Volume 19, No. 2, Art. 23 – May 2018



The Lone Mother Resilience Project: A Qualitative Secondary Analysis

Elizabeth C. Watters, Sara J. Cumming & Lea Caragata

Abstract: Although qualitative secondary analyses are conducted across the social sciences, supra-assorted analyses that involve both the re-use of existing data and the collection of new, primary data are relatively uncommon. Additionally, discussions regarding qualitative secondary analysis have tended to ignore the re-use of researchers' own data (i.e., auto-data). Thus, with this article, we aim to contribute to this discussion by providing an example of a supra-assorted analysis in which we re-used data from one of our previous studies, Lone Mothers: Building Social Inclusion. This earlier, longitudinal study was conducted with 104 poor lone mothers across Canada. We supplemented this dataset with data from three focus groups and 20 semi-structured interviews engaging a total of 38 lone mothers. Both studies were informed by a feminist and social inclusion lens, and recruited a diverse sample of women in three cities across the country: Vancouver, British Columbia; Toronto, Ontario; and St. John's, Newfoundland. In addition, most of the lone mothers who participated in the secondary analysis had also been involved in the original study as interviewees and/or research assistants. We conclude the article by discussing the strengths and limitations of, and lessons learned from, the secondary study's design.

Key words: qualitative secondary analysis; secondary research; supra-assorted analysis; auto-data; qualitative research

Table of Contents

1. Introduction

2. Secondary Analysis

2.1 Types of secondary analysis

2.2 Risks and benefits of secondary analysis

3. Original Study

3.1 Background

3.2 Methodology

4. Secondary Study

4.1 Background

4.2 Methodology

5. Conclusion

Acknowledgments

Appendix: Illustration of Resiliency Profile Building Process Using Existing Dataset

Notes

References

Authors

Citation

 

1. Introduction

Secondary analysis of qualitative data can be described as "any research activity in which a researcher uses data for purposes not defined or predicted in the original study design" (YARDLEY, WATTS, PEARSON & RICHARDSON, 2014, p.102). Unlike the re-use of quantitative data (e.g., statistical datasets), secondary analysis is a less established and more contested methodology (HEATON, 2004). Much of the debate concerning secondary analysis pertains to the archiving of qualitative data (i.e., in a database), and its subsequent re-use by those who were uninvolved in the original study (for example, see CORTI, DAY & BACKHOUSE, 2000; HAMMERSLEY, 1997; IRWIN, 2013; MAUTHNER & PARRY, 2009; MAUTHNER, PARRY & BACKETT-MILBURN, 1998; THORNE, 1998). The issues associated with qualitative archiving have been discussed at length in the literature, and will not be revisited here. Instead, we focus on issues that are relevant to the re-use of researchers' own data (i.e., "auto-data")1) (HEATON, 2000, p.6). This is the approach we utilized in our secondary study, and is one that appears to be more common in social research (HEATON, 2004). Nonetheless, discussions regarding secondary analysis have largely ignored this approach (ibid.), which risks overlooking its potential (COLTART, HENWOOD & SHIRANI, 2013). With this article we aim to contribute to this discussion by providing an example of a secondary analysis in which we re-used our own data. Moreover, the type of secondary analysis we utilized, supra-assorted, is one that is less common as it involves the collection of new data (HEATON, 2004). Our study may also differ from other supra-analyses, as the new, primary research engaged participants from the original study. [1]

Introducing this discussion in Section 2, we first summarize the different types of secondary analysis, as well as the risks and benefits associated with the re-use of auto-data. Then, in Sections 3 and 4 we provide descriptions of the original and secondary studies, respectively; while in Section 5, we conclude the article with a discussion on the strengths and limitations of, and lessons learned from, the secondary study's research design. [2]

2. Secondary Analysis

2.1 Types of secondary analysis

As noted above, data utilized in a secondary analysis are pre-existing and may or may not be the researcher's own. Also, the data can be derived from different sources or methods. Most often, re-used data are derived from observations, focus groups, and interviews (HEATON, 2004). A dataset can then be analyzed in its entirety, or sorted to better fit the focus of the secondary analysis; for example, to focus on a particular subsample of participants, or certain topics/themes within the data (ibid.). Different types of data can also be separated within a dataset; for example, interview data from observational data (LONG-SUTEHALL, SQUE & ADDINGTON-HALL, 2010). Thus, one dataset can potentially be sorted in several different ways depending upon the focus of the secondary analysis. Further, secondary analyses can vary in other respects, thereby giving rise to different types of secondary analysis (HEATON, 2004; HINDS, VOGEL & CLARKE-STEFFEN, 1997; THORNE, 1998). HEATON (2004, 2008), whose work is commonly cited, developed a typology of secondary analysis based on a review of the health and social care literature. This typology is comprised of 5 categories, namely re-analysis, supplementary, supra, amplified, and assorted. We summarize these different types of secondary analysis below. [3]

Within this typology, the purpose of a "re-analysis" is to verify a study's findings by re-examining the raw data, or comparing the published findings with new, primary research. Also, when re-analyzing raw data, researchers may use an analytic method that differs from that of the original study. HEATON (2004) found that re-analyses are the least common type of secondary analysis, and are mainly conducted by researchers who were uninvolved in the original study. Unlike re-analysis, the purpose of most secondary studies is to conduct new research. This is most commonly achieved through the use of supplementary analyses, which "extend" the focus of an original study by exploring a theme or subset of the data in more detail. The theme may have been one that was partially addressed in the original analysis, while a subset of the data might focus on, for example, data from a particular phase of the research, or participants who share certain characteristics (e.g., motherhood). Given their similar foci, a supplementary analysis may appear to be part of the original research study, particularly if the dataset is the researcher's own, and it is re-used in its entirety. Conversely, the focus in a supra analysis "exceeds" rather than extends the scope of the original study. Thus, the focus of a supra analysis is less closely related to the original study's than that of a supplementary analysis, and the research questions may or may not arise from the original research. Also, these questions can be empirical, theoretical, or methodological in nature and thus may require the application of theories and/or data analysis methods that are different from those applied in the original study. Such new applications, as well as the divergence between the original and secondary studies, render supra analyses more controversial than supplementary analyses (discussed in Section 2.2). HEATON also found in her literature review that supra analyses were less common than supplementary analyses. [4]

