researcher can simply turn on the computer, join a group, start reading the community member’s contributions without any obstruction, and post some own contributions (Kozinets, 1998).
2) Data collection and analysis: Two kinds of data are generated by netnographer: (i) data that are directly copied from the virtual community and the researcher’s notes about his observations. Data directly copied typically consist of downloaded text files, e-mail exchange, picture and sound files (Kozinets, 1998) and (ii) researcher’s inscription of his observations. “The netnographer’s choices of which data to save and which to pursue are important and should be guided by the research question and available resources (e.g. the number of online members willing to be interviewed, the ability of online members to express themselves, time, researcher skill)” (Kozinets, 2002 p. 64). By following the “Pareto” rule of 80-20 the most important and relevant information for the specific study is likely to be captured (Kozinets, 1999). A taxonomy of collected data will be developed in order to classify the postings. They will, for instance, be classified by social or informational content and if the messages are regarding the topic or not. A further categorisation could be made based on members categories, such as ‘tourists’ (low consumptive, low social interaction), ‘minglers’ (low consumptive, high social interaction), ‘devotees’ (high consumptive, low social interactions) and ‘insiders’ (high consumptive, high social interactions) (Kozinets, 2002).
3) Ensuring trustworthy interpretation: As stated by Kozinets (2002) in most qualitative consumer research, instead of the concept of validity, the researchers use the concept of trustworthiness (Lincoln and Guba, 1985; Wallendorf and Belk, 1989). In contrast to ethnography where the researcher balances observations and discourse, in netnography the analysis is primarily based on text messages (Arnould and Wallendorf, 1994). As thoroughly discussed in Chapter 2 and 3 text only computer-mediated messages have their peculiarities such as overestimation of messages and idealised self-presentation. Therefore, “every aspect of the “game” (the ct, type, and content of the posting; the medium, and so on) is relevant observational data in itself, capable of being trustworthy” (Kozinets, 2002 p. 65). Thus the crucial difference between ethnography and netnography is that the researcher is not analysing the “complete set of observed acts of consumer in a particular community” (p. 65) but the “content of an online community’s communicative acts” (Kozinets, 2002 p. 65). Therefore, to be trustworthy this limitation must clearly be stated.
4) Research ethics: It is not clear if virtual communities are private or public space or hybrid. That means that in contrast to other methods such as interviews, focus groups surveys, the researcher is not explicitly given the information for research. According to Langer and Beckmann (2005) “it has to be decided from case to case… whether we deal with (semi-) private communication or public communication. The key to this decision is the access criteria for observation of and/or participation in such communication: if access is restricted (e.g. by use of passwords) and thus reserved for members only, we can talk about a (semi-) private communication within the community and should apply the guidelines and procedures, Kozinets recommends” (p. 194). Those four recommended procedures can be summarised as follows: (i) researcher’s presence must fully be disclosed, (ii) confidentionality and anonymity must be ensured, (iii) community member’s feedback should be sought and included, and (iv) permission to quote postings should be obtained (Kozinets, 2002).
5) Member check: To authenticate meanings community members will be allowed to check reports and interpretations (Arnauld and Wallendorf, 1994).
The researcher will adopt those steps for the study.
Netnography can be adopted as a rigorous method for studying online communities when following the subsequent guidelines based on ethnography: “Prolonged engagement and persistent observation, triangulation of sources, recording of field notes, and member checks seem to me to be the most important methodological techniques” (p. 370). The researcher considers this method as appropriate in order to understand the communication activities in the virtual community, finding key informants for conducting individual interviews, and also conducting the individual interviews.
Stage three of this exploratory research is a very important stage of the whole research and merits some more considerations. Thus, the process of measurement development is described in the following section.
1.3.2 Measurement development for the focal construct – VCI
This part of the paper will provide discussion about the development of measurement for the virtual corporate image construct. Based on the paradigm formulated by Churchill (1979), the following scale develop measurement will be conducted. Since, there are very limited or no literatures available about virtual corporate image, the measures that will be adapted in this research will be based on the context of corporate image. Accordingly, corporate image referred to how a corporation or industry is being perceived. Generally, it is an accepted image of what an industry stands for. The generation of a corporate image is a practice in perception management which is primarily established by marketing experts who use public relations and other types of promotion to recommend a mental picture or image to the general public. Consequently, a corporate image is designed as appealing to the public in order for the company to generate an interest among their target audience, create share of mind, establish brand equity and facilitate sales on products or services.
