As we discussed in Chapter 2, choosing a research method is not just a case of picking the one that seems the easiest but picking the most appropriate relative to the knowledge you require. Qualitative data consist of words, pictures and sounds and are usually unstructured in nature. As such, qualitative data are not easily converted into a numeric format and, in general, need to be analysed using a different set of techniques from quantitative data. Studies which utilise, generate, analyse and interpret qualitative data are quite complex to design. You should be under no illusions - the production and analysis of qualitative data is no easier than for quantitative data. Admittedly, data produced using qualitative techniques are generally not analysed using dreaded statistics. However, qualitative studies require just as much planning as quantitative studies and in many ways are more demanding. It is therefore important that you plan your project carefully (Chapter 2) and read this chapter and Chapter 8 before you try to undertake a project which utilises qualitative data. This chapter is designed to provide you with a knowledge of how to produce qualitative data, the different approaches concerning the use of qualitative techniques, and appropriate sources of secondary data.
Just as there are many different schools of thought concerning how research in human geography should be conducted (see Section 1.4), there are a variety of opinions concerning how qualitative research should be undertaken. As such, there is more to undertaking qualitative-based data production than just conducting interviews, observing people, or undertaking secondary analysis of archival sources. There are also a multitude of ways in which to approach these tasks, depending upon your perspective and purpose. Tesch (1990), for example, identifies a multiplicity of labels that researchers in the social sciences have used to define their qualitative research (Table 7.1). She suggests that the main problem associated with these labels is that each is defined in a different way. For example, some labels refer to the perspective that a researcher adopts (e.g., naturalistic, clinical), others to the field in which they are based (e.g., phenomenology), others to the research approach used (e.g., case study, discourse analysis) and some to the data type, method or research location (e.g., field research, participant observation). To go through each of these labels and approaches would take a whole book in itself and might be a fruitless exercise when so many of them overlap in nature, or are synonyms for others,
and have more relevance to other social sciences than to human geography. The fact that all these perspectives have been grouped together as qualitative in nature suggests that they share common attributes. Tesch (1990) herself divided these labels into three more manageable groupings: language-orientated approaches concerned with the use of language; descriptive/interpretative approaches concerned with experience and meaning; and theory-building approaches concerned with identifying the connections between phenomena.
Tesch's classification is by no means the only attempt to pigeon-hole different studies. Other researchers have also tried to classify qualitative research. This book, however, is not the place to become enmeshed in trying to explain the differences between all the approaches to qualitative research. Needless to say, the differences are often tied to wider theoretical debates, as discussed in Chapter 1, and to specific areas of research (see Cresswell, 1998, for an extended discussion on five different qualitative approaches at various stages of the research process). In our discussion, we therefore concentrate on the qualitative techniques themselves (interviewing and observation), how to undertake qualitative analysis in relation to human geography, and three different approaches to qualitative analysis which have gained some currency in geographical enquiry (ethnography, action research and case studies).
When designing your study there are two important considerations: first, 'How am I going to produce data?', and second, 'How am I going to approach data production?'. Researchers can produce qualitative data from primary sources in a number of different ways. For the purpose of our discussion we have classified qualitative techniques into two generic classes: interviewing and observation. Within the accounts detailed below, interviewing and observation are considered from a traditional scientific perspective. That is, it is assumed that the researcher is an objective scientist producing data in a neutral fashion and for no other purpose than to increase understanding of a particular phenomenon. We have chosen to frame our account in such a fashion for ease of description. Later in this chapter (Section 7.4), three other approaches to interviewing and observation are detailed. These are ethnography, where the researcher attempts to understand the world as seen through the eyes of the participants, action research, where the researcher is seeking, along with the participants, to explicitly alter a situation, and the use of case studies, where research is directed towards specific cases in real-life settings.
The interview is probably the most commonly used qualitative technique. It allows the researcher to produce a rich and varied data set in a less formal setting. Like its cousin the questionnaire (see Chapter 3), the interview comes in many forms, ranging from the highly structured to the completely unstructured. The interview differs from the questionnaire in the nature of its questions and its manner of presentation. While the questiomiaire is useful for asking very specific questions concerning quantifiable information such as age, income and sex or for converting general information into a closed form through rating or ranking, the interview allows a more thorough examination of experiences, feelings or opinions that closed questions could never hope to capture. Dey (1993) characterises this difference such that questionnaires concern numbers or facts and interviews concern meanings or beliefs, although this difference is not mutually exclusive. The interview is also more informal in nature than the questionnaire and cannot be self-administered. While a questionnaire has a very formal question-then-answer structure, interviews are often described as entering and maintaining a conversation. This is not to deny that some interviews do follow a highly structured path, but the nature of this path does differ in the nature of presentation. Oppenheim (1992) suggests that the interview is really a precursor to a larger questionnaire survey, with the interview providing the basis for the closed-ended questions on the questionnaire. Used together, the interview provides a pilot for formulating relevant questions and the questionnaire ensures a larger sample size and data that can be analysed quantitatively. While interviews and questionnaires can be used in this way, many researchers would disagree with such a strategy for two reasons. First, some would argue that an interview alone will provide a sound data set. Second, some would argue that the closed questions on a questionnaire filter out meaningful information. As such, they would contend that questionnaire surveys are fundamentally flawed as it is meaningless or pointless to try and measure and analyse experiences, opinions and attitudes in such a manner.
Interviews can provide rich sources of data on people's experiences, opinions, aspirations and feelings. There is, however, more to interviewing than simply asking a participant questions. Interviewing can be a complex social encounter. As a researcher, before you undertake any interviews, you need to understand the dynamics of interviewing and the various different interviewing strategies, and be aware of both the strengths and limitations of interviewing.
There have been various different classifications of interviews. We will use a variation of Patton's (1990) classification. Patton identifies four different interview strategies:
These categories are not mutually exclusive and two or more interviewing strategies can be used during the same interviewing session depending on what sort of information is sought. For example, we might combine a closed quantitative interview designed to gain background facts with an interview guide approach designed to elicit a greater depth of information about specific topics. The differences between these categories essentially concern the degree to which the interviewer controls the conversation and the degree of standardisation in the questions asked. Effectively, the closed quantitative interview is discussed in Chapter 3 and is essentially a verbal questionnaire. All the questions are predetermined and the questions closed so that respondents have only limited choices for a response.