The last two types of secondary analyses in HEATON's typology, amplified and assorted, refer to the use of multiple datasets, and the addition of new, primary research, respectively. There are different reasons why a researcher may utilize more than one dataset in a secondary analysis. In one type of amplified analysis, datasets are combined, or pooled, if the study populations and foci are similar. The purpose of such analyses is to increase the sample size and produce a synthesized analysis. Another type of amplified analysis searches for similarities and differences across datasets. Studies used for comparative purposes may differ in several respects, for example, the characteristics of the participants, or the country in which the studies were conducted. In her review, HEATON found that amplified analyses are the second-most common type of secondary analysis, while assorted analyses are the second-least common. In the latter type, the secondary analysis is augmented with additional primary research, which may involve different data collection and analysis methods from that of the original study. An assorted analysis may be conducted if the foci between the original and secondary studies are less closely related (as in the case of supra analysis), since the additional research can help draw more accurate conclusions. As such, both amplified and assorted analyses are utilized in conjunction with supplementary and supra analyses. For instance, a secondary analysis may be a supplementary-amplified analysis, or, like the study presented in this article, a supra-assorted analysis. [5]

2.2 Risks and benefits of secondary analysis

The literature identifies several concerns regarding secondary analysis that apply, but are not specific to, the re-use of auto-data. The first is a methodological concern regarding the "fit" between the research questions of the primary and secondary studies. HINDS and colleagues (1997) explain that the fit may be lacking if too much data on the phenomenon of interest is missing, which can occur if the focus of the original study narrowed, or the new topic arose late in the data collection process. Also, if the foci of the two studies differ too greatly, the data may lack sufficient detail and depth to draw new conclusions (ibid.); consequently, the risk of poor fit is greater for a supra, rather than a supplementary analysis (HEATON, 2004). Similarly, MAUTHNER and colleagues (1998) found that the data from their previous studies were unsuitable for secondary analysis because their perspectives, and the context of their lives had changed significantly over time. They argue that since data are the "product of the reflexive relationship between researcher and researched" (p.742), they cannot be treated as discrete and separate from the conditions under which they were produced. However, as each had returned to their data after many years, the ability to re-use auto-data may partially depend on the length of time between the original and secondary studies (HAMMERSLEY, 2010). In addition, significantly different foci raise an ethical issue since the secondary study could "violate the conditions under which consent was [originally] obtained" (p.551). However, a blanket consent acquired in the original study cannot account for every possible direction of subsequent analyses (YARDLEY et al., 2014, p.108), and re-contacting participants to obtain their consent for a secondary analysis could cause psychological, social, or other harm (CANADIAN INSTITUTES OF HEALTH RESEARCH, 2002; GRINYER, 2009). Thus, methodological and ethical concerns exist even when re-using auto-data; yet, primary researchers' proximity, or closeness, to the dataset "can be seen to grant a certain amount of privilege," including a greater ability to judge the fit of the two studies (COLTART et al., 2013, §22), or the best approach regarding informed consent. [6]

Indeed, proximity is the greatest benefit afforded by the re-use of auto-data. Primary researchers possess both contextual and tacit knowledge of their data, and are thus are more likely to properly interpret their own notes, and participants' meanings (HAMMERSLEY, 1997, 2010; THORNE, 1998). Since much that occurs in the data collection process is tacit and experiential in nature, there are limits to what a researcher can capture in text (HAMMERSLEY, 1997, 2010). THORNE (1998) also suggests that being closer to the data may help protect participants' privacy and confidentiality as it enhances primary researchers' ability to determine which information might be identifying. Additionally, secondary analyses, in general, can reduce the burden on participants, and save time, money, and effort (CORTI & BISHOP, 2005; REDMAN-MacLAREN, MILLS & TOMMBE, 2014). It is also argued that secondary analyses allow for greater maximization of datasets (HINDS et al., 1997), including those on hard-to-reach populations or sensitive topics (FIELDING & FIELDING, 2000). For some, utilizing datasets to their fullest is considered an ethical and moral responsibility, especially for researchers conducting publically funded research (YARDLEY et al., 2014). [7]

Although the re-use of one's own data is not without considerations and potential challenges, there are also several advantages if concerns can be adequately addressed. Again, this is the approach undertaken in the supra-assorted analysis featured in this article. [8]

3. Original Study

3.1 Background

The original study, Lone Mothers: Building Social Inclusion, was a five-year, national research project conducted between 2006 and 2011. Ethics approval was obtained from Wilfrid Laurier University's (WLU) Research Ethics Board in Waterloo, Ontario, and the primary research question for this study was: How are changes in the labor market and the social assistance (i.e., welfare) system impacting lone mothers across Canada? These changes include a labor market with fewer stable, well-paying jobs (VOSKO, 2006), and a social assistance system that increasingly reflects a workfare (or work-for-welfare) approach (CARAGATA, 2003; LIGHTMAN, HERD & MITCHELL, 2006). This study was funded as a community-university research alliance, by the Social Science and Humanities Research Council of Canada. It involved 21 investigators from five universities and eight community agencies, including municipal social assistance agencies, as well as four doctoral students and 22 lone mother research assistants (RAs). [9]

