We are observing the world around us all the time, so what is the difference between ‘looking’ and ‘observing’ in research terms? The answer, according to Wood and Ross-Kerr (2006: 171), is that observation stops being part of normal human activity and becomes a research method when it is systematically planned and recorded and when the results are checked for their accuracy. In other words, observation is different from looking when it is carried out systematically for the purpose of answering a research question to develop knowledge. Observation can be defined as the collection of data that are visible to visual sensors, whether that consists of the researcher’s eyes or the use of some means of visual recording. Watson et al. (2010: 382) add that as a tool of data collection, ‘observation is an active process by which data are collected about people, behaviours, interactions or events’.

As with interviews (Chapter 10), observation varies depending on the amount of structure used to record the data. At one extreme is the highly structured observation checklist that produces quantitative data and at the other is the unstructured observation of situations that are used to produce qualitative data.

In the last chapter, researchers were seen as the tool of data collection in the way they channelled the verbal information in the form of an interview. In this chapter, researchers are again the tool of data collection in the way in which they select and record the data they ‘see’ to answer the research aim. Indeed, Mcilfatrick (2008: 310) suggests that observation ‘epitomises the idea of the researcher as the research instrument’.

We have repeatedly noted that although we can ask people what they do, we may not always get an accurate answer. This is because people are not always aware of what they do, or they are unable to accurately describe or articulate their actions. Some actions are carried out at a subconscious level and are difficult to describe or write down. Explaining how to tie a shoelace to someone over the phone is a good example. Data gathering using observation can overcome the problem of verbal descriptions. It can be applied to numerous midwifery activities where the best way to find out how someone does something or ‘what happens’ in certain situations is to watch it unfold. One example of this is the work by Kemp and Sandall (2010) who observed five 36-week birth talks in women’s homes by their caseload midwife. The aim of this was to obtain a detailed description of the 36-week birth talk, and how it is delivered to, and perceived by, women and their birth partners. It would have been difficult to only ask midwives or women about this, as the answers would have missed a lot of the detail, so observations were made on five talks and then phenomenological interviews used to talk with the midwives, the women and their birth partners to answer the question on perceptions of the talk. This is also a good example of triangulation, that is, the use of more than one data collection tool to increase the validity of the results.

The next two sections will look first at the use of structured observations, followed in the next section by the use of unstructured observation. In structured or systematic observation, data are often recorded on a checklist observation sheet that itemises the kinds of activities to be observed. The researcher will indicate, often with a tick, each time one of the items on the list occurs, for example, a checklist of the times in an antenatal group setting a midwife asks a direct question as a way of gaining involvement from those present. The results are usually presented numerically, in the form of the number of occasions (frequency distribution), a percentage, or displayed in a table or figure such as a bar chart. Polit and Beck (2008: 433) suggest:

The items or variables to be recorded need clear and unambiguous concept definitions, that is, a precise description of its meaning or form to ensure accuracy of recording, especially where there are a number of observers involved. This helps to reduce the problem of the observer having to make a large number of inferences regarding an item such as ‘gives emotional support’. What exactly does that mean in terms of what would be observed? The greater the degree of inference required, the less reliable the outcome. The ideal type of item for checklist observation includes those capable of being explicitly defined so that there is no question in the mind of the observer as to whether or not they have been identified.

As events and activities unfold so quickly, there is a limit to the number of different aspects the researcher can observe at once. Care has to be taken to avoid the checklist becoming too complex. For example, it may not be possible to accurately record the type, duration and form of touch between a midwife and woman in labour, as well as the duration of eye contact, and any additional non-verbal forms of interaction.

Not only should the optimum number of elements be considered in the checklist, but also the form of recording must be simplified to enable speed and accuracy. Simple ticks or crosses are the best form of recording items. Before using a checklist, the researcher should thoroughly practice with a pilot study. This may suggest ways of reducing the complexity of the list, as well as providing the researcher with an opportunity to develop the skill of observing and recording at the same time.

The limitation of an observational checklist is the depth of information that can be achieved, and the limited complexity of interactions that can be accurately observed. This type of approach is also restricted to predicted behaviour, and does not cope well with unexpected activities not included on the checklist. This means this form of observation is not appropriate where the researcher knows little about what will be observed (Polit and Beck 2008).

In contrast to the checklist approach of structured observation are the unstructured methods found in qualitative studies. This means that the researcher records in a more open and flowing way the events and behaviours they ‘see’. Anthropologists and sociologists first developed this form of observation to examine the actions and interactions of people in their natural social world. One helpful description of researchers using this type of observation method is provided by Polit and Beck (2008: 402), who say:

One frequently used typology of observer roles when carrying out data collection was developed by Gold (1958) cited in Holloway and Wheeler (2010). This can be seen in Box 11.1. The different roles vary in the extent to which the researcher becomes directly involved with those observed, that is, whether the role is participant or non-participant. A further variation is the extent to which the observed are aware that they are being watched. The term ‘overt’ signifies that those in the setting are aware of the observer’s role, and ‘covert’ to situations where they do not know that observation is taking place. Most health care research now takes the overt approach as there are considerable ethical issues raised by the concealment of data collection, as in covert research where the individual has not given their informed consent to take part. This is despite the possible advantage of reducing the ‘observer effect’, that is, people changing their behaviour because they know they are being observed (Holloway and Wheeler 2010).