Mixed Methods for Hard-to-Reach Populations: Lessons from Studying Economic Vulnerability in a Rare Ethno-Religious Community

Introduction

Studying economically vulnerable populations presents numerous methodological challenges, particularly when these populations exist within small, comparatively affluent ethno-religious minorities where poverty contradicts dominant cultural and social narratives. American Jews exemplify this challenge. A rare population, American Jews comprise only 2% of the U.S. population, have median household incomes above the national median (Horwitz 2025; Pew Research Center 2021), and are situated within and shaped by public and communal perceptions of economic success and affluence. These generalizations and perceptions, however, mask significant pockets of economic distress, with the most reputable probability survey of American Jews indicating that approximately a third have experienced economic vulnerability at some point in their lives (Kotler-Berkowitz 2022; Pew Research Center 2021). To deepen our understanding of this often-hidden vulnerability, we deployed a mixed methodological approach that combined nonprobability survey research and in-depth interviews (Creswell and Plano Clark 2017). This allowed us to examine and analyze directionally broad patterns across survey respondents, life-course narratives of those who have experienced economic vulnerability, and the work of human services professionals addressing economic precarity among Jewish clients.

The Challenges of Studying American Jews and Jewish Economic Vulnerability

Traditional probability survey methods face multiple difficulties when studying Jews. The Jewish population’s small size means most nationally representative probability surveys of the U.S. population contain far too few Jewish respondents for analysis. To address this problem, researchers have employed various stratified probability sampling designs to produce customized representative samples of Jews nationally (Pew Research Center 2013, 2020) and in local metropolitan areas (Kotler-Berkowitz 2016) that have provided abundant data on American Jews. However, stratified probability sampling of a rare population like Jews is extremely expensive, requiring major investments from funders. In many circumstances where researchers and funders want to examine aspects of Jewish experience in more depth or from different perspectives than secondary analysis of current national and local probability studies can provide, stratified probability sampling is not feasible due to costs.

Beyond probability survey-based difficulties, researchers studying economic vulnerability among Jews encounter cultural and recruitment challenges. Shame and stigma around poverty in a community celebrated for its success (Kranson 2017) create barriers to recruiting research participants. Economically vulnerable Jews often actively avoid Jewish organizational spaces and personal networks where their economic struggles may become visible (Horwitz and Lascar 2021), which are often the very spaces and networks researchers rely upon for recruitment and referrals. Other groups that are likely to experience elevated levels of economic insecurity can also be comparatively difficult to reach. Ultra-Orthodox Jews typically resist secular research participation. Jewish immigrants (commonly from Latin America and the former Soviet Union) tend to face linguistic and cultural barriers to participation. Young, secular Jews struggling with economic hardship may lack both Jewish institutional connections and traditional cultural markers that identify them as Jewish to researchers.

A Mixed Methods Approach

We were part of a research team at Tulane University and Rosov Consulting, funded by the Harry and Jeanette Weinberg Foundation, that wanted to learn more about the lived experience of economic precarity among American Jews than was available from secondary analysis of current probability-based surveys (Horwitz et al. 2024). Furthermore, we were committed to giving qualitative voice to those who have struggled with economic insecurity and to helping inform communal policy around economic vulnerability.

Because budget constraints precluded a probability design, we deployed a nonprobability survey from two online survey panels, explicitly recognizing in our reporting that the survey could not be used to generate estimates of the size of the economically vulnerable population nor could it be subject to statistical significance testing for inference to all US Jews (Hirschauer et al. 2020). Instead, we utilized the sample and extended survey instrument to examine broad, directional patterns among survey respondents about their experiences with economic precarity, and to compare those who have and have not faced economic insecurity during their lifetimes.

Simultaneously, we aimed to push beyond survey data, adding qualitative research into our design through follow-up, in-depth life-course interviews with survey respondents and interviews with human service professionals working with economically vulnerable Jewish clients. The follow-up interviews with survey respondents provided narrative depth that validated survey patterns and revealed pathways and life courses shaped by economic vulnerability that quantitative data alone could not fully capture.

The interviews with human service providers added an innovative element to the study, generating important insights that triangulated survey data patterns and the life-course stories we heard from interview participants. This integrated mixed methods design proved essential not just for documenting vulnerability’s existence, but for understanding its varying manifestations across an ethnically, religiously, and culturally diverse population where economic precarity remains largely hidden.^[1]

Phase 1: Nonprobability Survey (August-October 2023)

The foundational quantitative phase employed a dual-panel online survey approach, ultimately yielding 1,958 responses from self-identified Jewish adults. We partnered with two established panel providers—Qualtrics (August 16-October 24, 2023) and Dynata (August 29-October 5, 2023)—to recruit survey respondents. This dual-source approach addressed coverage limitations of single panels but introduced complexity requiring careful de-duplication based on IP and email addresses. The de-duplication process retained first submissions, yielding 1,440 Qualtrics responses and 518 Dynata responses.