Also, to account for social, political, and economic differences, Lone Mothers: Building Social Inclusion was conducted in three cities across the country, namely: Vancouver, British Columbia; Toronto, Ontario; and St. John's, Newfoundland. Important regional differences exist between social assistance programs in terms of availability and generosity, which significantly impact poor lone mothers. In addition, the lone mother population in Canada is very diverse, varying, for example, in terms of Indigenous2) heritage and ethno-racial identity (MILAN, 2015). These are important differences that significantly shape the experience of lone mothers in the labor market and social assistance system; thus, this study attempted to capture these experiences by reflecting diversity in the sample of women who participated. [10]

3.2 Methodology

This study utilized a feminist and social inclusion lens to understand the impact of racialization, immigration, poverty, and other factors on the lives of the lone mothers accessing social assistance. Consistent with this lens, the work was grounded in a feminist, participatory action research methodology. Lone mothers were hired and trained as RAs to help shift the power relations typically found within research. Involvement in the research also helped develop the lone mothers' research skills, and provided them with an opportunity to be involved in advocacy efforts. Further, the lone mother RAs played an integral role in the study. They contributed to the development of the interview questions, conducted approximately half of the interviews, and also shaped the data analysis. They also served as an ongoing reference group and were involved as research participants. [11]

The 22 lone mother RAs were recruited through grassroots and non-profit organizations, and social assistance agencies. The required qualifications for the role were: lived experience with social assistance, a basic understanding of the social assistance system, and good communication skills. After recruitment, the lone mother RAs participated in an intensive training program. They also received support (e.g., debrief sessions) throughout the research project. Considered remunerative work, the women were paid for their time in training, and spent interviewing, debriefing, and analyzing data. In addition, childcare and transit costs were reimbursed to further reduce financial barriers. Research participants were also recruited to the study with the help of local organizations that posted flyers in their buildings, and informed their clients of the study. To participate, lone mothers had to have at least one child under the age of 18 living with them, be parenting without a live-in partner, and be accessing social assistance at the time of recruitment. Lone mothers that were interested in the study were invited to call the local research office for more information. [12]

In terms of research methods, this longitudinal study involved approximately 400 interviews in English with a total of 104 lone mothers (38 in Vancouver, 42 in Toronto, and 24 in St. John's). The lone mothers were interviewed once every 12 to 15 months over a five-year period, and the majority had their four-interview sequence conducted by the same researcher to help facilitate trust building. The sample of women in the study was diverse in terms of age, Indigenous heritage, immigrant status, and education level. The women were also carefully matched with researchers (i.e., lone mother RAs, doctoral students, community researchers, and academics) in terms of demographic and geographic factors to reduce participation barriers and facilitate data collection. The researchers also maintained contact with the lone mothers between interviews by phone, which further supported the research relationship, helped ensure awareness of changes in the women's lives (e.g., home address or major life events), and allowed the researchers to re-affirm participants' consent to further contact. The first interview was broad in its focus, and included questions on employment, income, health, housing, and social networks. Subsequent interviews began by discussing any key changes in the women's lives, followed by questions on specific life dimensions. Round 2 focused on experiences in the paid labor market and with social assistance; round 3, social networks; and round 4, goals, accomplishments, and reflections on the past four years. Also, per an iterative process, interview questions were revised after each round of interviews, which allowed the interviewers to follow up on specific issues or questions that arose in earlier interviews. [13]

All interviews were audio recorded and transcribed verbatim as they were completed, which facilitated revisions to subsequent interview guides, as well as data analysis and knowledge translation throughout the duration of the project. Transcripts were stripped of identifying information, uploaded into NVivo (a qualitative data analysis program), and initially coded (BRAUN & CLARKE, 2006) by doctoral student researchers using a descriptive code book. The code book was co-constructed by six research team members who had first independently developed a coding frame. This common coding frame contained descriptors that were derived from the data (i.e., data-driven codes), and based on previously identified areas of focus (i.e., theory-driven codes) (ibid.). For example, one such (broad) descriptor applied to the data was "housing." Also, since the common coding frame was descriptive only, the initially coded data offered various directions for analytic inquiry. Thus, after initial coding, researchers were free to apply analytic codes to the data based on their particular areas of interest (for example, see CARAGATA & LIEGGHIO, 2013; CUMMING, 2014; LIEGGHIO & CARAGATA, 2016; POLLACK & CARAGATA, 2010). This analytic strategy reflected the collaborative nature of the research project, and was necessary given the massive volume of data. In addition, the results from the initial coding were used to compile summaries after each round of interviews. These summaries were then shared with the lone mother RAs to gather feedback on the categories and developing analyses. [14]

4. Secondary Study

4.1 Background

Within the same year of completing the original study, we undertook a supra-assorted secondary analysis entitled The Lone Mother Resilience Project. The research question for the secondary analysis developed after the original study, and although it arose out of the principal researcher's broader program of research, the original study was identified as a relevant source of data. The purpose of the secondary analysis was to explore resilience and adversity in lone mothers' lives, including their understanding of resilience, and protective and risk factors in different life dimensions. The primary research question for this study was: What part does resilience play in lone mothers' coping in spite of persistent life ‘hassles' and/or major traumatic life events? While the foci of the two studies differed, we saw evidence of resilience and adversity in the original study's dataset. However, because the original study did not specifically explore these phenomena, we conducted additional primary research with some of the lone mothers from the original study to meet that purpose. The lone mothers who participated in Lone Mothers: Building Social Inclusion had consented to be contacted for future studies, and ethics approval was obtained by the Research Ethics Board at WLU. [15]

4.2 Methodology

Similar to the original study, the secondary analysis utilized a feminist and social inclusion lens. This lens was applied to better understand how experiences of resilience and adversity were related to spheres of exclusion (i.e., socio-political, geo-spatial, economic, and subjective) (SEN, 2000). This theoretical lens also informed the feminist methodology for the additional primary research, which was qualitative in nature and privileged the lived experiences of women; however, due to time and financial constraints, this study did not include a participatory component as in the original study. This study was comprised of three parts: 1. additional primary research through focus groups, 2. the re-use of data from an original study, and 3. additional primary research through in-depth, semi-structured interviews. A flow chart outlining the three parts of this study is provided (see Figure 1).



