4. CAPITULO IV - DESARROLLO
4.7. DETERMINACIÓN DEL FACTOR DE PROPOCIONALIDAD Y CRÉDITO
4.2.9.1 Qualitative data
We utilised a qualitative content analysis approach. Krippendorf (1980) (cited in Elo & Kyngas, 2008, p 108) described this approach as a research method useful in making valid and replicable conclusions from data as related to their contexts. This method serves to represent facts, to gain knowledge and new insights all leading to a practical guide for action.
Thus it appeared to be extremely well suited as a method for analyzing the sensitive and multifaceted psychosexual/vaginal changes experienced by women post PRT for gynaecological and anorectal cancer. Rosenthal & Rosnow (1991) defined content analysis as a method of categorizing information using frequency of occurrences. According to Stemler (2001, p1), content analysis is a systematic replicable method in which many words of the text can be compressed into fewer categories using explicit rules and coding. The approach is considered to be at the juncture of qualitative and quantitative techniques, allowing for quantitative examination of ostensibly qualitative data to access both manifest and latent meaning from the text (Kondracki et al., 2002; Elo & Kyngas, 2008). Quantitatively, message elements are counted to identify foreground themes and emphasis in certain areas. Qualitatively, the manifest and latent/inferred meanings of the messages under inspection can be elaborated on, leading to the development of hypotheses or theories based on knowledge and evidence drawn from the study (Kondracki et al., 2002).
Furthermore, content analysis can be applied both inductively and deductively; these options are not mutually exclusive but rather are considered useful when combined (Berg, 2004). In other words, in content analysis the coding scheme can be decided a priori (White and Marsh, 2006; Stemler, 2001) or once emergent themes are evident. Thus, as noted above, a directed (deductive) content analysis can be undertaken, as with the current study, when incomplete information exists on a phenomenon from prior research or requires elaboration (Hsieh & Shannon, 2005) because it affords the focusing of questions. With this method, a more structured process guides the coding (or categorising) scheme, which helps to identify key
concepts and/or variables to be used as initial coding categories (Hseih & Shannon, 2005). This is of a higher standard than traditional content analysis where codes are derived after data collection in an inductive approach. Data collected through interviews, as in the current study, used open ended questions which were followed by targeted questions on predetermined categories. However, as was the case in the current study, text can also later be given a new code/category based on analysis of inductive/latent/emergent meanings over and above the initial deductive/a priori codes/categories.
The summing of responses, that is the quantitative aspect of this method, can also extend further than the manifest meaning of the data (Hseih & Shannon, 2005) and allow for interpretation of the content. Here, as noted above, underlying meaning is discovered, such that it is inferred from the data rather than directly addressed at the onset. Examples from the current study of non a priori format, content and utility of booklet categories that emerged post data collection were: acceptability of the booklet rehabilitation strategies, booklet validating sexuality; experiences of care in the healthcare setting. However, frequencies of response occurrences held equal importance in this content analysis, making it a versatile and relevant research method for assessing the feasibility of a psychosexual/vaginal changes rehabilitation booklet to women undergoing PRT for cancers in the pelvic region. Thus, both inductive and deductive content analyses were utilized: the former where there are no prior studies or information is disjointed on an experience and the latter used for testing a previous theory in a different type of setting or for comparing different groupings at different time periods (Elo & Kyngas, 2008).
Furthermore, in accordance with pilot aim ii) the quantification of responses was highly relevant, as noted above, to inform clinicians in supporting and participating in the later RCT. In health sciences, content analysis is accepted and widely used as a standard qualitative research method (Hsieh & Shannon, 2005). It is particularly well- suited to studies involving the practice and education of nurses and other helping professionals because of its focus on communication (Downe-Wamboldt, 1992). Thus content analysis appears well-suited for clinical practice and patient education in the healthcare setting, which was pertinent to the current study.
Based on existing theory and specialised/expert colleagues’ consensus regarding categories, the current study initially used a priori coding. Here, the coding is applied to the data with revisions made as required to ensure mutual exclusivity and exhaustiveness (Stemler, 2001; Weber, 1990). Each question from the telephone interviews and the study questionnaire was entered into a spreadsheet and frequencies of response occurrence were recorded. Responses that fell into broader emergent categories were later allocated to those new categories.
Other qualitative methods, such as thematic analysis, were considered but later excluded because they did not appear to meet the pilot study research questions/aims, taking account of their different foci (Harper, 2012, p. 85). For example, a thematic analysis – which aims to map out the range of concepts or summarize unstructured data in thematic categories – was excluded because unlike content analysis, it does not examine frequencies of responses. This was needed to meet the study objectives to determine how many participants liked the booklet format and content and found it helpful (utility) and thus its feasibility for women post PRT treatment. Other qualitative methodologies excluded were Grounded Theory, Phenomenology and Narrative approach because again content analysis appeared to relate to the research aims best.
4.2.9.2 Trustworthiness
A major advantage of content analysis is the fact that validity and reliability can be tested and maintained, since it is replicable and systematic (Potter & Levine- Donnerstein, 1999; Stemler, 2001). A content analysis categorises data according to a coding scheme which includes rules of data analysis that are logical and scientific. This systematic approach is central to trustworthiness (Hsieh & Shannon, 2005) and made content analysis an appealing, dependable and relevant research method for the present study.
However, results will not be correct if faulty definitions and non-mutually exclusive and exhaustive categories are used (Stemler, 2001). Thus, the following measures were put in place to ensure trustworthiness in the present study:
Reliability is aided by the demonstration of the data and the analysis, since the interpretations and findings can be easily reproduced by others examining the data and the process and procedure of the chosen examination (Elo & Kyngis, 2008). In the current study, bona fida citations (i.e., patient quotes) were liberally included, in accordance with Patton (1990) and Sandelowski (1993) s’ assertion that they increase the trustworthiness, reliability, and validity of the analysis by exemplifying how categories arose from the original data. Confidentiality/non-identification of the participants was observed (Ford & Reutter, 1990).
Numerous other strategies were also implemented in the current study to ensure rigour, validity and reliability of the qualitative approach, most notably an a priori coding scheme that guides coders, test-re-test procedures, and multiple coders. Emergent categories were subject to the same scrutiny and processes. Further, to ensure face, criterion and construct validity, and reliability, the coding scheme had definitions which were clear, instructions which were easy to follow, and clear examples. Doing this allows for methodical and reproducible analysis of the data and ensures inter-coder and intra-coder reliability (White and Marsh, 2006).
4.2.9.3 Quantitative data
The quantitative data on demographic and disease characteristics, from the standardised measures and regarding perceptions of the booklet, were entered into SPSS (IBM SPSS Statistics Premium, Version 19.0) and analysed using frequencies and descriptive statistics.