Understanding delayed access to antenatal care: a qualitative study Rosalind Haddrill

3.3 Considerations of bias and quality

3.3.1 Bias

As Polit and Beck (2012) identify, in any study, whether qualitative or quantitative, it is the researcher’s role to:

• 
  Use a variety of strategies to eliminate/minimise bias and strengthen study rigour as far as possible;
  
• 
  To establish mechanisms to detect and/or measure bias, and
  
• 
  To take known biases into consideration when interpreting any findings;
  

and the reader’s role to scrutinise and draw conclusions about whether bias exists and if it undermines the findings.
Careful consideration of appropriate methods for all aspects of the study, from sampling and data collection to analysis, was undertaken by the members of the research team at the outset of the project, using accepted protocols for research design, ethics and governance, to avoid design bias and systematic errors. The design and methods used were monitored and adapted throughout, as part of the project management and supervisory process, involving all members of the research team, to respond to possible bias and ensure the study methods were adequate to answer the research question. Purposive and theoretical sampling methods, and the use of triangulation, aimed to ensure comprehensiveness and reduce the potential for sample bias, whilst acknowledging the boundaries of the qualitative approach.
Recognition of the potential for interviewer bias has influenced my approach to interviews. I have attempted to convey a neutral and non-judgemental approach, using interviewing skills and empathy. The development and application of the interview guide for the study tried to avoid leading or biased questions and probing, and to be responsive to the participants. This and the naturalistic setting for the study aimed to encourage the participants to be as open and honest as possible, and to reduce the potential for reactive (Hawthorne) effects and social desirability bias. Being open with the participants, assuring them of anonymity, and not being directly involved in the care of the women interviewed, have also helped to reduce the potential for lack of candour. In addition, interviewing women as close to their booking appointment aimed to reduce the likelihood of recall bias (Bowling 2009, Tod 2010, Polit and Beck 2012).
I have acknowledged that my own subjectivity, based on my own experiences and expectations, has the potential to distort the findings of the study. Consequently, the recognition of personal biases, beliefs and assumptions, and if and when they have intruded into data collection and analysis, have formed a critical part of considerations of reflexivity, to be discussed further in the following section. Corbin and Strauss (2008) stress the importance of establishing enough distance between the interviewer and the research, in order to be able to think clearly and analytically, and to the need to reflect on and question everything, which I have attempted to do.

3.3.2 Quality

There is much discussion of the relevance and adaptation of quality criteria for qualitative research. Many authors have argued that qualitative research, with its naturalistic approach, cannot and should not be judged by quantitative measures of validity, generalisability and reliability, though there is an acknowledgement that the application of such measures to research does not necessarily presuppose a positivist approach. Several authors have suggested that some definitions of rigour do not fit with an interpretive approach that values insight and creativity, and that qualitative research can be considered both an art and a science. This needs to be balanced with the consideration that the rejection of reliability and validity could suggest a lack of concern with rigour, which is naturally undesirable. They argue that there is a need for a balance between creativity and insightfulness and scientific excellence, producing research which is descriptively sound and explicit whilst also interpretively rich and innovative (Polit and Beck 2012).
Some authors believe that though there is no definitive set of quality criteria that can (or should) be rigidly applied to qualitative research, some quality criteria are fairly universal (Whittlemore, Chase and Mandle 2001). As Murphy et al (1998) observe, the hallmarks of high quality qualitative research, namely a commitment to rigour, clarity and systematicity, are the hallmarks of all good research. However, despite some broad similarities in purpose, the fundamental differences in the research goals and the knowledge each approach generates require that quality is assessed differently (Popay, Rogers and Williams 1998, Bryman 2008). As Spencer et al (2003) comment:

“the concerns which lie behind customary conceptions of quality have relevance for qualitative enquiry but need to be reformulated – and assessed quite differently – within the domain of qualitative research… qualitative research should be assessed on its ‘own terms’ within premises that are central to its purpose, nature and conduct” (p4).

Kisely and Kendall (2011) suggest rather that quantitative concepts such as reliability, validity, statistical power, bias and generalisability have qualitative equivalents such as triangulation, trustworthiness, saturation, reflexivity and applicability, which need to be explicit in any study.
The most cited quality criteria for qualitative research, Lincoln and Guba’s framework for developing ‘trustworthiness’ (Lincoln and Guba 1985, Guba and Lincoln 1994): identified five criteria, with parallels to standards of validity and reliability in quantitative research. These are:

Credibility: ensuring confidence in the truth of the data and any interpretations: carrying out study using methods to enhance believability, and demonstrating credibility in research reports;

Dependability: demonstrating the stability and reliability of the data over time: ensuring the study could be replicated;

Confirmability: giving a fair representation of the information and interpretations (not invented). Accurately reflecting the participants’ voices and the context of the study, not the researcher’s bias;

Transferability: the extent to which findings can be transferred: providing enough data for readers to evaluate the applicability, for example using ‘thick description’;

Authenticity: fairly and honestly showing a range of realities, reflecting participants lives, for example in terms of tone, mood and/or experience. (cited in Polit and Beck 2012: 584-585).