Within a structured open-ended interview the conversation is highly controlled by the interviewer. Like the closed quantitative interviews, the questions are also highly structured and standardised. However, rather than consisting of closed questions which would transform this type of interview into a questionnaire, a series of open-ended questions are asked. Open-ended questions mean that the interviewee's responses are not constrained to categories provided by the interviewer; respondents can give whatever answer they wish. It is hoped that open-ended questions better reflect a person's own thinking. Using this strategy all the interviewees are asked the same basic questions
Box 7.1 An example of a structured open-ended interview
I am interested in problems of access for disabled people in urban environments. I am going to ask you a series of questions relating to this topic. Your answers to these questions will be treated in strictest confidence.
in the same order. The exact wording and sequence of the questions is determined before any of the interviews are conducted. This structured strategy is meant to try and increase the comparability of responses and ensure responses to all questions for every interviewee. It is also thought that this strategy might reduce interviewer effects and biases introduced through free conversation. Furthermore, this structured approach provides a 'natural' basis of organisation for analysis of data. However, this strategy does have its weaknesses. Because the interview is so highly structured and standardised, it allows little flexibility in relating the interviews to particular individuals or circumstances (it removes individuality) and may also constrain and limit the naturalness and relevance of questions and answers (the question might not be relevant to an interviewee but requires a response, and a particularly interesting response cannot be followed up in more detail). Box 7.1 provides an example of a structured open-ended interview (note how this differs from the questionnaire questions in Box 3.4). The respondents for the study outlined in Box 7.1 are known to have experienced problems of disabled access.
An interview guide approach is less structured than that taken in a standardised open-ended interview.
Box 7.2 An example of an interview guide approach
I am interested in problems of access for disabled people in urban environments. I am not going to ask you specific questions but rather just want to discuss with you some general topics. Your answers to these questions will be treated in strictest confidence.
Topics to be covered (for the eyes of the interviewer only)
Topics and the issues to be covered are specified in advance in an outline form but the interviewer can vary the wording of the questions and the sequence in which the questions are tackled. As a result, the interviewer has much greater freedom to explore specific avenues of enquiry, and logical gaps within the data can be anticipated and closed. The interview also takes on a more conversational feel while ensuring that all the topics of interest are explored. Because the interview is more free-form there is, however, the possibility that specific topics may be inadvertently omitted. Furthermore, because of the flexibility in sequencing and wording, the questions posed to interviewees may vary, thus reducing the comparability of the responses. Interview guide approaches therefore require the interviewer to have the ability to keep the conversation based around specific topics, within a more informal interview style, and not to let the conversation take off on wild tangents. Box 7.2 provides an example of a interview guide approach.
An informal conversational interview is generally considered to lack any formal structure. The questions the interviewer asks are meant to emerge from the immediate context of the conversation and are asked in the natural course of a discussion. Similarly, there is meant to be no predetermination of question topics or wording. With little or no direction from the interviewer the respondents are encouraged to relate their experiences, describe events that are significant to them, and reveal their attitudes and opinions as they see fit. The great strength of such an approach is that the interviewees can talk about any issue in
Box 7.3 An example of an informal conversational interview
I am interested in problems of access for disabled people in urban environments. I would like to discuss this subject with you in detail. Your answers to these questions will be treated in strictest confidence.
No predetermined agenda (questions are asked in the context of the conversation).
any way they feel, thus challenging the preconceptions of the researcher. The unstructured format allows respondents to talk about a topic within their own 'frame of reference' and thus provides a greater understanding of the interviewees' point of view. Thus, this method is said to increase the salience and relevance of questions as they emerge within the natural flow of the conversation. Furthermore, the interviewer is also given a great deal of freedom to probe various areas and to raise specific queries during the course of the interview. Informal interviews can vary in nature from discussing specific topics to constructing life history or biographical accounts (see Cresswell, 1998, for discussion). While this method of interviewing does provide a very detailed and rich source of information, the data produced can vary substantially from one respondent to the next. Further, some aspects that may be of interest to the respondent might not arise naturally. As a result, comparability across interviewees is more difficult. Data produced in this fashion can also be difficult to organise and analyse. While an informal conversational interview produces the 'richest' source of data, it also demands a relatively high degree of interviewer skill. You must be able to keep the conversation flowing naturally and show a strong set of interpersonal skills to try to elicit deeper insights into the interviewees' thoughts or experiences. Box 7.3 provides an example of an informal conversational interview format.
A group discussion can sometimes be a useful alternative or supplement to one-to-one interviews. A group discussion generally consists of a set of three to ten individuals discussing a particular topic under the guidance of a moderator who promotes interaction and directs the conversation. The dynamics of a group discussion often bring out feelings and experiences that might not have been articulated in a one-to-one interview. For example, in a group discussion workers may be more likely to express grievances against management, as they feel 'safer' within a collective environment. The group dynamics might also work in a negative way as well, with some participants reluctant to voice an opinion through shyness or fear of embarrassment. As a result, a group discussion might produce different perspectives to one to-one interviews concerning the same issues. The challenge for the interviewer is to keep the conversation flowing and to try to involve all members of the group. The selection of people for the group is also crucial. The individuals chosen should generally be from the same background and have the same characteristics. For example, there is no point in trying to run a group discussion on racial abuse on a housing estate with members of both the abused and (potential) abusers present, as both may be reluctant to speak freely in each other's company.
