Mixed Methods Sampling Procedures: Techniques, Methodological Problems, and Meta-Inferences |
|
| Chair | Dr Andrea Hense (Sociological Research Institute Göttingen (SOFI) ) |
| Coordinator 1 | Dr Leila Akremi (TU Berlin) |
Mixed-methods research on health-related issues with heterosexual couples raises several ethical and methodological challenges regarding sampling and data analysis. The decision about whom to interview is a major factor influencing the recruitment process and the data obtained, conditioning the knowledge researchers can produce. However, literature grounded on researchers’ experiences in this topic is scarce.
This study aims to discuss methodological problems regarding sampling procedures when interviewing parental couples together about their experiences in Neonatal Intensive Care Units (NICU) and the effects of a very preterm delivery on parental quality of life. Such discussion is informed by fieldwork conducted within a mixed-methods study, relying on the results obtained from: 1) ethnographic observation, carried out in a level III NICU located in the North of Portugal, during 6 months; 2) NICU-based surveys with 121 mothers and 91 fathers (83 couples) of very preterm infants admitted at the seven public level III NICU of the Northern Health Region of Portugal, inquired by trained interviewers 15 to 22 days after birth (July 2013 - June 2014); and 3) 27 couple semi-structured interviews with a subsample of these parents, four months after hospitalization (November 2013 - July 2014).
Interviewees were purposively sampled to include 10 parental couples of extremely low birth weight infants (<1000g) and 17 parental couples of non-extremely low birth weight infants (≥1000g), in accordance with the results obtained during ethnography and the NICU-based survey regarding the influence of infant’s birth weight on parents’ experiences and quality of life. Heterogeneity sampling was used to obtain maximum variation of views and experiences until reaching thematic saturation.
However, health professionals suggested that gestational age was the best evidence-based indicator of infant’s vulnerability. In fact, the narratives emerging from content analysis were determined by father’s occupation: 1) ‘emotionally-driven narrative’ enacted by fathers employed as health or teaching professionals, who talked about their experiences in depth, sharing an intense emotional repertoire with mothers; 2) ‘control-need narrative’ enacted by fathers working in business, civil construction or armed forces, who tended to control the interview, both by leading the couple’s narrative and questioning the study aims and its implications for clinical practice; and 3) ‘emotionally-silenced narrative’ enacted by fathers working as craftsmen or providing clerical support, who tended to silence their intimate experiences and emotions. The influence of occupation was a buffer in mother’s narratives. Mothers tended to provide detailed answers to questions about parental roles and their discourses revealed the reproduction of intensive motherhood and gender stereotypes according to which women are mostly guided by emotions.
Research methods training programs must elicit empirically grounded discussions about criteria for sampling and the potential effects of couple interviews in mixed-methods research dependability and transferability. Sampling strategies and their implication on data analysis in mixed-methods studies that consider the couple as the unit of analysis need to be systematically discussed in the literature.
Mixed methods research is fitting for engaging with vulnerable populations. Complex inequities exist for families living with low income, affecting social relationships, well-being, and mental health. This explanatory sequential mixed methods study was grounded in the philosophical perspectives of Dewey's pragmatism, and Bronfenbrenner's bioecological model of human development. A two-generation preschool program in Calgary, Alberta, Canada, has been ongoing since 2001. Program objectives include improving early childhood development and school readiness through strengthening children’s environmental influences. Program eligibility includes living with low-income, and mental illness, addiction, or social isolation. Study eligibility included enrollment in the program between 2002 and 2008, for a minimum of 3 consecutive months.
The aim of this study was to develop an understanding of the change processes in families that attended the preschool program. Mixed methods sampling strategies were used to understand how children’s developmental outcomes affected mother’s experiences, and/or how mother’s experiences in the program changed children’s environmental influences. Phase I (quantitative) included data collected at program intake (age 2.5 to 5 years), exit (age 6 years), and age 10 years, informing participant selection for Phase II (qualitative). In Phase I, children’s receptive language scores were identified, using the Peabody Picture Vocabulary Test 3rd Edition (PPVT-III). In Phase II, Charmazian constructivist grounded theory was used to explore experiences of mothers as they moved through CUPS One World.
The initial sample consisted of 134 children (and caregivers) and to ensure wide-ranging participant experiences in the interviews, maximum variation sampling was used. Six groups of children were identified in Phase I: (a) children in the top and/or bottom 25th percentile of receptive language intake scores who exited the study before age 10 years; (b) children in the top and/or bottom 25th percentile of receptive language change scores between intake (age 2.5 to 5 years) and exit (age 6 years); and (c) children in the top and/or bottom 25th percentile of receptive language change scores between program exit and age 10 years. Caregivers with children in the top and/or bottom 25th percentile of PPVT-III change scores or intake scores, were contacted for interviews. Eligible caregivers were biological mothers, and interviews were completed with 14 of them. Initial sampling goals were to interview two to four mothers from each of the six groups, however, only one mother of a child whose score fell in the top 25th percentile of scores and exited the study early consented to an interview. One from the top, and two from the bottom 25th percentile of PPVT-III intake scores of those exiting early; (b) three from the top, and three from the bottom 25th percentile of PPVT-III change scores from intake to exit; and (c) two from the top, and three from the bottom 25th percentile of PPVT-III change scores from exit to age 10. Integration occurred in design and method, and the grounded theory "Stepping Stones to Resiliency" emerged.
Although the term sampling is used in qualitative research, in almost all cases, the sampling is not a result of probabilistic theory but invoked from theoretical or empirical grounds. It is common to distinguish between two approaches for sampling in qualitative research designs: (1) theoretical sampling, which is based on grounded theory, where researchers choose their cases inductively as the research proceeds; (2) purposive sampling which relies on existing theoretical or empirical knowledge that leads to the research questions. There are numerous purposive sampling techniques taking the forms of typical cases, outlier sampling, and min/max variation sampling. One of the strengths of mixed method design is the use of multiple methods to define the question or phenomenon, to measure or test its extent, and to study the question in-depth. In a mixed-method design where qualitative analysis is followed by a phenomenon that can be represented quantitatively, often referred to as sequential explanatory design, researchers seek to define the cases based on previous findings while looking to understand the findings’ significance and validity. This design is a typical case for purposive sampling techniques.
These methodological issues identified during research project, aim to understand why undergraduate students in Israel choose to cross the disciplinary borders in higher education by studying for double-major degrees, pairing fields of study of varying disciplines. The study used mixed method with explanatory design. The first part used quantitative methods to define the disciplinary structure of higher education; it was based on a dataset provided by the Israeli Central Bureau of Statistics including all undergraduate students pairing fields of study between 2001 and 2010 (N=761,167). Based on these data, a Multidimensional Scaling analysis was used to identify clusters of fields and to map the structure of double-major combinations in Israel. The MDS analysis tends to locate spatially close, frequently joined fields, and less frequently joined fields, physically distant. The MDS map allows exposing typical combinations of fields and the outlier (rare and extreme) combinations of fields. The results of the quantitative analysis were used to draw a purposive sample of 30 students from different degree years that span the disciplinary borders, for in-depth interviews. A message about the study’s aim and the target population was sent via students' networks and social media, direct call in the classes, and by snowball sampling. The semi-structured interview had three main themes: (1) motivations for crossing the disciplinary borders, (2) the difficulties experienced, and perceived advantages of their choices, and (3) their expectations for post-graduation. All interviews were coded and thematically analyzed by content analysis.