Accordingly, corporate image is an essential organizational resource which enables to generate, strengthen and sustain competitive position. In order to determine whether an organization has a good corporate image, corporate image measurement or audit can be considered. There are some factors defined to be the desirable corporate image. The image profile is known to be a detailed description of present or desirable corporate image. Image profile can be divided into three elements: visual; verbal; and behavioural. The visual elements include the corporate style, corporate layout, brand, personnel or employee appearance, interior, exterior, lighting, goods, cleanliness and others. Verbal elements includes corporate advertisements, public discussion with specific stakeholder groups and the behavioural elements include financial results, management, quality of the products and services, social responsibilities, strategic positioning and others.
After identifying and considering corporate image contexts, the next step is the development of scales for measuring Virtual corporate image. Since this study focuses on virtual community, the corporate image that would be given emphasis includes employee appearance, management and the strategic positioning.
In order to identify items in-depth interviews with key players of the virtual community will be conducted.
A pool of dimension and items coming directly form the qualitative interviews will be used to develop a questionnaire. A first measurement testing will be conducted in order to analyse the quality of the measurement instrument and the usefulness of the results. This is critical to success of any good measurement system. Statistical analysis will be conducted to assess reliability and validity. The reliability of each scale will be estimated by generating its coefficient alpha.
A pilot test will be conducted in order to refine the questionnaire. The data will again be analysed via a series of statistical methods to assess the validity and reliability. For this study, the questionnaire survey will be used to measure Virtual Corporate Image based on the corporate image elements. In this regard, this scale of measurement will be provided to the managers and selected employees of a company which uses virtual communication.
The survey questionnaire includes the following Scale development measure?
1. Do the virtual community give value to their employees?
2. Does the virtual community have been able to manage the company effectively?
3. Do the virtual community consider employee trainings and development?
4. Does virtual community value stakeholders efficiently?
5. Does the virtual community use strategic management approach?
6. Do the virtual community are able to construct a strong virtual brand?
7. Do the virtual community use efficient strategic positioning to stay competitive in the virtual environment?
8. Does the virtual community create good employee relationship through effective virtual communication?
The following questions are samples of the actual survey questionnaires to measure Virtual Corporate Image.
The second phase 1.4 Research instrument and scale development
Virtual community members will be interviewed with an online structured questionnaire (population: participants in VC). Several studies have attempted to determine the validity of email surveys (comparable to Internet surveys) by comparing the results of surveys sent by post and surveys sent by email (Kiesler and Sproull, 1986, Bachman et al., 1996). No difference between both methods of delivery could be found in terms of response rate and the mean and rang of Liker-type responses. However, the electronic surveys had a faster response time and were less expensive. Further, respondents of electronic surveys seemed pay less attention to social desirability and gave more open and broad responses to open questions.
A further issue to consider is the question if some people of the population would be inhibited from participating at the survey and if respondents would answer questions differently than if they would use another method of delivery. The target population consists of virtual community members and it is assumed that they are computer literate. Using standard software ensuring an easy-to-access survey site and a easy to navigate survey site. This should assure that even less computer literate virtual community members will not be inhibited from participating at the survey. For the same reason, it is assumed that it is unlikely that respondents will answer questions differently than they would do if they were surveyed in a traditional manner.
Online questionnaires have become commonly accepted for collecting data on the Internet. Areas as varied as interpersonal (Tidwell & Walther, 2002; Wright, 2004), group (Hobman et al., 2002), organizational (Ahuja & Carley, 1998), and mass communication (Flanagin and Metzger, 2001) have been studied using online surveys. Compared with offline questionnaires they bear several advantages, such as low cost (Baines et al., 2005; Schonlau et al., 2003; Taylor, 2000), high speed of delivery (Baines et al., 2005; Sills and Song, 2002)) and honest responses (Baines et al., 2008) ease of data cleaning (Sills and Song, 2002). However, online surveys are also related with sources of bias, such as survey item non-response, non-response and sample selection bias (Burkey and Kuechler, 2003).