Robson (1993) suggests that the interview is a commonly used method because it seems a relatively straightforward and non-problematic way of finding things out. However, from the discussion so far it should be clear that there is more to interviewing than just talking to people. An interview is a complex social interaction in which you are trying to learn about a person's experiences or thoughts on a specific topic. You must remember that the interview is not just a passive means of gathering information but is also a social encounter. As with all social encounters the interview is rule-guided with both parties bringing expectations about the content and role they may play. Interviewing requires a high level of interpersonal skills such as putting the interviewee at ease, asking questions in an interesting manner, an ability to listen to the responses and act accordingly, recording the responses without upsetting the conversational flow, and giving support without introducing bias (Oppenheim, 1992). In general, you must also be able to try to balance the establishment and maintenance of a rapport with the interviewee, so that a trusting relationship is developed, while maintaining a neutral position about the topic under discussion so that you can be objective in your analysis. You should be aware that the adoption of neutrality depends upon your position as a researcher. If you subscribe to a feminist approach you may reject the concept of researcher neutrality and try to adopt a more emancipatory or empowering approach (see the section on critically appraising interviewing below). Without these skills the data set you produce may be underdeveloped or lacking in depth because the respondent may be unwilling to impart information to you.
Compounding the interview process are a series of personal characteristics such as age, skin colour and accent which may affect how the interviewee will react or 'take' to you. Clearly there is little you can do to alter these characteristics. You can, however, dress accordingly and by treating the respondent with respect try to gain their confidence. In other words, you must dress and act in an appropriate manner to the situation and interviewee. An interviewer who turns up at an 'eco-warrior' or new-age traveller camp wearing a suit and tie is not likely to get very forthcoming responses to their questions. Remember to try to listen more than you speak and to look as if you are enjoying the conversation. If you look bored the interviewee is likely to be less forthcoming with their views.
The first decision you have to make is which style of interview you are going to use. As discussed, all have their advantages and limitations. One of the main considerations at this point might concern your interviewing ability. To be able to conduct informal conversational interviews requires a great deal of interviewing skill. You must be able to keep a conversation flowing while probing the interviewee for relevant information. If the interview concerns a sensitive issue, using an informal conversational or interview guide approach means that there is more scope for causing offence through a badly phrased question. These approaches also require you to listen carefully and be able to think quickly 'on the go'. If you are a novice interviewer it may be wise to start with a standardised open-ended interview or an interview guide approach as you develop your interviewing technique. If time is of the essence then a standardised open-ended interview may be a more suitable interview method to use, as the data are more formally organised and easier to analyse. Remember, interviews are time consuming to undertake and even more time consuming to analyse.
The second mam concern relates to the medium in which you are going to conduct the interviews. There are three basic choices. The most common interview medium is a face-to-face meeting. This sort of meeting has distinct advantages, not least of which is that it is more personal in nature and that you can more easily gauge the interviewee's reaction to a specific topic through their body language and facial expression. A second option might be to conduct the interview using a telephone. Telephone interviews might be a viable option if it is difficult or expensive to meet face-to-face. Alternatively you might require a random sample of the general public which may be drawn from a phonebook, or speed might be the essence (telephone interviews take less time than face-to-face because there is no travelling). While less personal in nature, a telephone interview does allow you to pick up voice inflections. A final option may be to conduct an interview via e-mail or Internet Relay Chat. The Internet's growing use has led to several studies which have utilised it as a medium for research. For example, Correll (1995) conducted several interviews with users of an online lesbian 'cafe' via the Internet. In this case the medium was particularly useful, firstly because the respondents were widely scattered across the United States, and secondly because some of the interviewees were initially reluctant to be formally identified: the Internet did allow them a degree of anonymity. Many studies of online interaction have found that this anonymity leads people to express themselves more openly and honestly (see Turkle, 1996).
Interviewing requires quite a large commitment from the interviewee, as they are not only giving up their time but also imparting significantly more information than they might otherwise do in another medium (e.g., questionnaire). This information is more likely to be personal in nature and may be sensitive. Therefore, you need to be able to persuade people to take part in your research project. This can be achieved in two ways. First, you need respondents to feel that taking part will be pleasant and satisfying. If you are dealing with a particularly sensitive topic that may arouse deep emotions, this might be more difficult. Second, you need interviewees to feel that your study is worthwhile. Convincing potential respondents that a student project is worth contributing towards your project mark might be quite difficult. Our advice is to prepare in advance a summary of why you have chosen that research topic. These two methods of persuasion can be achieved by formally briefing respondents when first contacted. This briefing should consist of four parts:
In terms of finding interviewees to take part in the study, the sampling strategies discussed in Section 3.4 are still applicable. However, some might have more relevance than others. In general, respondents are not drawn from the whole population but from very specific groups of people. In these cases, individuals might be contacted through institutions (sampling strategies still apply) or through snowballing.
Whichever type of interview you decide to undertake, through whichever medium, how you phrase your questions is of critical importance. 'Good' interviews are those which ask 'good' questions. A good question is generally one which is clear, concise and easy to understand. The questions throughout the interview should vary in format to try to keep the conversation flowing and minimise boredom. For example, in Box 7.1 we could have phrased the first three questions in the same manner, just changing the last few words. Instead, we have tried to vary how these essentially similar questions have been phrased. Questions which start with Who? Why? What? Where? When? How? help to establish the basic framework. Mikkelsen (1995), however, suggests that Why? questions be used sparingly because they can put the interviewee on the defensive. Also, care must be taken to ensure that questions are open-ended in format to ensure explanation rather than 'yes' or 'no' answers. The more specific a question to the interviewee's situation, the 'better' the information that can be expected. In other words, try to personalise the questions. For example, rather than ask 'what motivates people to move house?', ask 'what motivated vou to move house?'.
In an informal conversation or an interview guide approach, probing questions (cross-checks) are also important to establish the depth of feeling and validity of a statement (that is, whether the opinion tallies both times). This should be done by further exploring answers (e.g., 'In your answer you mentioned do you mean . . . or were you referring to . . .?'; 'Could you go over that again?'; 'Anything more?') and should not amount to a cross-examination that will put the interviewee ill at ease. Remember, you are not trying to interrogate your participants. Probe questions might also be used to get the respondent to elaborate further on a specific subject. For example, in the following passage the interviewer has repeated a phrase in order to try to get the interviewee to elaborate further on a point. The probe has resulted in a great deal more information.
INTERVIEWEE: The main problem I encounter is steps. Being in a wheelchair basically means that anywhere that has steps at the entrance is inaccessible. Public buildings are the worst.
INTERVIEWER: Public buildings are the worst?
