What have sampling and data collection got to do with good qualitative research? My current research project is a mixed phenomenological and meta-analysis of declining membership and participation in the church. Operating on the presumption that sampling and data collection are critical to a study (Gibbs, 2007). Like Gibbs (2007) I want to be guided by the research goal developing theoretical outcomes Gibbs (2007), covering intrinsic participant cognitions, and clearly explaining any limitations (Gibbs et al, 2007).
I have decided to reduce the scope of my study to the a case study approach with a Pastor and five Associate Ministers within a single church to which I happen to belong, in the Midwest.
I believe these five observers are in the best “position” to observe this phenomenon and its effects.
The information obtained in this first week has led me to the following conclusions concerning sampling and data collection. According to the work of Gibbs, Kealy, Willis, Green, Welch, & Daly (2007), sampling and data collection are intrinsically germane to generalizability (Gibbs et al, 2007). These authors, in agreement with other exceptional researchers, use designs like those of Daly, Willis, Small, Green, et al (2007) who also note that generalizable studies provide a comprehensive analysis of experience (Daly, et al, 2007). There is an imperative for the allowance of immersion to investigate context and population, along with practical constraints operating against sampling and data collection (Gibbs et al, 2007). Qualitative research begins with justification of the research problem with reference to the literature (Gibbs et al, 2007). Qualitative research then according to Willis, Daly, Kealy, Small et al (2007) provides theoretical framework to identify the theoretical concepts relevant to and employed in the study Willis, et al, 2007).
Data is then collected according to a sampling plan, as suggested by Green, Willis, Hughes, and Small, et al, (2007), thus the most acceptable evidence possible, through data analysis (Green, et al, 2007). The hierarchy of evidence model proposed by Gibbs, et al (2007), offers studies that differing evidences such as the single case study, the descriptive study, the conceptual study, the generalizable study and the interview study (Gibbs et al, 2007). Accordingly transcribed data from verbatim recordings is the most common method of data collection (Gibbs et al, 2007). In these instances individual case studies, are limited by small samples but, capable of provide more information on setting (Gibbs et al, 2007); and Descriptive studies, describe experiences or activities but do not describe their differences (Gibbs et al, 2007). Case and descriptive studies provide good information as long as their limitations are clearly acknowledged (Gibbs et al, 2007).
According to Suri (2011), informed decisions concerning sampling are necessary to improving the quality of research (Suri, 2011). Suri additionally points out that data may be retrieved through group discussion, personal journals, follow-up in-depth interviews and researcher field notes (Tuckett and Stewart 2004a, 2004b; Suri, 2011). According to Tuckett, et al 2011 and in agreement with Rubinstein (1994), no rules governing the numbers in sampling apply; however, experiential methods have been used for choosing samples from 1 to 100, with clustering. Some have suggested as few as 12-20 data sources, for the best variation, because no definite rules apply (Baum 2002). Suri notes that according to Patton (1990), some research relies on small samples aiming to study provide depth and thoroughness (Miles and Huberman 1994, Patton 1990). Purposeful sampling is seen as a means for developing rich data, derived non -randomly (Ezzy 2002, Mays and Pope 1995, Reed et al, 1996), Also, according to Lincoln and Cuba (1985) and Higginbotham et al (2001), the desired sample size may unfold, depending on previous studies, allowing the support of emerging theory (Baum 2002, Kuzel 1992, Miles and Huberman1994, Reed et al, 1996).
Another issue in data analysis is presented by Sandelowski (2011), when he suggests alternative interpretations of data do not conform to the parameters between methods (Sandelowski, 2011). Sandelowski suggests that taking a view of inquiry as dynamic and flexible rather than static and unchangeable might prevent researchers from succumbing to that follow (Sandelowski, 2011). Sandoelowski also notes that Alvesson and Skoldberg (2009) coined extreme terms such as grounded theory ‘‘dataism’’ (p. 283), the hermeneutic ‘‘narcissism’’, and critical theory “reductionism’’ (p. 269). Sandelowski further suggests that data analysis and presentation do not have to be considered as discrete independent operations (Sandelowski, 2011). Recognizing Spalding and Phillips (2007, p. 961), Sandelowski proposed that the use of vignettes will reveal the often concealed author’s vision which Phillips expects will produce doubt’ (p. 961), inevitably serving to enhance the validity of interpretations (Phillips, 2007, p. 961; (Sandelowski, 2011).
Sandelowski finally concludes that recognizing the need to account for problems associated with cognitive flexibility validating qualitative or quantitative inquiry Sandelowski, 2011). In addressing the issue of “presentation”, I found an article by Simundic (2012), concerning some “Practical recommendations for statistical analysis and data presentation”. The table below gives a suggestion for what should be included in any presentation of data. In working on the definition of “saturation” I was able to find the differentiation between the various qualitative methods. The following table is a representation of my findings based on the article by Walker (2012). I was impressed with the definitions provided by this author as he explained the different methods of determining saturation. I found the definitions of to be succinct and to the point, and very helpful in making a decision about which methods to use and when.