Research methods are "the techniques used to structure a study and to gather and analyse information in a systematic fashion" (Polit & Beck, 2008, p.765). In a mixed methods study, as discussed in section 3.3.1.2, using explanatory sequential mixed methods design,
quantitative data are collected in the first phase of the research and qualitative data are collected in the second phase. The former is associated with the positivistic paradigm, utilising structured methods, precise measurement and quantification of variables, so that numeric data can be analysed. The latter is associated with more flexible methods such as observation and interview, and data analysis is built up inductively from specific to general themes (Kumar, 2011; Polit & Beck, 2008).
Data sources can be broadly classified into primary data and secondary data. Primary data refers to data collected from primary sources, e.g., determining user-satisfaction with a computer system by undertaking interviews. Conversely, secondary data is collected from existing sources of data such as hospital records. Data collected from secondary sources are used in both qualitative and quantitative research. In qualitative research the secondary data could be historical, e.g., from diaries or letters (Polit & Beck, 2008), or narrative in nature, e.g., when an individual provides stories about their lives (Creswell, 2009). On the other hand, secondary quantitative data from medical records could consist of categorical and numerical variables. An example of a categorical variable could be 'gender' or whether or not a specific vital signs has been recorded or not. An example of a numerical variable could be the actual value of a recorded physiological vital sign such as body temperature or blood pressure. Polit and Beck (2008) emphasised that existing records are an important source of secondary data in healthcare research; hospital records and patient charts are considered to be rich sources of data.
Data collection from secondary sources can be relatively straightforward when compared to data collection from primary sources; primary source data utilise questions and
interviews and this can cause problems stemming from people's awareness of a study, the so-called 'Hawthorne effect' (Kumar, 2011), which is when people respond or act
differently because they know they are being studied (Polit & Beck, 2008). A problem with collecting data from secondary sources can be that sometimes it is difficult to gain access to institutional records so it is important to ensure that the required data are available and accessible (Kumar, 2011; Polit & Beck, 2008). Similar to planning questionnaires and interviews for primary data collection, researchers collecting data from existing records must make important decisions about the data to be gathered.
3.5.1 Appropriateness of mixed methods research
Venkatesh (2013, p. 45) suggested that the purpose of employing a mixed methods approach should be explained to "demonstrate the appropriateness of conducting mixed
methods research". There were several reasons for employing a mixed methods approach in the current research. It allowed the research problem to be perceived from several angles and therefore was not constricted in the way that single research methods might be. By conducting mixed methods research, the researcher is more likely to find useful answers and provide valuable knowledge that will benefit society as a whole (Feilzer, 2010). By examining the initial aims and objectives of this research, the reasons for choosing a mixed methods approach can be further clarified. As stated in section 1.4, the aim of this research was to investigate documentation of physiological vital signs in electronic health records. More specifically, the objectives included to investigate the completeness of documentation of vital signs in an EHR and to report on the documentation of vital signs in an acute care setting. These objectives could be examined from a positivist perspective and a quantitative approach to identify exactly how vital signs were represented in the EHR. The remaining objective was to identify the specific problems related to documenting vital signs in the EHR and to examine the reasons for the existence of these problems. These perspectives required a naturalistic approach to enquiry and thus a qualitative study was deemed
appropriate and designed for this investigation. This approach would allow problems to be investigated at close range, in which every detail related to measuring and documenting vital signs could be scrutinised. Combining the two approaches into a mixed methods approach was the most appropriate way to address these research questions in order to provide a holistic understanding; the quantitative research could provide breadth (Johnson et al., 2007; Venkatesh et al., 2013) by examining the representation of vital signs in the EHRs of a number of high risk patients. Qualitative research could provide a deep understanding of the practices and attitudes of the people who were caring for these patients and identify the reasons for problems associated with using the EHRs for documentation.
3.5.2 Quantitative research method
When collecting data in a quantitative method, a vital step is to identify the specific information that is needed, pragmatically, the data that are needed to solve the problem. With this in mind, the first practical step is to design a research instrument or data
collection tool (Kumar, 2011). The data collection tool should be developed with research objectives in focus so that it is linked directly to the objectives. Further, the researcher needs to decide upon a broad framework of what needs to be found out and design the data collection tool accordingly (Kumar, 2011). Questionnaires and structured interviews
are typical examples of such instruments and designing these requires careful consideration so that all aspects of research questions are addressed (Bryman, 2012). Similarly, data collected from secondary sources such as existing records need a detailed and functional instrument to ensure that appropriate data are generated to answer the over-arching research questions. In the current study the data collection tool provided a standardised form that could be used systematically for collecting data from existing records. Details of the development and use of the data collection tool are described in Chapter 4 (Section 4.3.4.8).