Figure 1: The three parts of the secondary analysis [16]

4.2.1 Preliminary focus groups

The first part of this study involved the generation of additional primary data through three focus groups. The purpose of these groups was to explore lone mothers' understandings of resilience and adversity, determine "cases" of resilience and non-resilience in the original study, and identify language regarding resilience and adversity that was understood and used by the lone mothers. Research with lone mothers supports the need to clarify such terms because there is often a disconnect between middle-class ideologies of resilience and adversity, and lone mothers' lived experiences (BRODSKY, 1999; LEVINE, 2009). Similarly, UNGAR (2005) argues that researchers should not define concepts associated with resilience without the perspective of those who participate in the research. As such, the focus groups explored these concepts and attempted to answer the following questions: What does resilience mean to lone mothers? What life events do lone mothers perceive as adverse? What factors do lone mothers consider to be "protective"? And lastly: What words or phrases do lone mothers use to discuss these concepts? [17]

First, we invited the lone mother RAs from Toronto to join a focus group, three of whom agreed to participate. We invited the lone mother RAs to participate in this focus group because we had developed relationships with them over the course of the original study, and we felt they could draw on both their lived experience in poverty, and their experience as RAs. We then invited another four lone mothers who had been interviewees but not RAs in the original study. These women were included in the focus group to increase the number of participants, and because they were active as a reference/advocacy group associated with the project. Originally, we planned to conduct just this one focus group, however, the discussion was so rich and site-specific, that we decided to conduct focus groups in all three study locations. As such, all of the lone mother RAs from the original study were invited to join a focus group, and 14 (64%) participated (five in Vancouver, three in Toronto, and six in St. John's). Thus, all 18 focus group participants had been interviewees in the original research study, and the majority (78%) had also been involved as RAs. All focus groups were approximately three hours in length, audio-recorded, and transcribed verbatim. [18]

Our approach to data collection and analysis was iterative and flexible (RUBIN & BABBIE, 2017). As noted above, we revised our research plan to conduct more focus groups than originally planned, and additionally, our observations from each focus group informed the data collection for subsequent groups. After each group, we noted our observations and thoughts (RABIEE, 2004), discussed initial impressions, and made minor revisions to the focus group guide as needed. Our notes also assisted us in being reflexive, wherein we questioned our thoughts and reactions and sought to be attentive to our assumptions and biases (HESSE-BIBER & PIATELLI, 2012). We then reviewed our notes and the transcripts to familiarize ourselves with the data (BRAUN & CLARKE, 2006), which helped to provide an overall impression of the dataset, before "breaking it into parts" (RABIEE, 2004, p.657). Next, the stripped transcripts were uploaded to NVivo and coded using a combination of data-driven and theory-driven codes (BRAUN & CLARKE, 2006). Data-driven codes were those generated from the focus groups, while codes derived from the literature included, for example, SEN's (2000) spheres of social exclusion. The focus groups were coded mainly at a sentence and paragraph level to retain context and meaning. From the initial coding, we were then able to generate lists (e.g., protective and risk factors). This level of analysis served our purposes at this stage in the research project (STEWART & SHAMDASANI, 2015), as the lists were helpful in identifying "cases" of resilience and non-resilience in the next part in the study, as well as terms and concepts used and understood by lone mothers. [19]

4.2.2 Re-use of existing data

The second part of our secondary analysis involved the re-use of the original study's dataset. First, we sorted the data by selecting only lone mothers who had completed at least three of the four interviews. This allowed us to better analyze their experiences of resilience and adversity over time, and reduced the original sample of 104 lone mothers to 70 (17 from Vancouver, 29 from Toronto, and 24 from St. John's). [20]

Next, we uploaded the interviews for the 70 cases into NVivo to facilitate the coding and organization of the data. Informed by both the literature and focus group data (BRAUN & CLARKE, 2006), we re-coded the interviews for: signs of resilience or positive outcomes (e.g., a new job), experiences of adversity and negative outcomes (e.g., relapse into addiction), and factors that could be either protective or a risk depending upon how they unfolded. Once again, our coding considered the various spheres of social exclusion, which ensured all aspects of the lone mothers' lives were analyzed for resilience and adversity. After reviewing and refining our initial codes (ibid.), we then typed up "profiles" for each of the 70 cases. Profiles were approximately one half page in length (single-spaced), and summarized the key positive and negative events and factors in each woman's life over the three to four year period. Also, per the logic of longitudinal coding (SALDANA, 2015), we organized each case's data in chronological order by interview date, which allowed for an assessment of change over time. For our purposes, we were interested in determining which of the lone mothers appeared to move toward resilience, and those who appeared to move away from resilience over time. For ease of analysis, we highlighted the key factors/events different colors as follows: protective factors/events, green; risk/adverse factors/events, red; and factors/events that could be either protective or adverse depending on how they unfolded in the future, yellow (see the Appendix for an example of this coding). This color coding served to more clearly identify cases of resilience and non-resilience, as well as the protective and risk factors in the women's lives. [21]

