2. CHAPTER 2: Literature Review and Theoretical Framework 13
3.5. Validity and Trustworthiness
Although some researchers caution that the concept of validity is more applicable in quantitative studies than it is to qualitative studies, (Golafshani, 2003; Stenbacka, 2001;
Lincoln &Guba, 1985), it is important to use appropriate criteria to ensure that a qualitative study has rigour and depth. Lincoln &Guba (1985) put forward alternative criteria that could be used in qualitative research, in the place of the traditional criteria that are more applicable to quantitative studies. Drawing from Lincoln &Guba (1985), Trochim (2006) outlined the criteria for qualitative research, in Table 3.1. as follows:
Table 3.1. Criteria for Qualitative Research
Traditional Criteria (Quantitative)
Alternative Criteria (Qualitative)
Application in this Study
internal validity credibility Phenomena are described and understood from the participants’ perspective and hence participants judge the credibility of results. Testing of credibility of findings and interpretations is done through using various sources, i.e. triangulation.
Participants were promised a copy of the final draft of the study.
external validity (generalizability)
transferability
Transferring results to other settings becomes the responsibility of the one who transfers them. The use of a thick description about each context makes it easier for the reader to judge whether the results of one context may be applied to the other.
47 reliability dependability An account for the ever-changing context
within which the research occurs is given by the researcher. Consistency is visible on the produced findings.
objectivity confirmability Examining data collection and analysis procedures to make judgement about potential bias or distortion. Data collection procedures and analysis has been certified not by the researcher but by the data itself.
It is important for the researcher to provide a trail showing the basis for layers of increasing abstraction in the analysis process.
In a qualitative study, the researcher is the key instrument of research rather than a tool. The researcher is also part of the world that is being researched. As argued by Cohen et al., (2007), “the involvement and in-depth responses of individuals secure a sufficient level of validity and reliability” (Cohen et al., 2007, p.134). Cresswell& Miller (2000), further claim that it is the researcher who determines whether the “data are saturated to establish good themes or categories, and how the analysis of the data evolves into a persuasive narrative”
(Cresswell& Miller, 2000, p.125). I am a peer to the participants, and so they did not feel pressurised to present themselves as highly successful and professional individuals if they were not.
I decided to conduct the interviews personally, to minimise perceptions of power that may have been held by the participants. For example, participants might have responded differently if they were interviewed by a Subject Advisor and may have felt that failure to give an expected answer would lead to some form of victimisation. This situation created a free and relaxed atmosphere for the interviewees, who were interviewed at a venue selected by them, at a time convenient to them and with no external interference of either an audience, a cameraman or any other distraction.
Cohen et al., (2007) suggest that crucial information may be lost during data analysis. For example, during the transcription phase of interviews, some important data may be lost or distorted. They warn that reliance on one method of data collection may distort the
48 researcher‟s picture of reality. To reduce the risk mentioned above, data triangulation was used in this study. As defined by Cohen et al., (2007), triangulation is “ the use of two or more methods of data collection in the study of some aspect of human behaviour” (Cohen et al., 2007 p.141). Triangulation is further defined as “a validity procedure where researchers search for convergence among multiple and different sources information to form themes or categories in a study” (Cresswell& Miller, 2000, p.126). Data from questionnaires, interviews as well as records of the participants‟ performance was triangulated.
When data that was collected through different methods of collection happen to yield the same results, a greater degree of confidence about the findings is achieved. Just as triangulation builds more confidence in the findings, it also serves as a tool by which the researcher checks the data collected. Triangulation also works to disconfirm emerging trends which did not hold across the data sources. However, Cresswell& Miller (2000) warn that the
“search for disconfirming evidence is a difficult process because researchers have the proclivity to find confirming rather than disconfirming evidence” (Cresswell& Miller, 2000, p.127). There also exists the risk that disconfirming evidence could outweigh confirming evidence.
Participants completed the questionnaires in the absence of the researcher. This, according to Cohen et al., (2007), is helpful because “it enables the respondents to complete the questionnaire in private, to devote as much time as they wish to its completion... and to avoid the potential threat or pressure to participate caused by the researcher‟s presence” (Cohen et al., 2007 p.344). Cohen et al., (2007) further assert that the absence of the researcher when the questionnaire is completed also makes it easier for participants to reveal sensitive matters more than they would have done in the presence of the researcher.
Lastly and importantly, I described the setting of the study in detail, where participants were fully described in rich detail. This was especially true in the case of interviewees, where the biography of each participant was given, to highlight why such a participant was included in the interview. This provided a thick description, whose purpose according to Cresswell&
Miller (2000), is “[to create] verisimilitude, statements that produce for the readers, the feeling that they have experienced or could have experienced” (Cresswell& Miller, 2000, p.129). This procedure of giving a thick description, “enables readers to make decisions about the applicability of the findings to other settings or similar contexts (Cresswell& Miller, 2000, p.129). Guba (1981) supports the notion that a thick description may “demonstrate an
49 essential similarity between two contexts”, thus making it “reasonable to suppose that tentative findings of Context A are also likely to hold in Context B” (Guba, 1981, p.81). The rich details given in this study helped to produce such validity.