3.5.3 Qualitative research methods
In qualitative research approaches, the two main methods of data collection are
observations and interviews. In the current study, the second phase of the mixed methods research employed a qualitative approach. An observational study was conducted first and this was followed up by an interview study. A description outlining the data collection in these two methods is provided below.
3.5.3.1 Observational method
An observational study can promote understanding of complex situations through the observation of actions and activities (Bowling, 2009). This type of study can be a rich source of information as it enables the researcher to capture what people do rather than what they say they do (Wisker, 2001).
When collecting the data during observational studies, the researcher should not assume to understand all aspects of what is observed. For instance, the researcher's previous
experience may affect the interpretation of what s/he observes. As a means of ensuring accurate data if there is any uncertainty whatsoever, questions should be asked. This means that a researcher should ask follow-up questions during observations in less busy moments to clarify and verify what is being observed, to assess thinking processes and to gain direct views. These types of questions are known as 'opportunistic interviews' (Saleem et al., 2011).
However, carrying out an observational study is not limited to 'observing' and research carried out in the field is more comprehensive than this and termed more generally as 'field work'. Field work can be described as "observing, participating, interrogating, listening and communicating, as well as a range of other forms of being, doing and thinking" and can be a demanding task for the researcher (Mason, 2002 p.87). As Mason (2002) also mentioned,
collecting data in the field demands special efforts to create a good rapport with those being observed.
In the current research, measures were taken to ensure a good rapport and are described in section 5.4.3. Data were generated by observing the behaviour and practices of the health care professionals within the setting and communicating with those observed (Mason, 2002). A factor that can affect data collection during observations is observer intrusion. Those being observed may feel ill at ease when they feel they are being constantly watched. For this reason, it is extremely important that the researcher takes measures to put those observed at ease as far as this can be achieved. One way to deal with this situation is to select the type of observer that the researcher will be. When conducting an observational study, researchers can choose to be a participating or non-participating observer. In non- participant observation, the researcher adopts the role of detached observer, meaning that there is no interaction with the people who are being observed (Punch, 2005). Participant observation by contrast is when the researcher is immersed in the field, becomes part of the group being observed and 'goes native' (Creswell, 1998). In the current study, the researcher's background in nursing facilitated collecting the data as a participant observer. The advantages of being a participant observer include that those being observed feel more comfortable in the situation and that it feels natural to have another person in the group, in contrast to an observer who watches and says nothing. In addition, being a participant observer may ameliorate the negative impact of the Hawthorne effect. The Hawthorne effect means that people may change their behaviour in some way if they know they are being observed (Bowling, 2009). On the other hand, a participant observer may make the research less objective because of a closer relationship with the participants.
3.5.3.2 Interview method
An interview study is an appropriate method for collecting data when "people's
perceptions, meanings, definitions of situations and constructions of reality" (Punch, 2005, p.168) are being investigated. Interviews can be categorised according to the degree to which they are structured. Structured interviews have several characteristics. They have pre-determined questions with limited response possibilities; standardised questions are posed in the same format for each responder and are most often used in quantitative research. The other end of the scale is unstructured interviews, which are non-standardised, open-ended, in-depth and flexible. A third type is the semi-structured interview, which lies between these two extremes. It is also flexible but questions are planned so that specific
topics are covered and the interviewees have considerable flexibility in how they reply. For semi-structured interviews, an interview guide is prepared but questions may not adhere exactly to the schedule (Bryman, 2012). This allows the interview to progress in a conversational manner, rather than having a strict script, but ensures that all topics are covered. It also enables the researcher to use follow-up questions based on responses from the participants to gather richer data.
Another type of interview method is focus group interviews, in which several participants can be interviewed at the same time and can generate richer data because of the
interactions among participants (McLafferty, 2004). However, for this study individual interviews were selected as it was reasoned that the participants may not have wanted to share all of their opinions and ideas with their colleagues. For instance, the topic may have been potentially sensitive if some participants were not documenting information. In addition, the interviews would entail participants temporarily leaving the clinical area for a period of time and it would not have been practical to ask several professionals to be absent at the same time.
In the current study, semi-structured individual interviews were conducted because they were felt to be the best way of generating rich data and enabling complex issues to be addressed. The aim was to provide an in-depth understanding of views and experiences of the personnel (Punch, 2005) and interviews, conducted after completion of observations in each clinical setting, would augment and corroborate data collected during observations. This section has explained the research methods used in this study. The next section describes sampling and recruitment.