After identifying cases of resilience and non-resilience, we then sorted the data into two subsets. In the first subset, we selected an equal number of cases from each location that most clearly demonstrated resilience, which resulted in a total of 18 cases (i.e., six from each location). As much as possible, we purposefully selected cases to obtain a diverse sample in terms age, location, education, ethnicity, Indigenous heritage, and immigrant status. These 18 lone mothers were then contacted regarding participation in the third part of the study. With regard to the second subset, we selected six lone mother cases (i.e., two from each location) who appeared to show a lack of resilience at that particular point in their lives, and also reflected the diversity of the lone mothers in the study. We selected these cases for comparative purposes; however, we only selected a small number, and we did not contact those lone mothers for an interview, because our focus was primarily on lone mothers' experiences of resilience. Nonetheless, the non-resilience cases enhanced our understanding of adversity that appeared to be less readily overcome by some of the lone mothers, as well as life dimensions that appeared to particularly lack protective factors. [22]

4.2.3 Semi-structured interviews

The third and final part of the secondary analysis involved additional primary research in the form of two interview panels with 20 lone mothers. The first panel was comprised of 15 of the 18 resilience cases identified in the secondary analysis (i.e., six in Vancouver, five in Toronto, and four in St. John's). Although all 18 women were invited to participate, three declined, thus resulting in an 83% participation rate. All of these 15 women were accessing social assistance at the time of the study. In the semi-structured interviews, we explored their subjective understanding of adversity as well as ameliorating factors that enabled their apparent resilience. We also updated the women's demographic information from the original study. Interviews were held in the women's respective cities, and lasted between 1.5 and 2 hours. All interviews were audio-recorded and transcribed verbatim. The second panel was comprised of five lone mothers who were not currently accessing social assistance, though they had done so in the past. Also, these lone mothers had not been involved in the original study. We convened this second panel because we wanted to further increase the diversity of the sample of lone mothers in our study. Though these lone mothers also had low incomes, we felt that contact with the social assistance system may significantly affect experiences of resilience and adversity, and thus wanted to capture some stories without that influence. All the women from the second panel were recruited from Toronto where we had extensive community connections. Specifically, they were referred to the study by lone mothers associated with the study, a municipal social assistance agency, and through informal networks. The interview guide for the second panel included some additional demographic and contextual questions, but otherwise was comparable to the first panel. Interviews were between 1.5 and 2 hours in length, audio-recorded, and transcribed verbatim thereafter. [23]

Our approach to the data collection and analysis of the interviews was similar to that of the focus groups. We sought to be iterative, flexible, and reflexive throughout the process by recording our thoughts and impressions after each interview, discussing our notes, and making minor revisions to the interview guide as needed. However, unlike the focus groups, we conducted a full thematic analysis of the interview data, which included the following steps: familiarization with the data, coding and refinement, categorization, and theorization (BRAUN & CLARKE, 2006; CONNOLLY, 2003). We familiarized ourselves with the interview data by reviewing our notes, listening to the recordings, and re-reading the transcripts (BRAUN & CLARKE, 2006). Stripped transcripts were then uploaded to NVivo for coding. At this point in the study, we had a well-developed coding frame that had been informed by the literature, the focus groups, and the re-coding of the data from the original study. Once again, we coded for resilience, adversity, and risk and protective factors, within a social exclusion framework. Per the objectives of our study, we were also interested in the meaning of resilience to lone mothers, and coded for such evidence accordingly. The data were coded mainly at the sentence and paragraph level to retain context, after which we began to search for patterns across the interviews. These patterns were organized into themes, which we refined through discussion and re-reading of the transcripts as necessary (ibid.). The findings from this secondary study, including our theorization of the data, have since been presented as working papers (see CARAGATA, CUMMING & WATTERS, 2017; CUMMING, CARAGATA & WATTERS, 2017), and thus for the purpose of this methodological article, are not summarized here. [24]

5. Conclusion

In this article, we describe The Lone Mother Resilience Project, which was a supra-assorted secondary analysis that engaged a total of 38 poor lone mothers in focus groups and semi-structured interviews. Additionally, we re-used data from Lone Mothers: Building Social Inclusion, which was a longitudinal study with 104 lone mothers across Canada. There were several strengths associated with the secondary study's research design. Since we had recently completed the original study, we were still able to connect with, and relate to, the dataset. Also, due to our knowledge of the data, we were confident in its fit with the focus of the secondary analysis in terms of relevance and depth. While the original study did not specifically address resilience and adversity, evidence of these phenomena were present in the data. In addition, the dataset was rich and comprehensive. It was generated from approximately 400 in-depth interviews over a five-year period, and was strengthened in several other ways. Specifically, and where possible, lone mothers were matched with interviewers based on demographics, and had their interviews conducted by the same individual to facilitate trust-building. Similarly, many interviewees felt comfortable sharing their stories with one of the 22 lone mother RAs, who conducted approximately half of all interviews, and also contributed to the data analysis process. [25]

In addition, our familiarity with the original study informed the design of the secondary analysis, and facilitated the re-use of its data. The secondary analysis was comprised of three parts: preliminary focus groups, sorting and re-coding of the original study's data, and semi-structured interviews. As this study was a supra-assorted analysis, the collection of additional primary research helped to ensure we had enough data, while the use of two different data collection methods contributed to a deeper understanding of the phenomena of interest (i.e., resilience and adversity). In addition, we engaged 33 lone mothers from the original study, of whom 15 had also been hired as RAs. This continued the relationship between the researchers and previous lone mother participants, which facilitated trust-building and data generation. Also, their experiences as former interviewees and/or RAs may have shaped their understandings of resilience and adversity in their own life, and other lone mothers' lives. This allowed us to draw comparisons, albeit limited, between this group of lone mothers and the five who had not been involved in the original study. Lastly, both the original and secondary studies engaged lone mothers across Canada, and reflected much diversity of the lone mother population. This helped to capture regional and social differences and allowed for comparisons to be drawn within the sample. [26]

