gms | German Medical Science

The combination of qualitative and quantitative research using what is known as the mixed methods approach has become increasingly important in recent years [1]. Internationally, this approach is used particularly frequently in the social sciences and health care [1]. In Germany, too, there is a lively debate, including different research fields, about the use of this methodology and method [1], [2], [8], [13], [18]. Midwifery science, which can be described as a relatively new discipline in Germany ([23] p.83), draws on other academic disciplines such as the social sciences, psychology, nursing sciences, health science, medicine and also biology. According to Kahl, researchers in the field of midwifery thus utilise the “portfolio of methods from adjacent sciences to study their subject areas” ([11] p.194). To do so, researchers in midwifery science principally employ the fundamental instruments of both quantitative and qualitative methods from empirical social research. However, the combination of these two techniques in the form of the mixed methods approach has so far received very little attention in this research field in German-speaking countries. A handsearch in the relevant German-language specialist databases did not reveal any research projects using this approach. Yet, according to Kuckartz, the mixed methods approach which integrates quantitative and qualitative data in one research context represents “a new contemporary understanding of methods” ([13] p.29), one which brings both advantages and challenges. This paper will illustrate the potential of mixed methods for midwifery science. It will do so by introducing the basic principles of planning, implementation and analysis in the form of a discussion of methods based on an application-oriented example of a research project from the field of midwifery science on the risk perception of obstetric health professionals.

The aim of the research project on the risk perception of obstetric health professionals is, firstly, to gain insight into what midwives and obstetricians perceive as risk during childbirth. In a second step, the project explores whether personal factors (e.g. age, gender, professional experience) or systemic factors (e.g. annual number of births or level care provided at the maternity hospital) influence the risk perceptions of the midwives and obste-tricians, and whether this, in turn, impacts their decision-making when caring for women giving birth. The project will use an exploratory sequential mixed methods design to answer the research questions shown in Figure 1 [Fig. 1]. This paper discusses the methodological and methodical considerations regarding the use of this mixed methods approach, including the definition of mixed methods, the rationale for this approach, different mixed methods designs and strategies for data collection as well as data analysis, in general, and applied to the given research project. A summary of these general and applied methodical aspects of the design and construction of mixed methods research projects is also presented in Table 1 [Tab. 1].

Kuckartz defines the mixed methods approach as follows:

Here it is imperative to draw a distinction between a multimethod research design and a mixed methods design. While a mixed methods design comprises a combination of qualitative and quantitative methods, multimethod designs employ several methods from the same category. The integration or combination of both quantitative and qualitative data is, however, expressly understood as a core characteristic of the mixed methods approach ([3] p.3), [4]. The sometimes contentious and antagonistic debate about the possibility of a general link between the two “paradigms of social science methods” ([13] p.27) is sufficiently well known. This has, however, now been somewhat “defused” and “a debate on the combination of the two has now emerged” ([13] p.29). When methods are being combined, a distinction is often drawn between the mixed methods design and triangulation, and Kuckartz emphasises that these are “very different approaches to combining methods” ([13] p.48). Given that authors submitting articles for publication in the Journal of Mixed Methods Research are now in fact advised not to use the term triangulation due to its problematic definition (as indicated in Fetters et al. [7]), this paper will not provide any further detail regarding the differences between the two terms in this paper.

Given that the field of health research encompasses many different professions and research disciplines and is also characterised by high scientific standards in view of the far-reaching consequences of errors, it can be considered a highly complex field ([14] p.417). As a result of this complexity, in the field of health research in particular, the use of the mixed methods approach is considered to have major benefits [14], [20]. The authors of the present article see midwifery science as part of this complex field of health research. Consequently, significant benefits from the use of mixed methods research designs are anticipated also for midwifery science. These research designs can address the complexity [14] and depict the requisite broad range of perspectives within the complex phenomena being studied [20]. Kelle and Krones [12] also call for researchers in the field of evidence-based medicine to draw on different research designs with their different processes of data collection and analysis. They see the use of a combination of methods as an opportu-nity to tackle the “blind spots” in qualitative and quanti-tative data collection and analysis and thus also to address possible “errors and threats to validity” ([12] p.635). According to Creswell, the use of a mixed methods approach is suitable if

1.

quantitative or qualitative research alone would be insufficient, based on their inherent weaknesses, to adequately illuminate or understand the research problem,

2.

to facilitate the acquisition of a “more comprehensive view of and more data about the problem”,

3.

to obtain two different perspectives, for instance if a phenomenon must first be examined in qualitative terms,

4.

and to complement quantitative data with more detailed information from qualitative data ([3] p.15, [6] p.12).