INTERVIEWEE: You know, churches, town halls, sports centres and so on . . . I mean, they are meant to be built so that everyone can get in them. But they aren't. Most of them are just little oases for the able-bodied. In my local sports centre they haven't got a lift so I can't visit the first floor without somebody carrying me up there. The weight-room, where I want to go to train, is up there. I've tried complaining but it's the same old story - no money. What can you do?
Another way to get an interviewee to explore a situation might be use prompts. Here, you offer the interviewee a few more alternatives and ask them to comment on them. For example, you might ask: 'You mentioned the rate of pay as being important, how about other factors such as work conditions or work hours?'. When using prompts, however, you must be careful not to bias the questions. Refer to Box 3.5 for a general guide to asking questions; Box 7.4 lists common interviewing mistakes.
Box 7.4 Common interviewing mistakes
Source: Adapted from Mikkelsen 1995.
It is essential that the interview is adequately recorded. Failure to adequately capture the interview discussion will lead to problems of analysis and weakens the validity of the study (remember that full capture might be difficult in some situations). There are a number of ways that interviews might be recorded depending upon the nature of the interview. A face-to-face interview can be recorded by either jotting down comprehensive notes, tape recording the discussion, or video filming the interview. If the person is deaf you may need to video their hands to transcribe the signing at a later date or get a third person to transcribe the interview for you if you are also signing. Telephone interviews can be recorded by taping the discussion or making notes. The Internet interview can also be recorded by capturing the written dialogue to a file or by saving the e-mails. Each medium of recording has its merits and limitations.
Audio recording of an interview allows you to accurately record an interview word-for-word with a minimum amount of effort. Recording the interview does allow you to concentrate fully upon the discussion rather than trying to balance conversation and note-taking. While it does provide a rich data set, you should be aware that there are problems associated with its use. For example, Hester Parr (in press), in her study of the geographical worlds of people who are mentally ill, illustrates that taping interviews is not always an easy process:
Taping an interview was also problematic in my study as some interviewees often experienced states of mind which incorporated paranoid thoughts about being monitored, recorded and bugged. The geography researcher who uses recording equipment in this context has to negotiate and be sensitive to these meanings. At times it was appropriate that I abandoned the tape, abandoned my quest for 'order', and was content to hold halting conversations that could take hours, over several cups of tea and periodic silences. This of course does not apply to all people with mental health problems and not to all of the people who I interviewed and some were happy to speak with the tape recorder running, be that in short and sporadic bursts.
Some interviewees may be uncomfortable knowing that they are being recorded. Also, while the data recorded on tape will be rich in detail you must also remember that its analysis will be time consuming. General estimates place fall transcribing time for a one-hour interview at about 6 to 9 hours. There might also be a tendency for you to lose concentration and not to listen fully to the respondent because you know that everything is being faithfully recorded. One way to retain concentration is to take short notes - key words and phrases - to keep account of what has been said and what still needs to be covered. These notes can also act as a backup if your tape recording has failed. Remember that the tape recorder will not record body language or prompts such as nodding of a head or gesturing with hands. Therefore you might have to articulate responses for the benefit of the tape or jot down certain responses. When using a tape recorder always make sure you have spare tapes and batteries. Position the tape recorder directly in front of the interviewee and try to use an external microphone. Be aware that background noise can make your job of transcribing much harder, so try to conduct the interview in a quiet (or quietest) location.
Video recording provides additional information to the spoken word. When reviewing the recording the interviewer can note body language that may have been missed at the time of interviewing. Again, interviews captured using video filming might also suffer because interviewees are self-conscious about being recorded. With film, a great degree of anonymity is also lost. When using the video, position the camera so that it includes both yourself and the interviewer 'in shot'. If possible, use an external microphone.
Note taking can provide a rich description of an encounter when well written. However, in the course of an interview where the discussion often moves at a pace much quicker than can be transcribed, the note-taker must be skilled in identifying and jotting down the most important aspects of the discussion. At the same time as noting down the interview you must be able to concentrate on what has been said and to keep the conversation flowing. The notes taken will probably be piecemeal and disjointed with just key concepts and phrases recorded. As a result, if you do take notes after the interview it is wise to go back through them and add in extra comments to give key comments context. This ensures that when you come to look at the notes at a later date they make sense.
Similar to the administering of a questionnaire in Chapter 3, given the range of issues concerning the interview style, medium and method of recording, it is advisable to undertake a pilot study to try to iron out any difficulties that might be encountered in a larger study. There is no point in starting your project to discover at some later date that there is a significant flaw in your data generation or analysis. For
Box 7.5 Interviewee evaluation of pilot study
Issues to be raised:
Source: Adapted from Barrat and Cole 1991.
example, you might find that some of the questions you thought were perfectly reasonable cause offence or that people interpret them incorrectly. You can assess your pilot study in two ways. First, you can self-evaluate the effectiveness of the strategy used in terms of response rates, answers given, ease of analysis, etc. Second, you can ask your interviewees to evaluate the interview for you (see Box 7.5).
From the discussion above it should be clear that there are advantages and disadvantages to interviewing. The interview can provide a fuller and richer data set than might otherwise be gained through highly structured, closed questions. This allows the interviewee to explain further their experiences, attitudes and opinions. Further, the less structured the interview the greater the flexibility the researcher has to direct the conversation and to explore specific issues in depth. On the negative side, interviews are more costly to undertake and analyse because of their time-consuming nature. There might also be problems relating to a greater likelihood of interviewer bias introduced through prompting and question phrasing. As with questionnaires, there is also the possibility that the respondent might try to predict what the interviewer wants to hear, and rather than articulating their own views, instead might forward an alternative they feel that will satisfy the interviewer.