There are also some limitations to the secondary study's design. The secondary analysis lacked a participatory component, which was a particular methodological strength of the original study. Although the data were analyzed using the same theoretical lens, and most participants had been involved in the original study, limited time and finances restricted our ability to engage lone mothers as RAs. In addition to strengthening the data, engagement of the lone mother RAs in the original research project helped to reduce the study's power imbalance, and to develop the women's research and advocacy skills. Thus, the original study not only produced rich and comprehensive data, but it also helped create 22 new lone mother activists. Another limitation relates to the secondary study's sample. Since most participants were recruited from the original study, which was limited to English-speaking, urban lone mothers, some sub-populations were excluded in both studies. Although we were not seeking a representative sample, and we captured the diversity of Canada's lone mother population in other ways, rural and non-English speaking lone mothers are sub-populations to include in future research. [27]

In terms of lessons learned, our experience with this secondary study reinforced the importance of seeking consent to re-contact participants for future studies. Having obtained such consent in the original study along with our prolonged engagement with participants, made recruitment for the secondary study relatively easy. We had current contact information for participants, and knew that they met the criteria for the secondary analysis. Thus, re-engaging some original participants saved recruitment time, and provided additional benefits regarding the trustworthiness of the data, as discussed above. [28]

The supra-assorted secondary analysis we discuss in this article made greater use of an original study's dataset, and privileged the voices of poor lone mothers in Canada. In so doing, we produced new knowledge about under-researched phenomena, and highlighted the struggles of, and supports needed by, lone mothers. Thus, it is our hope that the published findings from both studies inform Canada's social policy, and contribute to the betterment of lone mothers across the country. [29]

Acknowledgments

The research upon which this article is based was funded by Human Resources Skills Development Canada.

Appendix: Illustration of Resiliency Profile Building Process Using Existing Dataset

"Alice"

Interview #1

Alice came to Canada 16 years ago from China. She left her abusive husband (R3)) prior to commencing studies, and lived in a shelter (G) for three months. The shelter helped her get legal aid, social assistance, subsidized childcare, and her husband removed from their apartment (G). Her ex-husband now has supervised visitation rights to see the children, which is stressful for her (R). She fought hard to get her kids in a school that is attached to a daycare. She sought out her local Member of Provincial Parliament for help (G). She has a brother and sister in the area, but they do not provide any support. She likes the area that they are living in, and has a good building superintendent. She has one good friend (Y). She does not use food banks. She has no cable TV or internet in an effort to save money. She is getting a recreation subsidy for her family (G) and loves it.

Interview #2

Alice finished a pharmacy assistant program paid for by social assistance (G), and is now looking for a job. She has subsidized childcare (G). She had a terrible social assistance worker (R), but just recently got a new worker who is helpful and she really likes (G). She lost her rent top-up of $100/month from social assistance, so is now finding that money is very tight (R). Her co-workers at her pharmacy placement are very derogatory towards social assistance recipients (R) (they do not know she is accessing it). Her university degree is not recognized in Canada (R) so she has only had several low income jobs, never making more than $8.00/hour. Her children's well-being is her first priority, and she feels that if she cannot find work that is conducive with their schedule, she would rather live in poverty (i.e., be unemployed and access social assistance) (Y).

Interview #3

Alice got a job at a pharmacy but the employer kept changing her shifts to evenings (R). She was earning $8.75/hour. Her mother and sister are being very negative (R), and her sister tells her that she should get back with her abusive ex-husband so that she and the children are not accessing social assistance. Her mother tells her that her job is not good because she is not there for her kids, which makes her feel very guilty. She feels defeated (R). She joins a support group for lone mothers (G) one night each week and finds it very helpful.

Interview #4

Alice has a full-time job, which she says changed her life (G). She does not get much more money than when she was accessing social assistance, but she has expanded her social network (G). She is paid $12.75/hour and gets a drug card from social assistance. She is meeting people from her culture every day in her new job. She has lived in the same place for 16 years so she feels secure (G) and likes her neighborhood. She feels very strong and highly motivated (G). She continues to be involved with non-profit groups (G). She has never received child support from her ex-husband and his supervised visitations with the children have been sporadic (R), however, he helps with driving the kids to their activities and appointments (G).

Determination: Resilience

Alice escapes an abusive partner to a women's shelter and then finds subsidized housing. She fights to get her children in a school that is attached to a daycare. She continues to upgrade her skills and English, and by round four has a full-time job that is supplemented by social assistance.

Notes

1) For the purposes of this article, we apply the term "auto-data" to primary research teams where at least one researcher was involved in the original study's data collection and analysis. In our case, the data re-used in the secondary study was derived from the principal investigator's (i.e., third author's) original study. Also, the third and second author were both involved the data collection and analysis for the original and secondary studies. <back>

2) There are three main groups of Indigenous peoples in Canada: First Nations, Inuit, and Métis (PARROTT, 2015). <back>

3) Key: G = green, Y = yellow, r = red <back>

References

Braun, Virginia & Clarke, Victoria (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101.

Brodsky, Anne E. (1999). "Making it": The components and process of resilience among urban, African American single mothers. American Journal of Orthopsychiatry, 69(2), 148-54.

Canadian Institutes of Health Research (2002). Secondary use of personal information in health research: Case studies. Ottawa: Public Works and Government Services Canada, http://www.cihr-irsc.gc.ca/e/documents/case_studies_nov2002_e.pdf [Accessed: February 20, 2017].