Another reason for using a mixed methods design is the development of contextualised instruments to first collect and analyse qualitative data and then, in a second step, to administer these instruments to the sample ([5] p.224).

All the reasons listed also apply to the research project described and lead us to the conclusion that a mixed methods design can be used in this context. To answer the research question of whether personal and systemic factors are associated with changed risk perception and whether there is an association between the risk perception and decision-making of obstetric health professionals when caring for women during labour, a survey is conduct-ed using self-constructed case vignettes from the obstetrics field. To the best of the authors’ knowledge, although there are international studies on the risk perception of obstetric health professionals [9], [10], [15], [16], [21], [24], in Germany there appear to be no studies on this phenomenon to date. To be able to construct valid case vignettes of situations perceived as risky thus requires prior investigation into the phenomenon in the context of the clinical setting. This will be conducted by means of focus group discussions with midwives and obstetricians. The insights from this initial partial survey, i.e. which situations during childbirth midwives and obstetricians perceive as risky, are used to construct valid case vignettes in the second partial survey that follows. Based on the insights from the first strand, the second strand can help to generalise the results and answer the other research questions on the influence of different variables of personal and systemic factors on risk perception and/or decision-making. Moreover, the findings facilitate a more comprehensive and thorough understanding of the results, in general, as the findings of the focus groups and the survey complement one another and enable an extended interpretation. The combination of qualitative and quantitative methods makes it possible to acquire a more comprehensive view of the research subject, i.e. risk perception and the association between this and decision-making while caring for women in labour, than would be possible using either of the two methods on their own. By using a mixed methods design, we can thus compensate for one of the drawbacks of a purely quan-titative survey, which is that it cannot generate “an understanding of the context or environment in which the people are speaking”, as the voices of the participants are “not heard directly” ([6] p.12). Combining the quan-titative with the qualitative data provides us with an inside view and also an insight into the case of the individual respondent. On the one hand, the risk of a purely quantitative approach resulting in bias and (mis)interpretations due to the researcher’s own preconceptions ([6] p.12) is countered by the qualitative research strand. On the other hand, qualitative research is sometimes considered flawed due to the subjective interpretation and, here too, there is a chance of bias and (mis)interpretations stemming from the researcher’s own preconceptions. In addition, due to the small number of participants, qualitative studies are criticised for their lack of generalisability ([6] p.12). The combination of the interpretive data analysis of the first qualitative strand and the statistical analysis of the subsequent quantitative second strand has the advantage that the shortcomings of the one method are compensated for by the strengths of the other ([3] p.15, [6] p.12). A research strategy that uses mixed methods thus seems ideally suited to investigate the complex phenomenon of the risk perception of obstetric health professionals.

The term (research) design denotes the overall strategy used to carry out an empirical study. The design defines the methodical approach that will be taken to answer the research question(s) and who or what will be studied, as well as the point in time at which this will occur and the methods that will be used. In selecting a mixed methods approach, a further choice can be made between conducting an exploratory, descriptive or hypothesis-testing study, or an evaluation study. This provides a systematisation of different designs. For this purpose, Creswell and Plano Clark [6] distinguish four dimensions within a mixed methods design: implementation, priority or weighting, integration or mixing phase and theoretical perspectives (see Table 2 [Tab. 2]).

Based on these four dimensions, various different design classifications can be described. Of these, there are three mixed methods designs that are the most commonly used or considered the standard “core” designs: convergent, explanatory sequential and exploratory sequential ([5] p.15). Burzan also refers to the sequential design forms as the Vorstudienmodell (preliminary study model), Verallgemeinerungsmodell (generalisation model) and Vertiefungsmodell (explanatory model) ([2] p.33, authors’ translation).

With these designs, the study strands take place consecutively. The key aspect here is that the findings resulting from the first method applied influence the design and implementation of the second study ([13] p.77). Either the quantitative study is conducted first, followed by the qualitative study, also known as qualitativ-vertiefendes Design (qualitative explanatory design), or the qualitative design comes first, followed by the quantitative study, referred to as quantitativ-verallgemeinerndes Design (quantitative generalising design) ([13] p.77, authors’ translation). More detailed information on other so-called complex design forms can be found, for example, in Morse [17], Creswell and Plano Clark [6] and Creswell [3]. These works also include tips on how to use and create graphs and diagrams to indicate time sequences and the prioritisation or weighting of the different me-thodical approaches within the various mixed methods designs.