Feminist commentators have been critical of textbook (cookbook) guides to undertaking interviews. They suggest that the description rarely conforms to the actual event. In textbook descriptions of an interview the interviewer is usually totally in control of the situation, asking the questions and responding to the answers. 'Normal' conversations do not take this form. In 'normal' conversations both parties ask and respond to questions. In most interviews the interviewee will also ask the interviewer questions about specific issues. Feminists argue that in such a case, if the interviewer refuses to answer an opinion-based question, they are compromising the rapport and trust of the interviewee. If the interviewer does respond to such a question, however, then they are breaking the notion of neutrality or objectivity. Furthermore, feminists would argue that the interview strategy described in most methods books represents a masculinist view of research. This strategy seeks to exploit the knowledge of the respondent rather than to empower them. When interviewing people from marginalised sections of the community, feminists would argue that traditional interview methods maintain and reinforce current social power relations. Feminists suggest that the power relations within an interview must be renegotiated and that the interviewer must recognise that they cannot be neutral and objective. Genuine trust must therefore exist between the interviewer and interviewee, and the outcome of the research must also be genuinely empathetic and empowering. The interviewer then must develop a genuine rapport with the interviewee based upon shared concerns. To achieve this rapport the researcher must become engaged in the life of the interviewee. In other words, if you are interested in homelessness, rather than just interviewing homeless people you need to either work in an organisation devoted to helping homeless people or actually live among them. Only in these ways can you hope to build a genuine trust between yourself and your homeless interviewees. Your experiences will also help you to put your research into context. Clearly, this may be impractical for a student project. Nonetheless, feminists would argue that every effort should be made to become involved with the research participants beyond an interview. Dyck (1993) suggests that researchers should be more reflexive in their approach. This means that you reflect fully on your assumptions, and your part in the research process. Before we discuss alternative approaches to qualitative research, however, we will examine the other component of qualitative study, observation.
Wolcott (1995) suggests that the difference between interviewing and observation is that in observation you watch as events unfold whereas with interviews 'you get nosy'. Interviews are self-reports of experiences, opinions and feelings, whereas observation relies on the observer's ability to interpret what is happening
Table 7.2 Types of observation.
| Straight observation | Participant observation | |
|---|---|---|
| Overt | Researcher does not engage with the group under study but makes no attempt to conceal fact of observation | Researcher joins a group as a participant in an event but does not hide fact that (s)he is observing them |
| e.g., observing how children learn a map in a classroom setting. | e.g., observing domestic labour relations by working as a cleaner where fellow cleaners know you are a researcher. | |
| Covert | Researcher does not engage with the group under study and does not reveal to the group that they are being studied | Researcher joins a group as a participant in a situation without telling them that s(he) is observing them |
| e.g., observing farming practices by walking down country lanes. | e.g., observing tourist behaviour on a coach journey by posing as another tourist. |
and why. Observation then 'entails the systematic noting and recording of events, behaviours, and artefacts in a social setting' (Marshall and Rossman, 1995: 79). Data can relate to observing conversations and overt behaviours. In particular, observation focuses upon people's behaviour in an attempt to learn about the meanings behind and attached to actions. Observation then assumes that people's behaviour is purposeful and expressive of deeper values and beliefs. Observation does not, however, need to be confined to observing people in contemporary settings. Much research within traditional cultural geography uses observation to study the cultural landscape. By observing human structures and practices upon the landscape, researchers try to determine the socio-cultural basis of society within certain time frames.
In general, observation is an inductive method of data generation (see Chapter 1). It works from data to refine hypotheses and produce a theory. This means that as observations are made the hypotheses of the study alter through a process of negative case analysis. When entering the observation phase a researcher should have a general hypothesis as to what will be observed. As the study progresses the researcher looks for cases which do not fit the hypothesis and .seeks to find a new hypothesis that will include these cases. In such a fashion a hypothesis that explains the observations is found. Theories thus emerge from data analysis rather than theories being tested by data analysis. Such a strategy is said to circumnavigate the problem of the researcher deciding upon what is important (e.g., what questions should be asked to test a theory).
Frankfort-Nachmias and Nachmias (1996) suggest that the major advantage of observation as a technique is its directness. Rather than asking people about their views and feelings, you watch what they do and listen to what they say. This directness provides a degree of validity as it concentrates upon what people really do as opposed to what they say they will. In an interview or when answering a questionnaire it is easy for a participant to claim a particular response. For example, it is easy for somebody to claim that they are not racist when completing a questionnaire. As studies have demonstrated, however, this does not always follow in real life. Saying something and doing something are not equivalent. Observation also allows you to record the lives of people as they live it rather than asking them to reflect critically upon their actions in an artificial social encounter such as an interview. This is particularly useful if the group who you are studying are unable to articulate themselves meaningfully or are not given to introspection, for example, young children and mentally impaired people. Observation then is generally a naturalistic (not necessarily naturalist) technique (although it is possible to observe people in an artificial situation created by the researchers). There are two sorts of observation: straight observation and participant observation. These can be divided into overt and covert types (see Table 7.2).
In straight observation the researcher is a visible and detached observer of a situation. For example, if we were studying children's ability to learn how to use maps we might undertake a classroom-based observation study. The researcher would be clearly identified as somebody who was sitting in the class to observe how the children learnt from a map. The researcher would make no effort to undertake a particular role within the classroom and would seek to be tolerated as an unobtrusive observer. By remaining in the background the researcher hopes that children will forget about her or his presence and act normally. By watching the children's actions and listening to their conversations the researcher seeks to understand the process by which a map is understood.
Marshall and Rossman (1995) note that this process of straight observation can range from a highly structured, detailed notation of behaviour guided by checklists (coded schedules) to a more holistic description of events (narratives). In the process of research they suggest that it is usual to progress from the latter to the former, although this is not always the case. For example, in our mapping example the researcher might enter the classroom with a broad area of interest but with no predetermined categories or checklists. After observing the behaviour and discovering recurring patterns of behaviour, the observer might start to construct a more context-sensitive checklist based upon the field notes (see coding data, Section 4.2.1). This checklist is then used to monitor behaviour over time. Some researchers would reject this structuring of the observation process, preferring instead to record their observations in a less constricting manner. Data generated using a formal framework such as checklists impose a large amount of structure on what is to be observed and consequently the data produced. This will increase the ease by which the data can be analysed, but only at the expense of complexity and completeness. Data generated using an informal framework such as note-taking and diary-keeping are relatively complex and more difficult to analyse. It is also possible to undertake straight observation by observing a situation without the observed community knowing or realising the true purpose. For example, you could observe the way farmers harvest their crops by watching over a gateway. In this scenario, your presence may not be suspicious, the farmers are not aware of the reasons for your observation, you remain firmly detached from the harvesting and do not seek to engage the observees in any way. Robson (1993) terms such a role as marginal observation.
