3

Choosing Methodological Approaches

Introduction

The choice of a general methodological approach is informed by many different issues, some subjective, some theoretical and some practical. For example, it used to be common, for undergraduate and pre-registration dissertations, and even up to PhD level, to encourage students to undertake studies using whatever methods most interested them. Intuitively, this seems a good course of action, because we want students to complete their projects, and have good reason to believe that they will be more likely to complete undertakings which involve them in things they are interested in. However, the counterargument is that this kind of supposed freedom has too many negative consequences to make it advisable. First, we might hope that student research will be useful to others as well as to the student, and following one’s methodological preferences in an unfettered way may lead to the undertaking of projects which lend themselves to a particular methodological approach, rather than to those of obvious benefit to patients. Moreover, if a student is sponsored by their employer, the employer will very likely have views as to the sort of project they would like to see carried out. This may not fit with student methodological preferences. Equally, there is the possibility that methodological preference may result in a researcher making a research question fit with a particular approach, regardless of whether that approach is appropriate (see the discussion of qualitative and quantitative approaches).

Perhaps most important, however, is the view of science that the ‘personal preference’ approach to methodological choice supports. This is the view that science is an individual, personal activity. Whilst this is certainly true in the sense that considerable personal skill and commitment are involved, science is mostly (and, arguably, most importantly) a social activity. It is carried out by groups of people, reviewed by and communicated to groups of people, and, at least in health research, undertaken for the good of society. In our view, any teaching of research which does not emphasise that gives at best a partial view of the nature of the research endeavour. Accordingly, we advise that the only reasonable rationale for deciding on a methodological approach is on the basis of its fitness for purpose.

When is a methodological approach fit for purpose?

Fitness for purpose requires that something does the job it is intended for. In research terms, this implies a number of things. First, the methodological approach must be capable, in principle, of answering the question it seeks to answer. Second, it must be practicable. Third, it must be within the expertise of the researcher. More broadly, there must, as we saw in the previous chapter, be a question which is worth answering (because it has not been adequately answered before and will tell us something of value).

Two different approaches to knowledge

Writers about epistemology (the study of the nature of knowledge) like to talk about deductive and inductive approaches to knowledge as representing two broad approaches to its generation. Although there has been debate about whether these two terms are truly separate, the distinction is certainly current in science. Basically, deductive investigation is said to proceed from a general standpoint (such as a known theory) and examine specific instances which confirm or disconfirm that general view. By contrast, inductive investigation is supposed to involve starting with specific instances and deriving a general conclusion from them. These are sometimes described as top-down and bottom-up approaches, respectively.

For example, investigating the general concept of habituation (the tendency to cease to respond to repeatedly presented stimuli) by conducting a series of experiments in which students were submitted to sudden, loud noises, and recording variations in their heart rates on successive presentations is an example of deductive research. Asking students about their experiences of such loud noises and seeking to find commonalities in their responses which might lead us to theorise about the nature of that common experience is inductive.

As we noted above, there is debate about the independence of these forms of inquiry. For example, the philosopher John Stuart Mill, in his discussion of deduction, regards induction as part of the process. Part of the issue here is that different writers use these terms in slightly different ways, and this has affected the way in which we talk about the distinction between induction and deduction. A detailed discussion of the relationship between deductive and inductive reasoning is beyond the scope of this book. However, many people are familiar with the deductions of Sherlock Holmes, and a very readable and enjoyable account of inductive and deductive reasoning using Holmes as an example is available at http://www.bun.kyoto- u.ac.jp/~suchii/holmes_1.xhtml.

Different broad methodological approaches and their appropriateness

In research, there has been a tendency to associate inductive reasoning with qualitative research and theory building, and deductive reasoning with quantitative research and theory testing. However, this is not a hard and fast rule. For example, a small survey might possibly be a reasonable way of generating ideas which would then lead to some general hypotheses about the world which would be tested in larger studies. Here, a quantitative method is proceeding from the specific to the general. Likewise, many qualitative studies end up by referring back to existing theories, although whether they could justifiably be called tests of such theories is another matter. Finally, mixed methods studies seek to combine qualitative and quantitative approaches, but do not necessarily combine theory building and theory testing.