The planned research project will use a qualitative partial study to generate insights on the conceptual understanding of the phenomenon of risk perception and these re-sults will influence the design and implementation of the second strand, i.e. the development of a questionnaire to capture the phenomenon and to investigate the association between risk perception and decision-making. An exploratory sequential mixed methods design was therefore selected (see Figure 1 [Fig. 1]). Here the qualitative study functions in a way that is similar to a preliminary study; the real emphasis is on the results of the quantitative study, which is why these are prioritised ([6] p.90, [13] p.65). The first strand is devoted to exploration and uses focus group discussions to examine the phenomenon of the risk perception of obstetric health professionals, which is an issue on which, to the authors’ knowledge, here have been no research studies in Germany to date. The aim of this is to gain insights into the situations midwives and obstetricians perceive as risky while caring for women giving birth in a hospital environment and in so doing to remedy the lack of “detailed knowledge about the subject being studied and the mindsets of the study participants” ([13] p.163). The subsequent quantitative study, the survey, is intended to help generalise the re-sults and, in particular, to provide information on the influence of personal and systemic factors on risk perception or on how these are associated with decision-making. With the first (qualitative) strand, it is therefore possible to generate “relevant (sub-)dimensions/hypotheses, indications for relevant items and assistance in accessing the field” ([2] p.34) for the second (quantitative) strand. In this way, the case vignettes and the survey can be developed and adapted based on insights from the field. To sum up, this study is thus geared towards a quantita-tively driven exploratory sequential mixed methods design, in which the quantitative strand has priority and the qualitative strand performs the function of a preliminary study (generalising design).

The sampling of participants for the study is of paramount importance, both for quantitative and qualitative research methods, as the reliability, authenticity and credibility of the research results are heavily influenced by the sampling [19]. Accordingly, when planning and conducting a mixed methods research project, especially careful attention must be paid to the sampling process ([3] p.75). The detailed considerations regarding the process of generating the sample depend on the research design and the intended integration of the findings [2], [26]. Teddlie and Tashakkori ([25] p.168ff) provide a comprehensive summary of sampling strategies for mixed methods research projects. Theoretical considerations for planning and generating the samples in a mixed methods research project must also give thought to whether the samples in the partial studies should be connected, partially the same or whether they only have to draw from the same population. For a Vorstudie, it is not relevant whether the samples in the quantitative strand overlap with the samples from the qualitative survey. If a study aims to generalise, however, the same population is required for both strands, although the participants in the qualitative survey are generally not the same as those taking part in the quantitative survey ([6] p.188). Accordingly, in this research project on the risk perception of obstetric health professionals, standard considerations were given to the target population and the inclusion and exclusion criteria. The participants in both research strands are midwives and obstetricians working in German hospital delivery rooms. For the first strand (the focus group discussions) participants were selected using the purposeful sampling technique. This therefore corresponds to a non-probabilistic and thus strategically targeted selection ([22] p.265) based on predefined selection criteria ([13] p.85). The use of such selection criteria, which in this case include age, gender, years of professional experience and work environment, takes into account the heterogeneity of those working in obstetric clinical settings. By combining different strategies within the purposeful sampling technique, i.e. typical case sampling with maximum heterogeneity sampling, a stratified purposeful sample is generated, which increases the relevance and credibility ([22] p.305). Here, the different characteristics (e.g. midwife or obstetrician, more or fewer years professional experience, place of work in a maternity hospital or a perinatal centre). According to Teddlie and Tashakkori [25], this stratified approach is similar to probability sampling. However, the small number of cases it generates is characteristic of purposeful sampling, in other words targeted selection of participants for a study ([25] p.186). For the survey, we plan to generate the largest possible non-probabilistic sample of midwives and obstetricians working in hospital delivery rooms in Germany. In order to achieve this, as many obstetric health professionals as possible should be encouraged to participate in the survey, but without it resulting in a random selection of participants. The strategy for the recruitment of the participants in both strands is based on self-activation, the snowball principle and gatekeeper [22].

In summary, strategies for qualitative or quantitative data collection generally include three elements: observations, interviews and documents (for qualitative data collection strategies) and survey, test, and some form of structured survey elements (for quantitative data collection strategies) ([25] p. 204). Strategies for the collection of qualitative data thus gather narrative data, which is explained using thematic analyses, and strategies for the collection of quantitative data generate numerical data which is statistically analysed ([25] p.204). The six main strategies for data collection (survey, interviews, focus groups, tests, observations and secondary data) are employed using three methodological approaches (qualita-tive, quantitative and mixed methods). Within a mixed methods research project a decision is made as to whether to conduct “within-strategy” or “between-strategies” data collection ([25] p.207). With within-strategy mixed methods data collection, one instrument is used to collect both qualitative and quantitative data. One example here is a survey with a questionnaire that includes both closed and open items. With between-strategies mixed methods both the qualitative and quantitative data are collected using more than one data collection strategy. This also applies to the research project presented here as two data collection strategies (focus group discussions and a quantitative survey) are used to collect the relevant data. According to Teddlie and Tashakkori ([25] p.206) mixed methods data collection strategies are used when individuals are asked for infor-mation or to share their experiences using self-reporting techniques. Accordingly, the data collection strategies listed are ideally suited for the research project outlined here.