Recruiters set quotas based on age, region, gender, and economic vulnerability. This quota approach was essential given the challenges of reaching economically vulnerable Jews who may have comparatively limited survey panel participation due to time constraints, internet access issues, or general disconnection from research opportunities, as well as ensuring representation across key demographics in the unweighted sample

Defining economic vulnerability required moving beyond federal poverty thresholds, which community observers note set unrealistically low-income levels to meet both secular and Jewish needs (Cooperman and Kotler-Berkowitz 2019). We ultimately used respondents’ subjective self-assessments of their current and past ability as adults to meet basic expenses, assessments of their family’s ability to meet basic expenses when they were growing up, and measures of income and household size to develop a three category measure of vulnerability: currently/recently vulnerable (experiencing hardship within five years), previously vulnerable (hardship during earlier adulthood or childhood), and never vulnerable. This temporal dimension proved crucial for understanding vulnerability as a dynamic rather than static condition.

To address potential selection bias from nonprobability sampling, we implemented calibration weights using targets from the 2020 Pew Research Center probability survey of US Jews for gender identity, region, age, education, marital status, Jewish denomination, Ashkenazi ethnicity, current financial assessment, and childhood financial assessment. This weighting strategy partially addressed representation concerns but could not mitigate nonprobability sampling’s inherent limitations on statistical generalizability.

Phase 2: In-Depth Life-Course Interviews (October 2023-May 2024)

The second phase transformed statistical patterns into human narratives through 175 semi-structured interviews with survey respondents. Participants who agreed to further research provided email addresses at the survey’s conclusion.

We employed purposeful sequential recruitment across two sampling frames. The Dynata recruitment (September 27-mid-October 2023) prioritized economically vulnerable respondents. The Qualtrics recruitment (beginning November 11, 2023) initially targeted vulnerable respondents before expanding to include marginalized identity groups (non-white, LGBTQ+) among the never vulnerable, ensuring intersectional perspectives. All participants received $100 gift cards through www.tremendous.com.

The 60-90 minute semi-structured interviews, conducted via Zoom with phone alternatives available, drew on the American Voices Project’s tell-me-the-story-of-your-life methodology—an approach designed to foster deep listening and capture everyday experiences holistically (Edin et al. 2024). Beginning with life histories, conversations progressed through family dynamics, neighborhood conditions, and current financial circumstances before examining political views, healthcare access, emotional well-being, and Jewish communal involvement. This progression captured holistic portraits of how economic vulnerability unfolds across the life course.

The interviews served dual analytical purposes. Individual narratives illuminated specific quantitative patterns. For example, several interviews detailed downward mobility pathways and revealed how chronic health conditions undermine economic stability. These deep portraits demonstrated how interconnected challenges compound over time, bringing statistical findings to life. We also synthesized responses across participants to explore broader themes. When examining clergy interactions or Jewish community support access, for instance, we grouped multiple perspectives to reveal the spectrum of experiences and challenges, identifying both commonalities and variations across the sample.

We adapted our interview protocol in response to the October 7, 2023, attacks on Israel. For interviews conducted October 7-December 31 2023, we added sensitive acknowledgment: “We wanted to start off by acknowledging October 7 and the war in Israel. How are you doing? Do you have any family or friends who are affected by this?” This modification captured current stressors while maintaining focus on economic vulnerability. From January 2024 forward, we removed this preface, allowing organic emergence of these concerns during the interviews.^[2]

Phase 3: Professional Stakeholder Interviews (January-February 2024)

To add another perspective to our study, we conducted 38 interviews with human service professionals who serve economically vulnerable Jewish clients (January-February 2024). We recruited through the Network for Jewish Human Service Agencies and The Weinberg Foundation. This yielded 33 interviews with professionals from Jewish agencies and five from adjacent agencies that have a significant Jewish client base. These 30–60-minute interviews confirmed patterns of vulnerability in the survey data and the follow-up in-depth interviews with survey respondents, while identifying populations our survey and follow-up interviews could not reach, such as individuals whose vulnerability is so severe that it prevents research participation. Professionals also helped explain mechanisms behind survey findings—for instance, why vulnerable respondents avoided synagogue assistance (e.g., shame, lack of clergy relationships, institutional barriers).