These are reflected in qualitative research appraisal tools such as CASP (2013), as used in chapter two. Several authors have reinterpreted Lincoln and Guba’s framework. Some argue that, there are two fundamental criteria, validity and relevance, which apply to both qualitative and quantitative research, but using different methods (Murphy et al 1998, Mays and Pope 2000, Malterud 2001b). Similarly, Whittlemore, Chase and Mandle’s (2001) synthesis of ten quality guidelines (including Lincoln and Guba) identifies validity as the overarching goal, including credibility and authenticity, but adding the need for criticality (critical appraisal of key decisions and self-reflection during the research) and integrity (on-going scrutiny of the validity of interpretations, ensuring these are grounded in the data). They provide supplementary benchmarks for quality relating particularly to explicitness and thoroughness, but also to vividness and creativity, many of which overlap with Lincoln and Guba.

Polit and Beck (2012) argue that such criteria may offer standards for readers to assess quality, rather than as a guide to researchers of how to do, but that the responsibility should rest with researchers to demonstrate the consideration of quality measures used throughout their research, providing evidence of rigour, integrity and trustworthiness. Mays and Pope (2000) agree with this comprehensive approach:

“as in quantitative research, the basic strategy to ensure rigour, and thus quality, in qualitative research is systematic, self-conscious research design, data collection, interpretation and communication” (p52)

This consideration is reflected in four key methods used: a commitment to multiple perspectives, triangulation, clarity of presentation and reflexivity; and an overall consideration of the relevance of the research, which are considered below.
3.3.2.1 Commitment to multiple perspectives

The study shows a commitment to the emic perspective, incorporating the views and experiences of a range of participants, contributing to the credibility and authenticity of the research. Several authors talk about the need for this ‘even handedness’ and ‘fair dealing’, to ensure that the research contains a wide range of different perspectives, so that the viewpoint of one group is never represented as ‘the truth’ (Murphy et al 1998, Mays and Pope 2000). The research attempts to present different sides of the discussion, including inconsistent and disconfirming evidence, which challenges accepted reasons for late booking and attitudes towards this (Polit and Beck 2012).

3.3.2.2 Triangulation

The use of multiple approaches is characteristic of qualitative research, and can be seen as a major strength, identifying patterns of convergence, enhancing the comprehensiveness of the findings and stimulating reflexive analysis (Mays and Pope 2000, Polit and Beck 2012). As Murphy et al (1998) identify, “emphasis is on counter-balancing the distorting effects of any single approach… the aim is to establish the convergent validity of findings drawn from complementary approaches” (P11). Data and investigator triangulation are key to authenticity and were achieved through the comparison of results from multiple data sources alongside independent reviews of the transcripts and themes developed during the analytical process, as part of the supervisory process, bringing a complementary blend of different perspectives. Such methods were used to examine patterns of convergence and to obtain a comprehensive understanding of the phenomenon of late booking, reaching a point of saturation, as discussed previously, and to reduce the risk of bias in the research (Polit and Beck 2012). However, the aim was to maintain a commitment to the context and complexity of the data, acknowledging the value of different contexts, rather than aiming for a single ‘master reality’ (Murphy et al 1998), an overall average where weaknesses or gaps in one set of findings might be balanced by strengths or areas of focus in another. There was also an acknowledgement that findings from different sources might corroborate and complement, rather than refute, each other.

3.3.2.3 Clarity of presentation

The clear exposition of methods and findings is critical to the dependability and confirmability of any study; Whittlemore, Chase and Mandle (2001) consider it key to a study’s integrity. I have sought to a provide a clear exposition of the study methods and the appropriateness of these for the study’s aims, at all stages from recruitment, interviewing and transcription, to data analysis and the presentation of the findings, supported with evidence from previous research and literature (Liamputtong and Ezzy 2005, Kisely and Kendall 2011). The systematic approach adopted seeks to minimise any anecdotal interpretation of the data, and a detailed consideration of the language and presentation of the findings forms a key part of the analysis and interpretation. The complexities of the process and the challenges involved are articulated.