As in other types of research projects, the data in mixed methods projects has to be processed, explored, analysed, presented, interpreted and validated for data analysis. However, depending on the mixed methods design selected (e.g. parallel or sequential), the approach to or the type and degree of influence of the different steps in the analysis differ. Unless this is self-explanatory due to the design selected, when planning the analysis, consideration has to be given to the form in which the findings will be processed and the point in time at which the analysis should be conducted ([13] p.121). Since the scope of this paper does not allow for a detailed description of the individual steps, reference is made to articles by other groups of authors who present the relevant steps of the process in detail based on different mixed methods designs ([6] p.204ff, [25] p.250ff). Instead, following Creswell and Plano Clark ([6] p.218f), we will summarise the steps and decisions relating to data analysis for an exploratory sequential mixed methods design like the one used in the current research project. Here, the infor-mation on planning the data analysis as applied to the research project on the risk perception of obstetric health professionals can be found in brackets. In summary, the data analysis process within an exploratory sequential mixed methods design is as follows:

1.

Collection of qualitative data (audio recordings and transcripts of focus group discussions),

2.

Analysis of qualitative data using the analytical approach best suited to the qualitative research question (structured qualitative content analysis),

3.

Development of the quantitative strand including the decision as to which qualitative data can be used in the quantitative follow-up (construction of case vignettes based on the findings from the first strand for use in the survey),

4.

Development and testing of the new instrument (pretest),

5.

Collection of quantitative data (survey),

6.

Analysis of quantitative data using the analytical approach best suited to the quantitative and mixed methods research question (descriptive statistics, inference sta-tistics),

7.

Interpretation of how the integrated findings answer the qualitative, quantitative and mixed methods questions and decision on how the quantitative findings build or expand on the qualitative findings.

One challenge when using mixed methods designs is the need to contend with different survey and evaluation methods. Not only do mixed methods research projects need careful planning, but they also require researchers to have strong methodological competencies in the field of quantitative and qualitative research methods [6], [13]. As a result, mixed methods projects are also often more time intensive, potentially requiring larger research groups and correspondingly greater financial resources [6], [13]. There is also the risk that the results and insights from the different strands will not be appropriately merged but will simply be presented side by side without being inte-grated [13]. Despite these limitations, the reasons listed above suggest that the mixed methods approach would be well suited to the midwifery science research field. In medical research, randomised controlled studies are generally seen as the gold standard for studying cause-and-effect relationships. However, in order to investigate and understand complex relationships, for which there are (still) very few available data sources, the integration of qualitative and quantitative methods in one research context is an excellent fit. In midwifery science, a discipline which is still relatively new in Germany ([23] p.83), the potential for mixed methods research designs is particularly great. Using the approach to gather infor-mation on the individual experiences, impressions and views of obstetric health professionals, on the one hand, and to link it with examination of correlations using (efficient) data analyses or possible cause and effect relations, on the other, can be particularly beneficial for this field. The considerable knowledge this method can generate regarding questions related to health care can help to optimise the quality of care provided to pregnant women, women giving birth and their families. The approach, described by Creswell and Plano Clark as “pragmatic”, of linking numerical data with words and thus inductive and deductive reasoning is just one of the many advan-tages of mixed methods research designs ([6] p.13). The research project outlined here demonstrates the potential of mixed methods by linking the knowledge acquired through the two research strands to generate insights on the risk perception of obstetric health professionals in Germany and by using qualitative and quantitative methods to allow a deeper understanding of risk perception within the research context. These insights can potentially be used to develop (training) concepts or reveal obstetric health professionals’ need for support in evaluating risky situations. In summary, it can be argued that midwifery science can benefit enormously from the use of mixed methods. For this reason, researchers in this field should aim to plan and carry out further mixed methods research projects.

The implementation of this study on the risk perception of obstetric health professionals is part of a dissertation conducted at Witten/Herdecke University in cooperation with the Hochschule für Gesundheit – University of Applied Sciences. The study received approval from the ethics committee of the Hochschule für Gesundheit – University of Applied Sciences and was self-funded by the institution with no third-party funding.

The authors declare that they have no competing interests.