Nevertheless, the broad alignment of qualitative research with inductive reasoning and quantitative with deductive is probably a useful rule of thumb. For the researcher, it may be most useful to combine such ideas with an examination of the researcher’s goals for the project. These will almost always be framed as research questions. Looked at in this way, our grounds for choosing qualitative versus quantitative methods are clearer. Broadly speaking, qualitative methods are better employed at the beginning of the life of a research question, when little is known about the subject. You can see that this is tied to the idea of theory building. By its nature, a problem we know little about is often unlikely to be associated with major existing theories. However, one thing which the novice should beware of is assuming that this is the case. For example, imagine we are examining the information needs of people who have experienced surgery which has caused a change in their facial appearance. As it happens, there is comparatively little research into this area. Nevertheless, there are any number of theories which are relevant and might bear testing in this group. We do not necessarily have to assume that a new series of qualitative studies using inductive methods is necessary to build new theory. However, it may still be the case that we will want to do some initial qualitative work to get an idea of these people’s experiences.

Quantitative approaches, by contrast, are best used when quite a lot is known about a topic area. Often, quantitative approaches, particularly treatment comparisons, come at a stage when a research question has been under examination for some considerable time, and is well integrated with a particular theory. Indeed, there may actually be competing accounts of what factors impact on the question, and these may be associated with different theories. Quantitative approaches attempt to tease out cause–effect relationships and establish predictive models associated with established theories, often allowing us to judge between the usefulness of competing models and theories. As with qualitative research, however, this is not always the way in which quantitative methods are used. Even in the absence of any theory, it would be perfectly possible to compare different approaches to care using, for example, a randomised controlled trial (RCT), to see which worked best for patients. The rationale for so doing might have no theory attached to it, but be driven simply by having observed these two different practices in the clinical setting. Given that a good deal of healthcare is based on tradition, this is quite likely, and finding evidence to support one practice rather than another can validly occur in the absence of any underlying theory.

Given that the associations between inductive reasoning and qualitative methods and between deductive reasoning and quantitative methods are not mutually exclusive, what other guidelines might we use to decide the general appropriateness of the two approaches? One distinction we have already met is the one based on the maturity of the research question. By maturity, we mean the amount of investigation it has received. As we noted, qualitative methods are better suited early in this process, and quantitative ones later. Associated with this notion are the ideas of exploration, explanation and prediction. Early in the life of a research question, much of the work is exploratory, and qualitative work is, arguably, better at this type of activity, because it examines individual cases in great detail. Some level of explanation of the characteristics is then possible, but that explanation is best pursued in further quantitative studies, because part of the notion of explanation is that it should hold good in general, not just in relation to the specific sample being studied. Qualitative and quantitative approaches make different suppositions about explanation and generalisation, and these are explored in the relevant sections of this book.

However, by and large, quantitative research places more emphasis on the generalisation of results to large populations. Moreover, this type of research is often concerned with the idea of the prediction of behaviour. For example, when we examine the action of a drug in a sample of patients we actually want to be able to predict how it will behave, in general, in patients in the future. Similarly, if we are looking at a healthcare intervention such as reassurance, we want to know, in broad terms, how patients will respond to it in the future.

This emphasis on prediction led to a reaction against quantitative methods amongst nurses in particular, but also other groups of HCPs, because it was felt that such an emphasis ran contrary to the view of healthcare practice as an individualised, patient-centred activity. Happily, much of this debate has now subsided, as most doctors and other healthcare professionals now recognise the complementary strengths of individualised care planning and effective general interventions which can be tailored to individuals. In the same way, at their best, qualitative and quantitative approaches can go hand in hand. Without the generalisability of quantitative research, we would have no rationale for using a treatment which had proved useful with one patient on another, but without qualitative research, we would often never have discovered the treatment to begin with.