In participant observation the researcher seeks to observe events and the behaviour of people by taking part in the activity themselves. This involves going beyond just being present at the same event to include sharing life experiences, becoming a member of the observee's social world. This might entail learning social conventions, the verbal and body language used, and establishing a social role within a community. The philosophy here is, 'if you want to know what it's like being an eco-warrior, then rather than just asking them, become an eco-warrior'. In living and acting out the life of the observed community the researcher becomes the research instrument. The observations made are usually recorded in a diary in a narrative form (unstructured). They will usually be accompanied by on-the-spot analysis and interpretation. Because the researcher is living with the researched community it may be impossible to take a break from the data generation to analyse data before going back into the field (as with our mapping example).
There are two forms of participant observation. The first is the participant as observer. In this form the observer reveals their intentions to the observed group from the beginning. The observer then attempts to build trusting relationships with the host community. This stance has distinct advantages because the observer can ask the community members to explain certain events to them and record events as they happen. If trusting relationships are not developed then the project will suffer greatly. It is important that the observees know why you are conducting the research and that they are happy to let you into their community willingly. If you are not accepted you will be marginalised from the community. The great danger with such an open approach is that the community will modify its behaviour for your benefit. If you are already a member of the group you are seeking to study then there may be problems relating to objectivity. Also if the community has a hierarchical structure then the problem of power relations applies. If you are a lowly member then senior members may be reluctant to confide in you or allow you to observe them. If you are a senior member, lowly members may resent intrusive observation.
In the second form, observation is undertaken covertly. The researched community has no knowledge that they are being studied. This strategy of research raises ethical questions concerning deception (see Section 2.5.2). It is usually justified, however, by arguing that the group would not have agreed to take part otherwise or would have acted differently if they had know about the researcher's presence. Covert research is relatively rare in geographical studies. One example is Moss's (1995) study of the oppression and exploitation in domestic waged labour, and in particular domestic cleaners. She took a job as a cleaner and while working chatted to the other staff about their experiences of the domestic labour market. In general, covert participation observation is frowned upon. It raises serious ethical questions and some researchers have questioned the extent to which the researcher can remain objective and neutral. Others suggest that postponing recording until one is safely alone may lead to incomplete and selective accounts.
It is generally assumed that the observer, although immersed in the world of their observed community, will remain objective and neutral. As such, they will record events as an impartial observer. Participation observation then can be considered 'good science' (Robson, 1993). By gaining access to the private life-world of the observed community the researcher can objectively validate their hypotheses with greater accuracy. As we have discussed elsewhere, some researchers would disagree with such claims to objectivity. They would argue that the research process is inherently subjective. An approach which explicitly recognises the subjective nature of observation is ethnography. This approach is discussed in the following section.
Participation observation requires a large commitment from the researcher both in terms of time and at a personal level. It may mean cutting off your own social relations in order to go and live with another community for an extended period of time. It may also mean getting involved in activities you might otherwise avoid. You should think carefully before committing yourself to such a venture.
Like interviewing, observation can seem deceptively simple. However, there is more to observation than simply watching people in a particular situation. As noted, if you are to observe people in an everyday setting (or laboratory) then a trusting relationship must be developed, unless you occupy a marginal position. At first you may find the process of building relationships difficult and frustrating and you should not expect immediate results. Some experienced researchers might spend months developing an environment conducive to their research. There are differences in opinion amongst researchers as to how much preparation work you should have undertaken before proceeding into the field. Some would argue that the theory should arise out of your observations. However, what you observe and why you are observing is framed within the context of wider knowledge. We suggest that it is best at least to have a good grounding in the relevant literature concerning the phenomenon you are observing, even if your exact theory is not fleshed out.
As we have discussed, there are two main ways of generating and recording your observations in the field. The first is to use a structured approach consisting of coding schemes (checklists). Coding schemes consist of predetermined categories for recording observations. In many ways a coding scheme is similar in format to a questionnaire but responses are interpreted and recorded by the observer. Great care must be taken when designing a coding scheme. The scheme must be exhaustive. That is, all possible relevant behaviours must be available for recording, in the format required (e.g., if timings of each behaviour are required then there should be a space to record these). Here, the keyword is relevant. In most cases not all behaviour is being recorded, only that which is relevant to the particular phenomenon or situation being studied. In general, you are recording specific events and their timings. Box 7.6 details four different sample coding schemes relating to traffic flows. It is extremely unlikely that you will be able to identify observation categories from the literature alone. It will therefore take much piloting time to construct a useful coding scheme. Robson (1993) provides a useful guide to designing a coding scheme (Box 7.7). The main advantage of this approach is that, once a scheme has been designed, it is fast and efficient and it minimises recording error by being highly structured and minimising interpretation. The approach is, however, limited to straight observation and provides a more constricted and limited set of data.
The second way of generating and recording observational data consists of a more holistic account. This approach, although less structured in nature than checklists, consists of more than just random note-taking. Here, the observer records observations using detailed notes rather than just ticking relevant boxes. This provides a richer and more detailed account. These notes should consist of more than random observation, and using a structured format may help you to record all the necessary information you might need later when analysing your records. Robson (1993) suggests structuring each entry into two accounts. The first is a descriptive account relating to the place, time, date, who is there, actions, etc. (see Box 7.8). The second is a narrative account which uses the elements within the descriptive account to construct the 'story' being told and a theory to explain what is being observed. The narrative then extends beyond the descriptive to detail your opinions and hypotheses as to what is happening.
Box 7.7 Considerations in developing a coding scheme
Source: Adapted from Robson 1993.