Methodological and Policy Insights From Mixed Methods Research

While the survey captured point-in-time vulnerability status, integration with life history interviews revealed three distinct trajectories: situational vulnerability, generational vulnerability, and downward mobility (Horwitz et al. 2024). Situational vulnerability proved more than twice as prevalent as generational vulnerability and four times as prevalent as downward mobility. This distribution only became meaningful through interview narratives explaining how respondents entered each pathway. For instance, one respondent’s survey indicated current vulnerability; their interview revealed a cascade beginning with spouse’s cancer diagnosis, leading to caregiving responsibilities, job loss, and eventual financial crisis. This reshaped our understanding from poverty as chronic condition to vulnerability as situational crisis.

The mixed methods design also revealed how economic vulnerability rarely occurs in isolation. Survey data showed that 74% of currently or recently vulnerable respondents experienced at least one co-occurring adverse factor, including health crises, divorce, caregiving responsibilities, or job loss. Interviews illuminated how these factors compound over time. As one respondent explained, “Since getting divorced, I’ve had to learn a lot of new things. I’ve come a long way, but I’m still figuring out the income part.” This co-occurrence pattern, visible in aggregate through survey data but only explicable through narrative, underscores the need for integrated services addressing multiple domains simultaneously.

The human service interviews helped triangulate our data. For instance, when our survey showed that government benefits were the most commonly cited factor helping formerly vulnerable respondents achieve stability, human service professionals independently confirmed this pattern, noting the critical role of expanded benefits during COVID-19. This triangulation across three data sources (survey responses, life history narratives, and professional observations) strengthened confidence in findings despite nonprobability sampling.

The mixed methods approach built on findings first seen in the survey data. For example, survey data showed vulnerable respondents had lower synagogue participation, which we initially attributed to membership costs. However, the follow-up interviews with survey respondents and with human service professionals revealed deeper barriers, such as shame. Interviews also uncovered how Jewish institutions’ family-centric programming systematically excludes vulnerable populations. Single parents couldn’t attend evening programs without childcare; older singles felt invisible in family-focused congregations; geographically isolated individuals lacked transportation. The mixed methods approach transformed our understanding from simple cost barriers to complex institutional-cultural exclusion requiring systemic reform.

The integration revealed that childhood vulnerability requires different interventions than adult vulnerability. For those experiencing childhood poverty, completing educational degrees and obtaining reliable employment as adults were the primary pathways to stability. However, for those experiencing situational vulnerability as adults, the interventions were more varied—from government benefits to therapy to legal services. This temporal distinction, first seen in the survey data and confirmed through interviews, suggests age-specific intervention strategies.

We are grateful that our research is being utilized by professionals at Together Ending Need (TEN), an initiative of the Jewish Funders Network and the Harry and Jeanette Weinberg Foundation, and has been highlighted in Jewish community news reports, journals, lectures, webinars, conferences, op-eds, and other online content. Our mixed methods approach provides an integrated and holistic way to communicate findings to communal professionals and advocates and through them to the donors and volunteers that they seek to mobilize to address economic vulnerability.

Both statistics and stories play an important role in this effort. The broadly-defined patterns that emerge from statistics are an important foundational source of information for rational policy making, planning, and resource deployment. In turn, stories, especially stories of lived experiences, can be more impactful and longer lasting than numbers in shaping people’s beliefs, attitudes, motivations, actions (Graeber et al. 2023). No matter the issue, researchers seeking to move beyond the academy and to inform policy and advocacy efforts can be well served by mixed methods that support, complement and cross-validate each other.

Conclusion

Studying economically vulnerable populations within rare populations requires methodological innovation. Our mixed methods design demonstrates how integration can address sampling challenges, validate subjective measures, reveal complex causal pathways, and capture heterogeneity within vulnerable populations, thereby transforming limitations into strengths. Nonprobability sampling becomes less problematic when triangulated through qualitative data. Cross-sectional surveys gain temporal dimension through life history interviews. Statistical patterns gain meaning through lived experience narratives. Understanding economic precarity within comparatively small religious and ethnic communities requires both breadth and depth—counting those affected while honoring their stories. Mixed methods research provides the methodological framework for achieving both goals, ensuring vulnerable voices shape both research findings and policy responses.

Acknowledgements

This article is based on a study conducted by a research team at Tulane University and Rosov Consulting, funded by the Harry and Jeanette Weinberg Foundation, and originally published as On the Edge: Voices of Economic Vulnerability in U.S. Jewish Communities (Horwitz, Kotler-Berkowitz, Jollymore, and Rosov 2024). A full list of the colleagues and institutions who contributed to the study and report—as researchers, funders, advisors, project managers, and operations and production staff—appears in the front matter to the study’s main report. We thank them all, while we retain sole responsibility for this article. We also extend our gratitude to the research participants who answered our survey and interview questions. In sharing their experiences, they provided us with the primary data we needed to meet the study goals.

Corresponding author contact information

Dr. Ilana Horwitz, 7031 Freret St, New Orleans LA 70118;
ihorwitz@tulane.edu