Caragata, Lea (2003). Neoconservative realities: The social and economic marginalization of Canadian women. International Sociology, 18(3), 559-580.

Caragata, Lea & Liegghio, Maria (2013). Mental health, welfare reliance, and lone motherhood. Canadian Journal of Community Mental Health, 32(1), 95-107.

Caragata, Lea; Cumming, Sara J. & Watters, Elizabeth C. (2017). Ameliorating adversity in the lives of lone mothers. Paper presented at the Pathways to Resilience Conference IV, Cape Town, South Africa, June 14-16.

Coltart, Carrie; Henwood, Karen & Shirani, Fiona (2013). Qualitative secondary analysis in austere times: Ethical, professional and methodological considerations. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 14(1), Art. 18, http://dx.doi.org/10.17169/fqs-14.1.1885 [Accessed: November 30, 2016].

Connolly, Marie (2003). Qualitative analysis: A teaching tool for social work research. Qualitative Social Work, 2(1), 103-112.

Corti, Louise & Bishop, Libby (2005). Strategies in teaching secondary analysis of qualitative data. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(1), Art. 47, http://dx.doi.org/10.17169/fqs-6.1.509 [Accessed: February 20, 2017].

Corti, Louise; Day, Annette & Backhouse, Gill (2000). Confidentiality and informed consent: Issues for consideration in the preservation of and provision of access to qualitative data archives. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 1(3), Art. 7, http://dx.doi.org/10.17169/fqs-1.3.1024 [Accessed: November 30, 2016].

Cumming, Sara J. (2014). Lone mothers exiting social assistance: Gender, social exclusion and social capital. Doctoral dissertation, University of Waterloo, Waterloo, Canada, https://uwspace.uwaterloo.ca/handle/10012/8444 [Accessed: May 4, 2018].

Cumming, Sara J.; Caragata, Lea & Watters, Elizabeth C. (2017). Conceptualizing resiliency: Lone mothers' understanding and experiences of resiliency. Paper presented at the Pathways to Resilience Conference IV, Cape Town, South Africa, June 14-16.

Fielding, Nigel G. & Fielding, Jane L. (2000). Resistance and adaptation to criminal identity: Using secondary analysis to evaluate classic studies of crime and deviance. Sociology, 34(4), 671-689.

Grinyer, Anne (2009). The ethics of the secondary analysis and further use of qualitative data. Social Research Update, 56, 1-4.

Hammersley, Martyn (1997). Qualitative data archiving: Some reflections on its prospects and problems. Sociology, 31(1), 131-142.

Hammersley, Martyn (2010). Can we re-use qualitative data via secondary analysis? Notes on some terminological and substantive issues. Sociological Research Online, 15(1), 1-7.

Heaton, Janet (2000). Secondary analysis of qualitative data: A review of the literature. York: Social Policy Research Unit, University of York, http://www.researchcatalogue.esrc.ac.uk/grants/R000222918/outputs/read/a4f1b793-44b2-4eec-9769-11ac5204d6aa [Accessed: November 30, 2016].

Heaton, Janet (2004). Reworking qualitative data. Thousand Oaks, CA: Sage.

Heaton, Janet (2008). Secondary analysis of qualitative data: An overview. Historical Social Research, 33(3), 33-45.

Hesse-Biber, Sharlene N. & Piatelli, Deborah (2012). The feminist practice of holistic reflexivity. In Sharlene N. Hesse-Biber (Ed.), The handbook of feminist research: Theory & praxis (pp.557-582). Thousand Oaks, CA: Sage.

Hinds, Pamela S.; Vogel, Ralph J. & Clarke-Steffen, Laura (1997). The possibilities and pitfalls of doing a secondary analysis of a qualitative data set. Qualitative Health Research, 7(3), 408-424.

Irwin, Sarah (2013). Qualitative secondary data analysis: Ethics, epistemology and context. Progress in Development Studies, 13(4), 295-306.

Levine, Kathryn A. (2009). Against all odds: Resilience in single mothers of children with disabilities. Social Work in Health Care, 48(4), 402-419.

Liegghio, Maria & Caragata, Lea (2016). “Why are you talking to me like I'm stupid?" The micro-aggressions committed within the social welfare system against lone mothers. Affilia, 31(1), 7-23.

Lightman, Ernie; Herd, Dean & Mitchell, Andrew (2006). Exploring the local implementation of Ontario Works. Studies in Political Economy, 78(1), 119-143.

Long-Sutehall, Tracy; Sque, Magi & Addington-Hall, Julia (2010). Secondary analysis of qualitative data: A valuable method for exploring sensitive issues with an elusive population?. Journal of Research in Nursing, 16(4), 335-344.

Mauthner, Natasha S. & Parry, Odette (2009). Qualitative data preservation and sharing in the social sciences: On whose philosophical terms?. Australian Journal of Social Issues, 44(3), 291-307.

Mauthner, Natasha S.; Parry, Odette & Backett-Milburn, Kathryn (1998). The data are out there, or are they? Implications for archiving and revisiting qualitative data. Sociology, 32(4), 733-745.

Milan, Anne (2015). Families & living arrangements, http://www.statcan.gc.ca/pub/89-503-x/2015001/article/14235-eng.htm [Accessed: November 30, 2016].

Parrott, Zack (2015). Indigenous peoples, http://www.thecanadianencyclopedia.ca/en/article/aboriginal-people/ [Accessed: February 20, 2017].

Pollack, Shoshana & Caragata, Lea (2010). Contestation and accommodation: Constructions of lone mothers' subjectivity through workfare discourse and practice. Affilia, 25(3), 264-277.

Rabiee, Fatemeh (2004). Focus-group interview and data analysis. Proceedings of the Nutrition Society, 63, 655-660.

