Oct 1, 2009 - Journals. Cancer Nursing Practice · Emergency Nurse · Evidence-Based Nursing · Learning Disability Practice · Mental Health Practice · Nurse ...
grounded theory
Using survey data to assist theoretical sampling in grounded theory research Kay Currie looks at ways to use theoretical sampling to help select interview participants for analyses using grounded theory Abstract Grounded theory methods can help to generate a theoretical explanation of social processes in specific contexts. This paper discusses theoretical sampling to guide interview participant selection and reflects on the practical application of this procedure to outline the value and limitations of using survey
▲ ▲ ▲ ▲
key words
data in this way.
grounded theory methods survey data decision trail sampling
Grounded theory methods are applied in nursing research to explore topic areas where little is previously known, particularly where there is a requirement to investigate social interactional processes at play in a specific context (Happ and Kagan 2001). Although a divergence in approach has emerged (Glaser 1992, Melia 1996), the main components of the method remain broadly similar. These include: n Constant comparative analysis. n Generating codes and categories from data. n Theoretical sensitivity. n Theoretical sampling. n Theory development (Glaser and Strauss 1967). While all components deserve critical attention, this paper will focus primarily on theoretical sampling. The aim is to describe and evaluate the experi24 NURSERESEARCHER 2009, 17, 1
grounded theory
ence of using survey data to identify participants for interview, based on theoretical sampling decisions. Principles of theoretical sampling Glaser and Strauss (1967) defined theoretical sampling as ‘the process of generating theory whereby the analyst jointly collects, codes and analyses his data and decides which data to collect next and where to find them in order to develop his theory as it emerges’. Thus, in the early stages of grounded theory, purposeful sampling is used (Cutcliffe 2000, McCann and Clark 2003) because the researcher begins with the conscious selection of certain subjects who can readily articulate their experience of the area under investigation (Burns and Grove 2001). As the study proceeds, theoretical sampling, selecting specific participants or other data sources, and directing interviews or alternative data collection strategies based on emerging theoretical concepts, are adopted (Glaser 1978). By using theoretical sampling in grounded theory, it is possible only to determine the size or nature of the sample retrospectively; sample choice is a function of the emerging hypothesis, and sample size a function of theoretical completeness (Cutcliffe 2000). However, depending on emerging issues, this sampling can be a potential limitation of grounded theory, with the researcher reliant on the ability of early participants to identify relevant aspects of the area under investigation – Cutcliffe (2000) describes the first interviewee as the ‘gatekeeper’ who determines the direction of subsequent sampling. Decisions relating to the breadth or narrowness of the sample and sampling to maximise or minimise comparisons also require consideration. Glaser and Strauss (1967) note that minimising differences in comparison groups helps to bring out the properties of categories and recommend that this is done before moving on to maximise the difference in comparison groups. To demonstrate these issues and the value and limitations of using survey data in this way, I will reflect on my PhD study (Currie 2006), which explored graduate specialist practitioner engagement in clinical practice development. Planning data collection At data collection planning, the 227 nurses who had graduated from the NURSERESEARCHER 2009, 17, 1
25
grounded theory
BSc specialist nursing programme at the host university served as the study population group. Ethical approval for the research was granted by the appropriate university committees. I wanted to generate a grounded theory of graduate specialist practitioner engagement in practice development, so using face-to-face interviews to seek information directly from graduates was appropriate. I used a questionnaire as a preliminary device to identify potential interviewees. Before further describing and evaluating the use of questionnaire data in this context, I will review the place of survey in grounded theory. The use of survey in grounded theory research To explore the reported use of survey data in grounded theory nursing research, I undertook a literature search. Of the 24 retrieved articles, only one study (Gunnarsdóttir and Björnsdóttir 2003) appears to have integrated survey findings as a data source in a well developed grounded theory study. The majority of the articles used mixed methods, blending survey data and a modified grounded theory approach to analysing qualitative data to produce interesting findings in a range of topic areas, rather than generating pure grounded theory (Curtis et al 2006, Russell et al 2006). Two articles (Raj et al 2005, Mishna 2004) used a survey tool to identify participants who had the primary characteristic under investigation. Both studies provided quantitative data, which was incorporated into the discussion, and located potential interview participants at the beginning of the qualitative phase of the study. However, in contrast to the study of specialist practitioners reported in this paper, neither Raj et al nor Mishna subsequently returned to their survey data to guide additional sample group selection based on evolving theoretical sampling criteria. Only Mactavish and Schleien (2004) reported using a ‘sequential-purposive sampling technique’ that appears comparable to the approach described below. The authors interviewed 16 families of children with a developmental disability. Eight families were selected according to pre-determined criteria, and eight selected ‘based on consistencies between their questionnaire responses and emerging interpretations of the interview data (that is, a theory based purposive technique)’. 26 NURSERESEARCHER 2009, 17, 1
grounded theory