A clear description of how the development of a simple system of codes evolves into a more sophisticated structure of themes and subthemes is provided by the analytical process (Mays and Pope 2000). A credible account and an accurate representation of the phenomenon of late booking is given, attempting to provide a good match between the data presented and the themes and theories developed in the study, and displaying enough data to allow the reader to judge these (Hammersley 1992). Geertz’s ‘thick description’ encapsulates this idea: a comprehensive and vivid picture of the participants, their experiences and context, using verbatim quotes and lucid descriptions, is presented (Mays and Pope 2006, Kisely and Kendall 2011, Polit and Beck 2012). However, as Sandelowski and Leeman (2012) identify, description should be ‘thick’ not ‘heaped’ if qualitative research findings are to be understandable and usable. In practical terms, adherence to recognised research governance protocols, and particularly maintenance of a study site file, provide a clear and auditable trail for the project.

3.3.2.4 Reflexivity

All researchers must ensure that they do not overtly allow personal values or theoretical inclinations to bias the conduct of the research and its findings. However, reflexivity, an essential component of credibility and confirmability in qualitative research, requires a sensitivity to the ways in which the researcher and research process and strategies have shaped the data, including prior assumptions and experience, personal and intellectual biases. It recognises the potential for professional status to influence the data and the ‘distance’ between the researcher and the researched (Mays and Pope 2000, Green and Thorogood 2014).

I recognise that reflexivity has been an integral part of, and my personal connection to, the study. Evidence of the ways in which the data have been shaped by my presence and background is part of the transparency of the study. Theoretical sensitivity forms a positive part of this reflexive approach. My knowledge and understanding of the subject of late booking and the participants’ world, from a theoretical perspective and professional experience, have informed the interviews and enhanced the analytical process. It influenced my ability “to see the research situation and its associated data in new ways, and to explore the data’s potential for developing theory” (Strauss and Corbin 1990:44). My personal and professional biases have been made explicit at the outset of the research, and I have attempted to be open and honest about the relationship between myself and the participants (Liamputtong and Ezzy 2005, Mays and Pope 2006). In method terms a balance has been sought between acknowledging my influence as a midwife and creating a research environment which was ‘value neutral’, to allow the participants to speak for themselves (Fox et al 2001). In terms of ‘situating myself’ in the data, all participants were made aware of my dual status as both midwife and researcher but I was not involved in the clinical care of the women.
Whittlemore, Chase and Mandle’s (2001)’s criteria of ‘criticality’ suggests rigour may be demonstrated through a critical approach adopted to research, exploring but also challenging taken for granted assumptions, for example about late booking. The critical approach has become part of the reflexive process, and an awareness that my identity and values, alongside the multidisciplinary influences from within the research team, inevitably influence any interpretation of the lives of the women, and the co-construction of the findings (Polit and Beck 2012). Critical thinking in relation to the study has involved situating the data within the context of the existing literature, and particularly the themes identified in the synthesis. My reflection has included a consideration of the status of the women in the study and their choices in relation to pregnancy; their assessment of risk and the relevance of care. Throughout the research there has also been a professional consideration (and reconsideration) on my part of the necessity and value of early antenatal care.
3.3.2.5 Relevance

The concept of relevance considers the extent to which research adds to existing knowledge about late booking, but is also linked to Lincoln and Guba’s transferability, the extent to which findings, though not statistically representative, can be generalised beyond the study circumstances, in empirical or theoretical terms (Murphy et al 1998). The interpretivist and constructivist thinking behind the study assumes a changing world socially and a multiplicity of ‘truths’, which inevitably influence the replication of any findings. In contrast to positivist or quantitative approaches, the study used sampling methods aimed at maximising diversity and which were contextually unique. As a result no real claims have been made about the generalisability of the findings to other settings. However, this is balanced with consideration of the purpose the study, which seeks in part to influence practice and policy generally, through new knowledge with the potential for wider applications (Fox et al 2001, Spencer et al 2003). The study aims for authenticity, representing the participants fairly and ethically, with the goals of producing findings with relevance and significance for both pregnant women and practitioners, to improve understanding and encourage change (Yardley 2000). To achieve this, the research has been presented with sufficient range and detail for the reader to judge if it can be applied elsewhere in similar settings; the rich accounts of ‘thick description’ and clarity of presentation considered previously. Also the use of theoretical sampling has ensured that the sample includes a full range of perspectives and contexts relevant to late booking, to consider as many factors as possible that might affect behaviour (Mays and Pope 2000, Liamputtong and Ezzy 2005, Mays and Pope 2006, Kisely and Kendall 2011).