Author: Áine M. Humble

Working with Data (continued)

Housekeeping
First, some brief points about data management/housekeeping. In my folder/file management system, I had folders for: (a) the initial grant application and any related financial documents; (b) the ethics application, which includes copies of informed consent forms, interview guides, and so on; (c) promotional material (newspaper ads, posters to be distributed at coffee shops, etc.); (d) the actual analysis (this can include quantitative data, if relevant - for this study, I also had an SPSS file with demographic information about the sample); (e) publications; (f) conference presentations; and (g) resources (e.g., downloaded research articles in pdf format). For this study, I kept note of various data management and analysis issues in one file, which was kept separate from the folders. However, since then, I have used MAXQDA's logbook for my audit trail (see Rodgers & Cowles, 1993, for a discussion of audit trails) for other projects and will continue to do so. Transcripts were typed out in a word processor (I use Microsoft Word) and changed to rich text format (with the file extension ".rtf"), which is a requirement before being imported into MAXQDA (once imported, the files are listed in the document system window).

With regard to backing up data prior to coding, two back-up copies of each transcript were made. First, each transcript was printed out and stored in a locked cabinet to which only I had access (microcassettes of the interviews were also kept here). Second, back-up electronic copies of the transcripts were stored on a CD in a separate location. I kept a back-up electronic copy of my MAXQDA file as well, but I admit that I did not save multiple versions of the analysis to show how it was changing over time. This has changed in more recent projects (I use MAXQDA's back up file option); on any day when new changes are made to the analysis, a new version is saved with that date (e.g., "Analysis- remarried- Jan 04-08.mx3"). Older versions are placed into a folder titled "previous analyses" and organized chronologically (e.g., a new folder for each week and/or month). Saving as different files rather than saving the same file over and over again serves two purposes: (a) it provides a history of the analysis, and (b) it also gives me confidence that I have a previous file to revert to in case the latest version somehow becomes corrupted.

Coding
To begin my analysis, I used the inductive process of topic coding (Richards & Morse, 2007) or open coding (Strauss & Corbin, 1990), staying as close to the data as possible, which is recommended by both Charmaz (2006) and LaRossa (2005) at the start of an analysis. This part of the analysis was carried out in MAXQDA. I should point out that I was already familiar with this program prior to beginning this study, and knew it was an appropriate QDA program (also referred to as a CAQDAS - Computer Assisted Qualitative Data Analysis System) to use for this particular analysis. I learned how to use this program on my own as a graduate student and I found its interface (consisting of four main windows called document system, text browser, code system, and retrieved segments, any of which can be viewed at the same time) and other aspects of the program very intuitive to use. I have described this program elsewhere (Humble, in press); MAXQDA also has a very helpful website that includes on-line tutorials and screenshots of the interface and other options.

In MAXQDA's text browser window, text segments from each transcript were coded for topics such as the type of work that was done, feelings experienced during the planning, and partner expectations (e.g., around involvement or level of involvement). As codes were created, they were listed in the code system window. Although I coded for many topics, my main interest was in the division of labour - who managed wedding work, who carried out tasks, and what kind of tasks were managed or accomplished by brides or grooms. I focused primarily on three work categories reported by the respondents: (a) work that women did on their own, (b) work that men did on their own, and (c) work that was done together. I also coded for when other people helped out. For example, when one groom reported that his wife had checked out ceremony and reception venues with her mother rather than with him (the text segment: "I didn't have a whole lot of time to go do site visits and that sort of thing, but we discussed a lot of things at home, and she actually went with her mother to most of the places"), it was coded as "women's work\visiting venues" and "received help from others\mother or mother-in-law". Throughout the process of topic coding, I used a constant comparison process (Strauss & Corbin, 1990), continually checking to see if the segment I had just coded fit with the other text segments assigned to that code. This procedure was facilitated by MAXQDA's retrieved segments window, which can show all of the coded text segments for a particular code once that code is "activated" for such analysis. Constantly comparing coded segments and codes helped ensure coding consistency within the codes. When needed, I renamed codes, deleted coded segments, moved coded segments, merged codes together, and so on (this was easily accomplished through various options in the code system window). It is important to see codes as flexible at this stage of the process.

Couple categorization
As I coded the transcripts for each couple, I developed a sense of what their wedding preparation was like, and immediately after coding each bride and groom from the same couple I described their wedding preparation (of both their most recent and first weddings) in a paragraph or two in a separate document (in Microsoft Word). In this document, I noted my interpretation of their gender construction in their descriptions of their most recent weddings. I then compared couples with each other in terms of their gender construction, working a bit more deductively at this stage, as per directed content analysis, whose goal is to "validate or extend conceptually a theoretical framework or theory" (Hsieh & Shannon, 2005, p. 1281).

Based on the topic coding and descriptions, I categorized each couple into a previously developed conceptualization (Humble et al., 2008 [PDF- 109 KB]) consisting of three wedding planning categories: (a) traditional (the groom has no involvement or very little involvement in wedding planning), (b) transitional (the bride and groom feel that there is equal involvement and the groom is more involved compared to traditional grooms, but details from the interviews indicate that the bride-to-be still does most of the work), and (c) egalitarian (the bride and groom plan and carry out wedding planning together). I also looked at couples within each category, seeking to describe the full range of gender construction - for example, seeking to determine who was the most egalitarian or least egalitarian of the couples who were all categorized as egalitarian. To carry out this analysis, I printed out each couple's sheet and placed all 14 sheets on the floor of my living room. This allowed me to move couples (their sheets) from categories or within categories as needed. Indeed, some placements did change as I thought carefully about each couple and compared them to the other couples. Thus, I actually continued with a constant comparative process here, but with the comparison now between conceptual categories rather than codes. During this stage of the analysis, I categorized six couples as traditional, three as transitional, and five as egalitarian.

Matrix analysis
What about individuals' first weddings (for those who remarried, which were 26 of the 28 participants)? As indicated earlier, the impetus for this study was to see if individuals did do anything differently "the second time 'round." I mentioned earlier that I did not have as good data for these weddings as I had for individuals' most recent weddings because some individuals were trying to remember details from events that occurred 20 or 30 years ago. However, I was able to get a general sense of participants' first weddings and the planning of those events. With that information, I developed a qualitative matrix (Richards, 2005) comparing individuals' first weddings with their second weddings. Matrix analysis involves "the crossing of two or more main dimensions. . . to see how they interact" (Miles & Huberman, 1994, p. 239, as cited in Averill, 2002). I created this matrix using a Microsoft Word table, cross referencing how involved (little to no involvement or a high level of involvement) individuals were relative to their partners for their most recent weddings with how involved they were in their first weddings (gleaned from reviewing their answers to the question about their first weddings), again, relative to their first spouses. This matrix is shown in Humble (2009) (PDF- 139 KB). It was very easy to create the matrix, and it was an effective way of simplifying the data to determine whether or not there were any patterns, which there were (although second or third weddings tended to be smaller, less formal events compared to first weddings, the results indicated that patterns were typically reproduced from first weddings to second weddings. That is, if a woman was largely responsible for planning and carrying her first wedding, she was still typically responsible for planning and carrying out her second wedding, even if it was smaller or less formal). Matrix analysis was new to me, however, and after I presented my results at a conference (PDF- 701 KB) in 2008, I decided that I could have presented the matrix results in a different-and what I thought a much clearer-way. This alternative matrix is presented on my website. I humbly submit that I am still learning.

Back to Project Home Page