Based on the findings of this literature review, using survey data to identify interview participants by applying theoretical sampling principles can be interpreted as a relatively innovative approach. Development of the survey questionnaire Glaser and Strauss (1967) note that although samples for data collection are not determined by a preconceived theoretical framework, the researcher may begin by designating a few ‘principal features’ of the situation they will study, based on their existing knowledge of the subject. My experience as a nurse and an educator suggested that certain characteristics might enable me to distinguish between respondents with respect to their engagement in clinical practice development. These were translated into a set of ‘principal features’ (Box 1) and incorporated into a postal questionnaire. The questionnaire items required respondents to tick a box which most closely corresponded to their experiences, with an opportunity to add comment if desired. Each item was assigned a numerical or alphabetical code, and transferred onto a computerised database (Statistical Package for Social Sciences), where potential interviewees were identified through their reported characteristics. To distinguish further between respondents, a score was allocated to each type of practice development activity, acting as a crude judgemental indicator of the overall level of activity. The questionnaire was piloted by nursing colleagues and students to evaluate the face validity of the contents and confirm that the questions were easily interpreted. Survey response In March 2004, the questionnaire and a stamped addressed envelope was posted in a single batch to all graduates listed on the BSc specialist nursing programme database from the years 1999-2003. A response rate of 45 per cent (n=102) was realised. This was acceptable as they showed a reasonable spread across all five years of graduate cohorts and a range of responses across the other characteristics explored by the questionnaire. This meant that it would be possible to identify potential interview participants NURSERESEARCHER 2009, 17, 1
27
grounded theory
Box 1. Principal features explored by survey questionnaire Clinical specialty the respondent worked in Length of experience in specialty prior to starting the BSc programme Clinical grade on starting programme Clinical grade at time of participating in the study Time lapsed since graduation Job moves since graduation (moved area or promotion) Type of practice development activity involved in: a) Searching for literature on a practice-based topic b) Arranging to share research findings with colleagues c) Initiating change in practice, based on evidence d) Developing clinical standards or protocols e) Initiating clinical audits (rather than doing audit for a doctor or manager) f) Being involved in developing research projects g) Developing teaching packs and programmes in the area h) Taking on a formal practice education role (beyond normal mentoring) i) Taking a leadership role, responsible for developing clinical care j) Any other type of activity you consider practice development n Self-described frequency of involvement in practice development n Self-described satisfaction with current level of involvement in practice development n n n n n n n
along a continuum of level of engagement in practice development, providing an opportunity for sampling to minimise and maximise comparisons of experience and other characteristics (Glaser and Strauss 1967, Cutcliffe 2000). Influence of theoretical sampling on data collection Having considered the need, or otherwise, for a guiding theoretical framework in the study, I decided that symbolic interactionist thinking was relevant (Benzies and Allen 2001, Crooks 2001, Jeon 2004). Used sensitively, it would allow elements of self, interaction, context and action to be explored, while still maintaining the flexibility to attend to the concerns of participants, rather than forcing data based on any preconceived framework. Thus, while an interview guide (Box 2) provided a framework for questioning, there was ample opportunity for the concerns of the participant to emerge and be explored. Interviews were conducted from June 2004 to January 2006. The first interview was with a locally accessible participant who was active in practice 28 NURSERESEARCHER 2009, 17, 1
grounded theory
development and who agreed to assist in debriefing. Thereafter, constant comparative analysis was used, with theoretical sampling supplanting the early purposive approach. According to Glaser and Strauss (1967), theoretical sampling begins by asking, ‘What group should be turned to next in data collection, and for what theoretical purpose?’ These questions were used to guide this study and the characteristics of successive sample groups were determined by their theoretical relevance at a particular stage in the research, rather than any attempt to ensure complete coverage of comparisons of all possible characteristics. Thus, Glaser and Strauss’s criteria of theoretical purpose and relevance were applied in determining sequential participants. Auditing of grounded theory requires the researcher to specify how and why participants were selected for inclusion (Chiovitti and Piran 2003), and to identify the influence of the researcher in participant recruitment (Neill 2006). This allows the reader to track the relationship between data sources and interpretations, demonstrating that data has undergone analysis, reduction and synthesis (Wolf 2003). A detailed audit trail was included in my thesis. Originally, I had intended to explore the experiences of graduates with respect to engagement in clinical practice development, anticipating two broad patterns of experience in graduates. However when data collection was complete, I had identified eight theoretical sub-groups of participants (Table 1). After 20 interviews, no new concepts were emerging and I was confident that the categories and developing theory were reflective of the experience and actions of graduates who had become actively engaged in clinical practice development. It would have been useful to have been able to interview additional graduates who had not actively engaged in practice development, Box 2. Interview guide based on symbolic interactionism framework Initial interview questions focused on exploring the following three areas: n Aspects of participants’ understanding of self and their roles before and after completing their degree. n The participants’ interpretation of the influence of external context and interactions with colleagues and managers on their practice development activities. n The participants’ decisions about action relating to practice development.