Redman-MacLaren, Michelle; Mills, Jane & Tommbe, Rachael (2014). Interpretive focus groups: A participatory method for interpreting and extending secondary analysis of qualitative data. Global Health Action, 7(1), Art. 25214, http://dx.doi.org/10.3402/gha.v7.25214 [Accessed November 16, 2016].

Rubin, Allen & Babbie, Earl R. (2017). Research methods for social work (9th ed.). Boston, CA: Cengage Learning.

Saldaña, Johnny (2015). The coding manual for qualitative researchers. Thousand Oaks, CA: Sage.

Sen, Amartya (2000). Social exclusion: Concept, application, and scrutiny. Manila: Office of Environment and Social Development, Asian Development Bank, http://www.adb.org/sites/default/files/publication/29778/social-exclusion.pdf [Accessed: November 16, 2016].

Stewart, David W. & Shamdasani, Prem N. (2015). Analyzing focus group data. In David W. Stewart & Prem N. Shamdasani (Eds.), Focus groups: Theory & practice (3rd ed., pp.115-140). Thousand Oaks, CA: Sage.

Thorne, Sally (1998). Ethical and representational issues in qualitative secondary analysis. Qualitative Health Research, 8(4), 547-555.

Ungar, Michael (2005). Introduction: Resilience across cultures and contexts. In Michael Ungar (Ed.), Handbook for working with children and youth: Pathways to resilience across cultures and contexts (pp.xv-xxxix). Thousand Oaks, CA: Sage.

Vosko, Leah F. (2006). Precarious employment: Towards an improved understanding of labour market insecurity in Canada. In Leah F. Vosko (Ed.), Precarious employment: Understanding labour market insecurity in Canada (pp.3-42). Kingston: McGill-Queen's University Press.

Yardley, Sarah J.; Watts, Kate M.; Pearson, Jennifer & Richardson, Jane C. (2014). Ethical issues in the re-use of qualitative data: Perspectives from literature, practice, and participants. Qualitative Health Research, 24(1), 102-113.

Authors

Elizabeth C. WATTERS is a doctoral candidate in the Faculty of Social Work at Wilfrid Laurier University, Ontario. Her doctoral research explores the impact of precarious employment on the health of racialized immigrant women in Southwestern Ontario. She obtained her master of social work degree at Wilfrid Laurier University, and her bachelor of science degree at York University in Toronto, Ontario. She has a decade of professional experience in the public and non-profit sectors, and has worked in the areas of health promotion, vocational rehabilitation, diversity services, research, and university teaching. As a doctoral student, she has provided research assistance to several projects, including knowledge mobilization support to The Lone Mother Resilience Project.

Contact:

Elizabeth C. Watters

Lyle S. Hallman Faculty of Social Work
Wilfrid Laurier University
120 Duke Street West, Kitchener, Ontario, Canada, N2H 3W8

E-mail: ewatters@wlu.ca

 

Sara J. CUMMING is a professor of sociology in the Faculty of Humanities & Social Sciences at Sheridan College, Ontario. Throughout her career, Sara has held multiple research grants in partnership with regional government and not-for-profits assessing the quality of services offered to marginalized populations. Her most recent grant is a collaboration with Dr. Michael McNAMARA which relies on qualitative research and creative problem solving to help community partners produce new, creative, and fundable projects aimed at ameliorating hardships for vulnerable populations in Halton Region, Ontario. Sara teaches and publishes in the areas of introductory sociology, participatory action research, gender, social inequality, xenophobism and social policy. Prior to joining Sheridan in 2013, Sara was a term professor at Brock University, Ontario and the University of Waterloo, Ontario, as well as a contract writer for a Canadian publishing company. Sara is the co-chair of Applied Sociology in Canada, a research cluster under the Canadian Sociological Association. She received her PhD in sociology from the University of Waterloo.

Contact:

Dr. Sara J. Cumming

Faculty of Humanities & Social Sciences
Sheridan College
Hazel McCallion Campus, 4180 Duke of York Blvd., Mississauga, Ontario, Canada, LB5 0G5

Tel.: +1 905-845-9430, ext. 2858
Fax: +1 905-815-4072

E-mail: sara.cumming@sheridancollege.ca
URL: https://www.sheridancollege.ca/academics/faculties/humanities-and-social-sciences/faculty-profiles/sara-cumming.aspx

 

Lea CARAGATA is a professor of social policy at the Faculty of Social Work at Wilfrid Laurier University, Ontario. She completed her PhD at the University of Toronto, Ontario, focused on the interplay between land use, social movements and the democratization of public space. Her return to academe to do a PhD followed an almost 20 year career that included grassroots community organizing, social housing development, public policy coordination, and public administration in non-profit community organizations and in government. She continues to blend academic and research interests with her interest and commitment to public policy change and community development. Her recent book "Not the Whole Story: Challenging the Single Mother Narrative" is an illustration of her participatory, activist work.

Contact:

Dr. Lea Caragata

Lyle S. Hallman Faculty of Social Work
Wilfrid Laurier University
120 Duke Street West, Kitchener, Ontario, Canada, N2H 3W8

Tel.: +1 519-884-0710, ext. 5219
Fax: +1 519-888-9732

E-mail: lcaragata@wlu.ca
URL: https://www.wlu.ca/academics/faculties/faculty-of-social-work/faculty-profiles/lea-caragata/

Citation

Watters, Elizabeth C.; Cumming, Sara J. & Caragata, Lea (2018). The Lone Mother Resilience Project: A Qualitative Secondary Analysis [29 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 19(2), Art. 23, http://dx.doi.org/10.17169/fqs-19.2.2863.



Copyright (c) 2018 Elizabeth C Watters, Sara Cumming, Lea Caragata

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.