One factor which can drive our decisions about research methodology is this understanding of the difference between individual and general experiences. This distinction precisely characterises the major difference between qualitative and quantitative methods. Researchers wishing to investigate personal experiences in detail will be most likely to find a qualitative approach fulfils their needs best, as this type of research allows the flexibility to work in depth with a few respondents and help them to give detailed accounts of their experiences. By contrast, researchers who are interested in drawing conclusions about what large groups of people experience will typically use quantitative approaches. Here, detail and, to a marked extent, flexibility, are sacrificed in order to allow this general picture to emerge. There has to be this trade-off for two reasons. First, if we want to know what groups of people experience, we usually have to research large numbers, and if we do this, we do not have the resource to get a great amount of detail from each respondent and then analyse it. Second, we will want to gather the data in a specified, systematised (usually, numerical) format, because this will allow us to examine the data in a way which permits us to draw conclusions about the population as a whole (see Chapter 5).

What sorts of questions for what general approaches?

Having explored the general rationale behind qualitative and quantitative approaches, we will now consider the kinds of specific issues you might use each approach to investigate. This discussion is by no means exhaustive, and much of it will be amplified in the sections of this book concerned with the two approaches. What we intend here is a series of illustrations which might help you in deciding how research questions might best be tackled.

We suggested that qualitative research was at its best when dealing with the detailed examination of individual experiences, and also noted its role in theory building and exploration. It follows from this that the following sorts of areas are best examined qualitatively, certainly in the first instance:

Patients’ and carers’ experience of illness and care. Typically, researchers will use unstructured or semi-structured interviews to approach this broad topic area. These allow detailed exploration and the opportunity for the respondent to tell their own story in their own way.

Research with excluded and hard to reach groups. Here, qualitative approaches using interview and observation have several advantages. Apart from allowing detailed exploration, qualitative approaches may be more acceptable to the target groups, as they are, arguably, less formal and more collaborative. Finally, by their nature, these groups have been less frequently researched than others and, in consequence, research from them will benefit from an inductive approach to theory building.

Pilot and feasibility studies. When we are developing a new intervention, we want to know as much as possible about precisely how it is being carried out and the reaction of patients. There may well be an element of quantitative analysis in such studies, but the detailed observation and conversational questioning familiar to qualitative researchers has a lot to offer in describing the exact circumstances which may influence effectiveness and acceptability of an intervention. Additionally, such approaches can tell us about potential methodological problems in mounting a treatment study.

Quantitative approaches, by contrast, are best at examining general trends in large groups:

Patients’ and carers’ experience of illness and care. The type of research appropriate here will probably be by means of surveys, and will almost always follow on from a qualitative study. The difference between qualitative and quantitative examinations of experiences is essentially that approaches like surveys give a broad-brush picture of groups of people’s experiences, rather than those of individuals.

Epidemiological studies. These studies are essentially survey approaches, possibly using elements of observation. Their aim is to identify trends in populations with regard to the behaviour (usually, an illness or response to it) under examination. Both the survey and observational components will be carried out using strict sampling protocols and standardised measurement procedures to ensure that accurate rates emerge which are generalisable to the whole population.

Treatment comparison studies. Typically, the researcher will be keen to ensure no bias which might alter apparent treatment effects intrudes into these studies. Accordingly, quantitative researchers have developed a series of different research designs which attempt to reduce bias so that the relationship between treatment and outcome is relatively clear. Probably the most well known and methodologically adequate of these (but also the most difficult to mount) is the RCT, in which patients are randomly allocated to receive one or another treatment. In some situations, it is possible to offer the treatment in such a way that neither the patient nor the clinician knows who is receiving which treatment. In situations where RCTs are not possible, other designs such as non-randomised designs and before–after designs maybe used. The same standardised measurement and treatment delivery are used as in RCTs, but there is greater opportunity for bias.

Do we need to do research at all?

All research is hard to carry out. Apart from methodological problems, there are difficulties with organising treatments across different sites, recruiting respondents, ensuring the safety of respondents, ensuring adequate expertise to analyse the data appropriately. For this reason, it is important to be clear that a piece of research actually needs carrying out. We can see two main reasons for doing a piece of healthcare research: it addresses an important question; it responds to a gap in the literature. In the first instance, we will agree that the question is important if it has consequences for our understanding of health and illness and if it is likely to lead to improvement in the delivery of care. In the second case, we will want to be sure that an issue has not been examined before, or has not been examined adequately. The principal way of establishing this is via review of the existing literature. This consists of two parts: searching and appraising. We examine these issues in Chapters 4 and 22, respectively.

Stay updated, free articles. Join our Telegram channel

Join