Your observations should be recorded using a research diary. Remember, recording in a more holistic manner is time-consuming. Where possible, you should record your observations there and then (even just in an abbreviated form to act as memory jogs at a later date). If this is not possible, record your entry at the nearest opportunity (visiting a restroom will provide you with a couple of minutes to jot down notes). You might find it helpful to get into a regular diary entry routine, setting aside a time every day to
Box 7.8 Descriptive observations
Source: Adapted from Robson 1993: 200.
flesh out your descriptive account and work on your narrative account. You should try to write a full account of your observations within 24 hours of the events while they are fresh in your mind. After this period, reflection and subsequent events are likely to cloud your memory and you may misinterpret some of your notes. If you are observing a situation covertly you may find it difficult to find the time and space to note down what you have observed and make attendant notes. In such a situation, note down your observations in your research diary whenever an opportunity arises. This might mean writing down snippets at irregular intervals. An alternative to a written diary might be a taped diary which can be transcribed at a later date. Harry Wolcott (1995) provides
Box 7.9 Some useful tips for undertaking observation studies
Source: Adapted from Wolcott 1995: 96-101.
some useful tips when undertaking an observation study (Box 7.9).
Observation has its merits and limitations. On the plus side is the directness and openness of the method. Observation is sensitive and receptive to the individuals within the study and also allows a certain amount of growth and progression of ideas and focus during the course of the study. While this might be limited in a structured format, ideas can progress during the piloting stage. As such, some lines of enquiry will wither and die while others will be followed up and observed more closely. Observation is the only methodology available to study what people actually do rather than what they say they may do. As such, its value to human geography research is great, although at present it remains relatively under-utilised for two principal reasons: first, observation is time consuming and difficult to instigate; second, it is not a trouble-free technique.
There are two major doubts concerning the use of observation. The first doubt concerns whether deeper meanings can be attached to overt behaviour. Here, we can get drawn into an age-old debate as to whether all behaviours are conscious acts. The second concerns the effect of the researcher upon the behaviour of the observees. Clearly in covert observation this effect is negligible as the observees do not know they are being observed. However, where the observer is known to be recording the actions of the observees, there is a question as to the degree to which the observees might be acting for the benefit of the observer rather than acting as usual. This is a difficult problem to resolve with no way of testing without covert observation. Here, we get drawn into the ethical problems of observing people covertly.
Observation is also very time consuming. This is particularly so for participant observation where strong trusting relationships often need to be developed. There are also methodological doubts concerning the nature of observer recording. It has long been documented that observational techniques suffer from a number of biases. When we observe people or a situation we do so from a selective position. Our position as a researcher but also our individual feelings and experiences affect what we see and also our interpretation of the events unfolding. As such, we have selective attention. Accompanying this is selective encoding. This is where our judgement concerning an action may be clouded and we pigeon-hole an observation into the category where we think it should go rather than the one where it should be placed. The risk of incompleteness or inaccuracy is increased with time and the dangers of a selective memory.
Ethnography seeks to understand the world as it is 'seen through the eyes' of the participants. The aim of ethnography is not to deceive or exploit your respondents but to empathise with them and to gain an understanding of their lives through a genuine trusting relationship. In general, it is a naturalistic approach where you engage with your respondents in their everyday life-world rather than an artificial setting or encounter. As such, ethnography is an empathetic approach that combines aspects of informal conversational interviewing with straight and participant observation. As in participant observation, ethnographers look, listen and enquire about certain aspects of people's lives, noting down and recording encounters. Unlike participant observation, the researcher is trying to describe a situation from the participant's perspective rather than just recording events and interpreting the social processes leading to an event from a neutral position. This is a crucial difference. Like most other approaches ethnography consists of many variants. For example, many ethnographers would use the interviewing and observational methods in the scientific way we have just described. However, an emerging strand of ethnographic work argues that it is impossible to assume an objective role within the research process. Rather it is recognised that the ethnographer is a subjective agent within the research process and therefore cannot adopt a neutral position. Within this alternative ethnography, the use of covert observation is also generally rejected. As such, the researcher exposes themselves in their true role. While this might make the process of developing trust more difficult it also allows a freedom to record data more openly. An exploration of the social politics of the research process is often incorporated explicitly within the research.
Action research has been a distinctive study approach since the mid-1940s. Action research is formulated upon the basis of trying to change a social system at the same time as generating knowledge about it. As such, rather than trying to produce new knowledge by tackling scientific questions, action research aims to create new knowledge through the solving of practical problems. Clearly, remaining objective within action research remains difficult, as a particular action is sought. In general, action research is aimed at practical and technical problems and has a stronger history within physical geography than in human geography. Action-orientated approaches with political and social consequences are generally confined to neo-Marxist and feminist approaches within human geography and are more abstract in nature than action research projects.
Case studies involve studying a phenomenon within its real-life setting. Rather than studying a phenomenon in general, a specific example within time and space is chosen for study. This allows a particular issue to be studied in depth and from a variety of perspectives. There are different types of case study approach (see Box 7.10). No one data generation method is used and quite commonly a number of techniques are employed. We have included case studies within the qualitative approach because, in
Source: Adapted from Robson 1993: 147.
the main, case studies are qualitative in nature, using observation and interviewing as methods of data generation. However, case studies can also be quantitative in nature or use a mix of both sorts of data. Often some of the data will be secondary in nature, consisting of summary statistics relating to the phenomenon or historical accounts relating to a phenomenon. It is therefore possible to undertake a case study which is also ethnographic in nature or action research-led.
Just as with quantitative studies, there are a whole series of archival, qualitative data which you can utilise in your research. These can range from the formally recorded, such as an historical inventory, to the more informal such as letters and photographs. In this section we outline briefly some of the main sources of secondary qualitative data you might use. There are different approaches to analysing secondary data. For example, there are biographical approaches that seek to reconstruct the life histories of people to understand their life-worlds; content analysis which seeks objectively and quantifiably to identify patterns within the text; and deconstructive or hermeneutic approaches which seek to tease out the wider meanings held within the sources. We do not detail these approaches here but rather just present possible secondary sources of data. As such, we do not present different methods of how to use secondary sources. It is suggested that observation and interviewing analysis strategies be applied to the secondary sources (see next chapter). These recognise the need to 'read' the text and to focus upon noting and interpreting what is said or displayed. It should be realised that these data sources are often used together rather than exclusively.