By following Dixon (1983) it can be controlled for item non-response bias. The sample will need to be split in two sub-samples for each variable with missing data. The first sample will contain all responses and the sub-sample will be built with the responses containing missing data. This allows controlling for significant differences in the means of the data of the two samples (Little, 1988).
Previous surveys have found that late responders answer differently than early responders, and that the differences may be due to the diverse levels of interest in the subject matter. Thus it further needs to be controlled for non-response bias by comparing the answers of early and late responders (Armstrong and Overton, 1977).
The over-sampling of highly active community member is another important issue. Because the questionnaire will be published on the community site, members who frequently visit the community site will have a higher exposure to the survey then less frequent visitors. Armstrong and Overton (1977) propose a time-trend extrapolation to evaluate the results of quick respondents to slow respondents by comparing the two groups’ demographic variables, socio-economic variables, Internet proficiency and the like. If both groups do not significantly differ in those variables, it can be concluded that the general characteristics of the sample is not corrupted.
The survey will collect dependent and independent variables with a single instrument. This poses the threat of common method bias (Podsakoff et al., 2003). Podsakoff and Organ (1986) propose an analysis of common method variances to dispel this concern.
1.5 Main survey
The virtual community also serves as the research site for the quantitative study. On this section we focus on the online survey that will be developed for the study. The goal of the study is empirically test the conceptual framework proposed in Chapter 3 as well as the measurement of VCI developed in the qualitative study.
1.5.1 Method
Survey will be announced on the virtual community homepage containing a direct link to the online survey site. If possible survey link will also be sent to population via e-newsletter or email. While some studies assert that too many rounds (in this case the direct link on the community homepage and an email) might offend the virtual community members (Lakhani and von Hippel, 2002) other studies claim the opposite stressing that it will increase the response rate (Cook et al., 2000).
In order to better understand community members and their membership characteristics, the following types of questions should also be included into the survey:
§ Why they became members
§ Membership duration
§ How often they visit the community and duration of the visit
§ Which days and times they visit the community
§ What kind of activities they engage in when visiting the community
§ Have they developed social ties to other community members
§ To what extent they are dedicated to the topic of the community
Further, demographic and socioeconomic data will be collected as well as Internet profiles and a member’s orientation toward others. Data about the orientation toward other community members are important for considerations about interpersonal communication and computer-mediated communications. Member’s Internet proficiency will be captured with questions about their Internet profiles such as, how long have they been using the Internet, for how many hours, etc.
1.5.2 Item generation
To remain consistent with previous research, measures from previous studies (where possible) will be adopted (see table X). All measures will be subjected to confirmatory factor analysis to assess their psychometric properties and uni- dimensionality.
1.5.3 Target population and Sampling
The population of the survey will be the virtual community members. As the number of active members is often considerably lower than the number of registered community members, the log-in and log-off behaviour of community members will have to be tracked during the whole survey period.
The sample is a volunteer convenient sample. The online survey will be announced through a direct link on the virtual community homepage. The purpose of the survey will be explained and confidentiality and anonymity will be guaranteed. To ensure that members who access the virtual community website less frequently will also have a chance of being in the sample, the survey will be online for at least 4 weeks.
In a second wave the link to the online survey will be sent by email to all virtual community members. This will increase the possibility that virtual community members who have not logged in during the 4 weeks the survey will be announced on the community website; still have a chance to be part of the sample.
1.6 Data Analysis techniques
Accordingly, the use of survey questionnaire is advantageous or beneficial for a researcher since the researcher can already evaluate the suitability of the data as it is directly provided by pertinent respondents. Hence, the analysis of the gathered or collated data is very important to achieve the objective of the study. In this research, the research adapts a data analysis process to measure and analyse the collected data. The first state is the assessment of the overall suitability of the data to research objectives and questions. In this stage, the researcher pays particular attentions to the measurement of the validity and coverage. The second stage is the evaluation of the suitability of the date for analysis needed to answer and meet the objective of the study. In t his stage, the researcher ensures of different aspects of the research methodology: Unidimensionality, different types of validity, reliability, fit indices, structure equation modelling. In addition, the researcher will try to be cautious so as not to commit measurement bias which can happen because of the deliberate distortion of data or changes in the manner data are being collected. And lastly, the researcher judged how the data gathered through survey questionnaire and through secondary documents will be incorporated.
