Using ATLAS.ti for Coding Ethnographic and Policy Data
Author: Joyce Weil
As part of my work on my book, The New Neighborhood Senior Center: Redefining Social and Service Roles for the Baby Boom Generation, to be released by Rutgers University Press this November, I used ATLAS.ti to organize and code themes from in-depth interview narratives with older persons attending a senior center. My data, or primary documents (PDs) as they are called in ATLAS.ti, also included fieldnotes, my participant-observations and other written and visual materials (such as newsletters, video clips, and newspapers) for various periods of time from 2008 until mid-2013. I also used ATLAS.ti to code interview narrative from heads of non-profit organizations and governmental officials in the field of aging and to organize themes in legislative and social policies concerning aging and senior centers for the last two chapters of my book.
As background on the project, my book explores the ways social structure impacts the lives of older persons at a local senior center at the micro level. And, at the macro level of analysis, I trace the role of political, social, and economic institutions, organizations, and neighborhood processes involved in the decision to close, or shutter, a senior center. Then, I expand the discussion to include issues related to the trend of re-branding senior centers to attract Baby Boomers. The entire modernization of centers is placed in a national context to incorporate recent changes in legislation such as the on-going re-authorization of the Older Americans Act and the current Affordable Care Act. I also discuss the feasibility of using existing federal programs as a way to fill existing gaps left from closed senior centers.
My Data Coding Approach
Here, I will describe the way I used Computer Assisted Qualitative Data Analysis Software (CAQDAS), ATLAS.ti.7, to do the data analysis for a portion of this project. The examples I will use here are the accounts of daily activities at a senior center, a community setting where older people may socialize, volunteer, and/or receive services.
I followed Virginia Braun and Victoria Clarke’s (2006) thematic data analysis process – from being really familiar with the data to offering the reader deep, rich description of the data with large portions of participants’ direct quotations. To begin, after full transcription of each interview into a Word file, each Word document was entered as a primary document in the ATLAS.ti program as part of one Hermeneutic Unit (HU). Fieldnotes, and all other printed material, were scanned and entered into ATLAS.ti as well as part of this same HU.
With my combined in vivo and open-coding approach, there were no preconceived codes or categories for the interview data. When working with ATLAS.ti, I felt as if I was writing in the margins of a page, which enabled me to have freedom to code my interview data as I read through it on screen. For my in vivo coding, I began to use elders’ exact words and phrases as quotations to create my codes. For my open-coding, I interpreted the meaning of text in the narratives to generate additional concepts (or codes) from the data. Using both coding strategies, I could either create a code from the direct quote of a participant or link a code to concepts.
Looking at the frequencies of codes led to the discovery of dominant (as well as unexpected or serendipitous) themes and patterns in the interview text. ATLAS.ti also allowed me to create a visual word cloud of codes and code families (groupings of codes together) along with code-mapping displays. Using the Word Cloud feature from the Analysis and Word Cruncher menu options, I produced a visual representation of the codes in the data. The more frequent codes appear in larger font in the cloud. This view showing all the codes helped me clearly see and uncover the general patterns in the data. (See Figure 1 below for a portion of my word cloud of codes in my project.) Note that the numbers in parentheses after each code refer to the groundedness (number of direct quotations linked to a code) and density (semantic link of the code of interest to other codes). Also note how I changed the color of a code of my particular interest at a point in time in the analysis. Here the active aging code appears in purple font.
An Example of Coding Themes and Using Code Families
In this article, I will use an example of the way one set of themes (for how center members use the senior center on a daily basis) was identified using ATLAS.ti. While reading through each account, I noted a code for each and every way a participant talked about activities that occur at the senior center. 58 activity codes could be compared and combined. Some of these 58 codes are shown in Figure 2 below. As I coded the interview text, I also kept track of my coding definitions and decisions and noted any differences in the narratives by using ATLAS.ti’s memo feature. These memos aided me in what Susanne Friese (2014) calls recursive coding or rethinking and examining the relationship between codes/concepts in a continual process throughout the project.
In addition to listing code counts or frequencies, the Code Manager feature let me combine individual codes via commonality of theme into code groups, or what ATLAS.ti calls Code Families. This process is analogous to analytical or axial codes being moved towards selective coding. So, some of the 58 subcodes for kinds of individual activities (partially shown in Figure 2) at the center were grouped together into a larger, positive “Activities for All Center-Goers” family. Another emergent “Negative Changes in the Center” family was created and included 15 codes (see Figure 3 that includes a partial code list). Again, here I used purple font to highlight my code of interest at that point in the process: gossip.
Using ATLAS.ti helped me explore my narrative data. I used the software to become familiar with the narrative text in the documents and create/review the codes and their relationships to one another – as seen in the various negative activities codes within the “negative changes at the center” code family. Word clouds offered me a way to initially, literally visualize the occurrence of codes/concepts in my work. Although not the main focus here, the memos feature let me define codes and create my own audit trail of coding and track general observations as I worked with the data. For me, ATLAS.ti was an integral organizational tool in my on-going evolving data-analysis process.
Braun, V. & Clarke, V. (2006). Using Thematic Analysis in Psychology. Qualitative Research in Psychology, 3 (2): 77-101.
Friese, S. (2014). Qualitative data analysis with ATLAS.ti. London: SAGE.
Weil, J. (Nov. 2014). The New Neighborhood Senior Center: Redefining Social and Service Roles for the Baby Boom Generation. NJ: Rutgers University Press.
About the Author
Joyce Weil is assistant professor of gerontology at the University of Northern Colorado with a Ph.D. in sociology from Fordham University and an MPH from Columbia University. Her research focuses on aging-in-place and social inequalities across the lifecourse. Her articles appear in the Journal of Aging, Humanities and the Arts, the Journal of Loss and Trauma, Social Forces, the International Journal of Aging in Society, and Research on Aging. She is a co-editor of Race and the Lifecourse: Readings from the Intersection of Race, Ethnicity, and Age and sole author of The New Neighborhood Senior Center: Redefining Social and Service Roles for the Baby Boom Generation (Rutgers University Press, Nov. 2014).