NURSERESEARCHER 2009, 17, 1
29
promoted into leadership roles.
Active in practice development;
(n=7)
Group 2
30 NURSERESEARCHER 2009, 17, 1
Active in practice development;
Random selection to explore emerging theory.
Group 8
(n=1)
Lowest scoring participants.
Personal journey; fit with emerging theory.
(Two invited, unable to recruit).
Personal journey; being blocked; constraining context.
relation to educational developments.
promoted into leadership role since graduation.
Not active, not promoted since graduation.
Personal journey; demonstrates ‘acting in the bigger picture’, particularly in
of persistence; being blocked; constraining context.
Personal journey; legitimate leadership; why have they not been active; lack
developing as an expert; moving on; acting in the bigger picture.
Legitimate leadership; getting the ticket; needing knowledge and confidence;
enabled; supportive manager; process of engagement.
Personal journey; why have they not had to ‘escape constraints’; being
practice; making a difference to patients; gaining respect.
Personal journey; being blocked; escaping constraints; trying it out in
Theoretical sampling: key concepts being explored/emerging
Not active in practice development;
started in leadership role (F/G).
Not active in practice development;
started in leadership role.
Group 7
(n=1)
Group 6
(10 invited)
(n=1)
Group 5
(n=4)
Group 4
(n=4)
Group 3
still in non-promoted posts.
Active in practice development;
Group 1
(n=2)
Theoretical selection criteria (from survey)
Group
Table 1. Theoretical group characteristics
grounded theory
grounded theory
including those who were in promoted roles and those who had not been promoted since graduation. Unfortunately, despite posting out repeat invites to several potential participants, no further volunteers for interview came forward, and thus this aspect of the developed theory remains incomplete. Reflections on using survey data in theoretical sampling The approach to participant recruitment had strengths and weakness. Using a structured questionnaire rather than entering the field to seek out participants is not described by Glaser, and is not common in grounded theory. However, while this approach may be considered a transgression from traditional grounded theory methods (Cutcliffe 2005), it proved to be a practical and effective aid to theoretical sampling. Descriptive analysis of the questionnaire results indicated that the characteristics of the respondents covered the spectrum of possible graduate features. However, participant selection was limited in three ways. First, while identifying the type of practice development activities participants were involved in was useful, the allocated ‘total practice development activity score’ was too crude to discriminate between high and low levels of engagement. For example, one interviewee, selected because his total score suggested he had relatively low levels of practice development activity, described significant engagement. Thereafter, specific activities were scrutinised to select interviewees rather than the total practice development score alone. This initial reliance on the total practice development activity score meant that additional participants who would have met the criteria for theoretical sampling at a particular stage were not identified. However, this did not affect the outcomes of the study as theoretical saturation was achieved for all categories except for those who had low levels of engagement. Second, while questionnaire responses provided a spectrum of practice development activity, those who were less actively involved tended to decline or not respond to the invitation to be interviewed. The concept of ‘giving in’ in the face of obstruction tentatively emerged from the data, but further details of the dimensions of that response could not be elucidated. The resultant theory therefore provides a portrayal of ‘successful’ graduate specialist practitioners, but an incomplete picture of the experience of those less actively engaged in NURSERESEARCHER 2009, 17, 1
31
grounded theory
practice development. This is a limitation of the study, rather than a limitation of the use of the survey per se, as those who were not active in practice development provided valuable data through the questionnaire. Third, this study achieved its objective to investigate the perspective of graduate specialist practitioners. But, in terms of understanding the social context and interactional processes of their engagement in practice development more fully, the study may be limited by not sampling to incorporate the perspectives of peers and managers. In essence, I have studied the individual, rather than what Schmitt (2005) would call the collective. Conversely, focusing on the personal experience of graduate specialist practitioners may have enabled a greater depth of understanding