In this research, the researcher will initially submit a sample of the set of survey questionnaires and upon approval, the survey will be conducted to seven respondents After the respondents provide the answer, the researcher will ask any suggestions and necessary corrections from the respondents to be use in the improvement and validity of the instrument used.
After which, the research will again examine the content of the interview questions to determine the reliability of the instrument used. Herein, the researcher will not include irrelevant questions and changed words which were deemed difficulty by the initial respondents into much simpler terms. As mentioned, the following will be given emphasis to ensure that the instrument used adhere to the questions and objectives of this study.
- UNIDIMENSIONALITY
One important aspect to consider in knowing whether the data gathered adheres to the objective of the study is the measurement of the unidimensionality. Unidimensionality is used when the researcher tries to measure a construct using a multiple indicator variables. In this regard, the researcher will demonstrate that the items used in the instrument measure the same thing. There are various approach to test unidimensionality. These include Cronbach’s alpha, factor analysis, confirmatory factor analysis in structural equation modelling, Guttmann scaling and other forms of scaling. Accordingly, Cronbach’s alpha is the most commonly used measurement testing which is a measure of the inter-correlation items, In this test, if alpha is greater than or equal to 0.6 then the items are noted as unidimensional and may be integrated in a scale or index. Some researchers utilise the more stringent cutoff of 0.7. This measurement can be found in SPSS 13.
The next measurement is the factor analysis approach which is conducted on all the indicators for all the constructs in the research. Indicators in these types should have higher factor loadings on their own constructs than the other constructs. Some researcher also considers that loadings be higher than those absolute cut-off like 0.3 while some also considers that indicators cross load on factors not their own.
The next one is the confirmatory factor analysis in structural equation modeling. The first step of structural equation modeling is known as confirmatory factor analysis in which the measurement approach is evaluated separately forms the structural model. In this approach, if goodness of fit measures for the measurement approach is acceptable, the author concludes that the indicators sufficiently measure the intended constructs. In addition, it is also plausible to evaluate the unidimensionality of a concept through the comparison of models in structural equation modelling. This can be conducted through AMOS.
The last is the Guttman scaling in which Mokken, proximity and other types of scales, indicator variables are measured to see if they form a specific relationship to each other, such that the researcher is justified in putting the items in the equivalent scale. Herein, the relationship and the meanings of unidimensionality, differ by scale form. It can be noted that the Likert scale do not normally include testing for unidimensionality.
- Testing validity and reliability
The context of reliability and validity has now been defined not only to suite quantitative research but also qualitative. Accordingly these are two contexts that should be given emphasis. The reliability is referred to as the degree to which a test is stable and consistent in measuring what is intended to measure. In this regard, a test is reliable if it is consisted within the research objective; To understand the basic of reliability, consider a bathroom scale which give a drastically different readings every time a researcher stepped on it regardless if whether one has gained or lose weight. If such a scale existed, the research would be considered not reliable.
On the other hand, the test of validity referred to the degree to which the test actually measures what it claims to measure. The test of validity is also the extent to which inferences, conclusions and decisions made on the aspect of test scores are meaningful and appropriate. Accordingly, the test validity is requisite to test reliability. If a test is invalid, then reliability is moot. On one hand, if the test is invalid, there is no point in discussing its reliability since test validity is needed before measuring reliability.
- FIT INDICES
The major objective of fit indices is to detect model misspecification rather than the adequacy of modelling the null hypothesis. Herein, those who use structural models and factor analyse need indices to aids in detecting well fitting from those badly fitting models. The evaluation of these indices, but have been related to the sensitivity of such fit indices to sample size or to the method of estimation.
Credit:ivythesis.typepad.com
0 comments:
Post a Comment