The study attempted to use the 'editing analysis style' proposed by Crabtree and Miller (1992) during data collection. According to Polit and Hungler (1999, p.574) the following is true of this approach to data analysis:
"The researcher using the editing style acts as an interpreter who reads through the data in search of meaningful segments and units. Once these segments are identified and reviewed, the interpreter develops a categorization scheme and corresponding codes that can be used to sort and organize the data The researcher then searches for patterns and structure that connect the thematic categories".
Given the nature of the study and the proposed methodology, it appeared to be the appropriate choice as it fitted snugly within the areas of phenomenology and grounded theory (polit and Hungler, 1999). This style of data analysis allowed the
researcher to feedback to participants during focus group interviews and semi- structured interviews, thereby ensuring verification of the information provided and the way in which it was interpreted. Examples of this can be seen in the excerpts below of data from an interview with an antenatal AMC:
Example 1:
Respondent: "Ya, they hurt you and when you say you know, 'it's sore '. 'Oh, relax, relax, relax '. And most ofthe time they tell the patient, like anybody,
'oh when you were having sex it was nice, now you are complaining '. "
Interviewer: "So when you complain about something that is done to you, then the response is, 'you didn't complain when you were having sex'? "
Respondent: "Yes, yes. "
(Antenatal AMC) Example 2:
Respondent: "The first timeI came there to make an appointment to come to the antenatal clinic and thingy, they were like ... that nurseI spoke to, she was like, 'oh, you so young and blah, blah, blah,andyou coming so late for clinicing and everything '. "
Interviewer: "You sound like they made you feel uncomfortable? "
Respondent: "Ya. "
(postnatal AMC) The approach to data collection also allowed for the researcher to constantly compare information obtained, and delimit dataina bid to enhance the emerging of concepts towards thefinalconstructs.Italso helped the researcher in theoretical sampling, by giving more direction in what to ask next and of whom to ask.
The following schematic representation explains the 'editing analysis style':
Figure 1: Editing Analysis Style
Interpreter
Report ...
1lII---~-~-:f
+ - - - ."/ ,," Ident1
units~ I .
: , , Deve op rtegones
... ... Interpretlvely Determine Connections
(Crabtree and Miller, 1992, pI8) In terms of this study the schematic representation was as follows:
Figure 2: Adapted Editing Analysis Style
Researcher
RepoAl--- \-'lficipants (AMC, HCp
1
& HCP) +----....,~~~ ~
Identify units relatedtfl~perceptions / improvements of care, ie. whole statement
~~~ ~
1/ Emergence of categories and themes
'... ~ ~ l
Feedback to participants , ,
to verify ...
Determine relationships that exist between categories / themes
Done by feedback to participants Verify relationships observed
(Adapted from Chetty, 2000, p.54)
As can be seen from the above diagrammatic representation, during data collection from participants, the researcher began simultaneous analysis, by carrying out a
process of datareductio~attempting to attach "meaning to elements" in the data that was being collected (Burns and Grove, 2001, p. 596). Units of information relating to perceptions of adolescent maternity care as it stands, and what the care should ideally be, were identified by the researcher in an attempt to bring to the fore emerging categories and their properties. Once this was done, the researcher focused attention on trying to determine relationships that may exist between the core categories and sub-categories. To ensure that the researcher was clear about the information that participants were providing, this information was fed back to participants during the focus group interviews and semi-structured interviews to verify correctness and allow for constant comparison. An example of this is found in the following excerpts from the data. In the first statement an AMC respondent alluded to the age of the HCp invoking a different response from her.
" No some ofthem that's old I just ignore them. The young ones make me cross because they know what they do. "
(Antenatal AMC)
The researcher then followed this trend in thought to ascertain how other members of this focus group interview felt about the age of the HCp. The researcher then decided to see if other stakeholder groups verified this information.
Interviewer: "Is there a specific age category you think the person should be?"
Respondent 3: "It would help, but
if
you are lawwledgeable you shouldn't have a problem being 20 or 30. "Respondent 2: "But I suppose it would be easier for them to talk to a person who is more or less their age,
if
possible. "(postnatal HCp) Areas of consensus amongst stakeholder data were used to construct the requirements and components of adolescent-friendly maternity services.
In keeping with the directions from Glaser (1998), data analysis was done manually, allowing the researcher to immerse herself in the data by way of memoing and developing categories with its related properties, and drawing thematic connections between categories. Once interviews were transcribed into text, the researcher began the process of reading through the data, coding (substantive and theoretical),
analyzing, making memos, and arriving at categories. With further surveillance and constant comparison, categories which appeared to be linked to more than one category became visible. Hence, core categories were arrived at, linked to sub- categories and their properties.
A category is used to succinctly catch fundamental patterns that appear in the data and is at a higher level than a property. Since a property is defined as "a concept about a category", every category will have related properties (Glaser, 1998, p. 135).
Categories can alsobe used as properties of other categories. Within Glaser's grounded theory, categories are arranged according to their conceptual levels. Core categories are the highest level, followed by sub-core and categories towards theoretical comprehensiveness. "The core category relates to most other categories and their properties, since through these relations the core category accounts for most of the ongoing behaviour in the substantive area being researched" (Glaser, 1998, p.
135). For the purpose of this study, a slight deviation from what Glaser purports was used, with the findings being arranged into core categories, sub-categories and properties.
Thus emerged the latent structure of the model for action (Glaser, 1998), which by constant comparison of data, was grounded into a modifiable model for action. This
process took place using a four level conceptual perspective analysis. Firstly, the data was scrutinized and coded. Secondly, the data was conceptualized into categories and their related properties, which is presented in Chapter Four. At the third level of this process, the categories were arranged into constructs based on their thematic patterns and links, as demonstrated in Chapter Five. Finally, these constructs were formalized into a model for action by the use of literature from the substantive area under study and presented as a discussion in Chapter Five (Glaser, 1998).