of the individual perspective, leading to the emergence of a theory of identity transformation, which may not have been possible by continuing to apply theoretical sampling principles in a broader investigation. Conclusions This paper has illustrated the process of theoretical sampling assisted by survey data. The questionnaire was designed to provide information on a range of possible characteristics of graduate specialist practitioners in relation to their practice development activities. It was not a particularly sophisticated tool, and the emergent theory did not require more than simple descriptive analysis of the presenting characteristics. Although it has limitations – relating primarily to the lack of validity of the total practice development score and the challenges of recruiting reluctant participants – the advantage of being able to consecutively select individuals from a sample of 102 questionnaire respondents meant the survey proved a useful aid for theoretical sampling of interview participants. I would encourage future application of this approach and further debate as to its relative merits as a means to advance scholarship in grounded theory research n Kay Currie PhD, MN, BSc, RN, RNT works in the Department of Acute Nursing in the School of Health at Glasgow Caledonian University, Glasgow, UK
This article has been subject to double-blind review
32 NURSERESEARCHER 2009, 17, 1
grounded theory
references Benzies KM, Allen MN (2001) Symbolic interactionism as a theoretical perspective for multiple method research. Journal of Advanced Nursing. 33, 4, 541-547. Burns N, Grove SK (2001) Qualitative research methodology. In: Burns N, Grove SK The Practice of Nursing Research; Conduct, Critique and Utilization. Fourth Edition. WB Saunders, Philadelphia PA, 591-622. Chiovitti RF, Piran N (2003) Rigour and grounded theory research. Journal of Advanced Nursing. 44, 4, 427-435. Crooks DL (2001) The importance of symbolic interaction in grounded theory research on women’s health. Health Care for Women International. 22, 1-2, 11-27. Currie K (2006) Making a Difference: A Grounded Theory of Identity Transformation In The Process of Graduate Specialist Practitioner Engagement In Practice Development. PhD Thesis. Caledonian University, Glasgow. Curtis P, Kirkham M, Ball L (2006) Management and morale: challenges in contemporary maternity care. British Journal of Midwifery. 14, 2, 100-103. Cutcliffe JR (2000) Methodological issues in grounded theory. Journal of Advanced Nursing. 31, 6, 1476-1484. Cutcliffe JR (2005) Adapt or adopt: developing and transgressing the methodological boundaries of grounded theory. Journal of Advanced Nursing. 51, 4, 421-428. Glaser BG (1978) Theoretical Sensitivity: Advances in the Methodology of Grounded Theory. Sociology Press, Mill Valley CA. Glaser BG (1992) Emergence vs Forcing: Basics of Grounded Theory Analysis. Sociology Press, Mill Valley CA. Glaser BG, Strauss AL (1967) The Discovery of Grounded Theory: Strategies For Qualitative Research. Aldine Publishing, New York NY.
setting. Scandinavian Journal of Caring Sciences. 17, 1, 66-73. Happ MB, Kagan SH (2001) Methodological considerations for grounded theory research in critical care settings. Nursing Research. 50, 3,188192. Jeon YH (2004) The application of grounded theory and symbolic interactionism. Scandinavian Journal of Caring Sciences. 18, 3, 249-256. Mactavish JB, Schleien SJ (2004) Re-injecting spontaneity and balance in family life: parents’ perspectives on recreation in families that include children with developmental disability. Journal of Intellectual Disability. 48, 2, 123-141. McCann TV, Clark E (2003) Grounded theory in nursing research: part 1 – methodology. Nurse Researcher. 11, 2, 7-18. Melia KM (1996) Rediscovering Glaser. Qualitative Health Research. 6, 3, 368-378. Mishna F (2004) A qualitative study of bullying from multiple perspectives. Children and Schools. 26, 4, 234-47. Neill SJ (2006) Grounded theory sampling: the contribution of reflexivity. Journal of Research in Nursing. 11, 3, 253-260. Raj A, Silverman JG, McCleary-Sills J et al (2005) Immigration policies increase South Asian immigrant women’s vulnerability to intimate partner violence. Journal of the American Medical Women’s Association. 60, 1, 26-32. Russell L, Nyhof-Young J, Abosh B et al (2006) An exploratory analysis of an interprofessional learning environment in two hospital clinical teaching units. Journal of Interprofessional Care. 20, 1, 29-39. Schmitt N (2005) Role transition from caregiver to case manager, part 1. Lippincott’s Case Management. 10, 6, 294-302. Wolf ZR (2003) Exploring the audit trail for qualitative investigations. Nurse Educator. 28, 4, 175-178.
Gunnarsdóttir S, Björnsdóttir K (2003) Health promotion in the workplace: the perspective of unskilled workers in a hospital
NURSERESEARCHER 2009, 17, 1
33