May (1993) reports that secondary sources of qualitative data have been classified in three main ways. First, sources can be primary, secondary or tertiary in nature:
Second, sources can be public or private sources. This refers to the ownership and regulation of a source. This is important in a study as this indicates whether you might gain access to read or observe the source. Those sources in private hands are generally inaccessible without permission. Access to sources in public ownership can also be regulated. This has led Scott (1990) to categorise access into four classes:
If you cannot secure access to a source you may have to reconsider your project objectives. Lastly, sources can be solicited or unsolicited. Solicited sources refer to those sources which have been created with the researcher in mind. An example might be a diary kept by a participant on the instructions of the researcher. Unsolicited sources refer to those that were not deliberately created for the purpose of the researcher. For the purpose of our discussion we have grouped the sources into three classes.
For those researchers interested in the life histories of certain figures or events, diaries, letters, autobiographies and biographies are valuable sources of information. Diaries and letters, in particular, provide a rich source of information concerning the personal feelings, opinions and experiences of the writer. Historically, many ' of note' wrote regular journal entries and letters to friends and many of these journals are held in public or private collections. This is particularly the case with fairly recent historical 'geographical' figures who kept logs of their explorations and whose exploits were sponsored by organisations such as the Royal Geographical Society. By linking the diaries and letters of several individuals together, rather than focusing upon a particular person, a particular event can be studied. Accompanying diaries and letters, many 'key figures' were also members of 'high society', regularly in and out of the press. Press clippings, particularly from the letters pages and gossip columns, are also sources of information. Many figures have also converted their diaries into more formal, edited autobiographical accounts of their lives. These accounts may contain retrospective narratives concerning certain events. Biographies, where a third party was writing at the time or from an historical perspective, may also shed important light on some figures or situations. These accounts benefit from the biographer drawing together several sources to document certain key stages within a life history. Historical geographers have long used such sources. In a recent example, McEwan (1996) studied the travel diaries and biographies of nineteenth-century female travellers to Africa. Using these sources, she has demonstrated the way in which colonial histories have been written almost exclusively by men. Similarly, Royle (1998) has studied the diaries, letters and press clippings of Alice Stepford Green, an aid worker in a Boer War prisoner-of-war camp on St Helena. It should be noted that these sources are necessarily subjective and in the cases of letters and autobiographies will most often be written in such a way as to flatter or argue the case of the writer. Diaries, if private, may allow a more detailed insight into the true, rather than public, opinions of the writer. Biographies similarly may not be objective studies of a person but may be written out of admiration or loathing, or for a fee.
As well as more personal accounts, documentary sources provide a wealth of secondary data. These can range from literature to government and quasi-government reports, academic studies, minutes of meetings, and the 'unofficial' reports of interested parties. While of particular interest to historical geographers, the importance of these sources for contemporary analysis should not be understated. For those interested in studying contemporary society, documentary data can provide valuable insight into the structures and mechanisms of socio-spatial thinking and practice. Literary sources are particularly useful for detailed accounts of the geographical imagination of the writer and her or his characters within the context of an area.
In contrast, more 'official' documents provide detailed accounts of how individuals and institutions thought and reacted to certain geographical contexts. For example, Chris Philo (1987) has studied 'official' documents relating to the siting and running of mental institutions. He is particularly interested in ideas and proposals forwarded in the Asylum Journal, a quasi-academic journal concerning mental health institutions. By sifting through back issues and analysing the articles and editorials contained within, he provides a detailed historical account of asylums in nineteenth-century Britain. When you use official sources it is important that you take account of who wrote the document and ask why it might have been written rather than just studying the content. All documents are subjective and represent a particular viewpoint. For example, Barrat and Cole (1991) note that the 1985 review of social security was undertaken by four committees, with a total of 18 members. Of these 18 committee members half were government (Conservative) ministers; the others were from private industry and right-wing organisations. There were no trade unionists or representatives from any of the leading poverty agencies. Is this report likely to be an objective assessment of the needs of those on social security benefits?
As well as textual sources of secondary information there are visual and auditory sources. Visual data can be classed into two general themes: stills and action. Stills are snapshot recordings of a particular scene. A painting would be an interpretative still, capturing a scene through the 'painter's eye'. A photograph, in contrast, would be a more faithful recording of a scene (although what is photographed and how is also subjective). This is not to say that a photograph should be preferred over a painting, particularly if you are interested in the geographical imagination of a certain period. Each medium records the same sorts of information but in different ways. In recent years, geographers have been increasingly recognising the importance of paintings and photographs as sources of information. For example, a number of studies within cultural and historical geography have analysed landscape paintings (e.g., Daniels, 1988). Visual data are not confined to stills. Films, home videos and television programmes can also provide useful observational data and an insight into the geographical imagination of the camera operator and prospective viewers. Although film studies are still relatively rare in geography, Aitken (1990) has studied the films of the Scottish director Bill Forsyth. Auditory data can also be used as a secondary source of information. Recordings might consist of the taped conversations of another researcher's study or of music and lyrics. McLeay (1995), for example, has studied the sense of place within the music and lyrics of U2.
Whatever the information source, Scott (1990) suggests that before you go ahead and use the data you should assess its usefulness in the context of four things:
In conjunction these four factors relate to the validity of the data - the confidence with which you can use it to draw valid conclusions about a situation or phenomenon. When you analyse and interpret data from secondary sources, you should consider carefully these four factors and keep them in mind throughout the project. In addition, you should also define some criteria by which you select sources and which parts of a source will be used in your research. These criteria should follow the same boundary-making principles that you might follow when deciding upon questions for an interview. Essentially you need to be able to identify what is, and what is not, relevant.
After reading this chapter you should:
In this chapter we have explored how to generate primary qualitative data and to identify and locate secondary sources of data. We have discussed a number of themes, such as strategies for data generation, recording medium and recruitment, that you will need to consider when generating primary qualitative data. Each aspect of the data generation phase needs careful thought and planning. You should only proceed with the generation of data when you are satisfied that you have considered thoroughly the options available to you. In the following two chapters, we will examine how to analyse and interpret qualitative data, first using a handworked example and second using a qualitative data analysis package (NUD-IST).
Cresswell, J.W (1998) Qualitative Inquiry and Research Design: Choosing Among Five Traditions. Sage, London.
Dey, I. (1993) Qualitative Data Analysis: A User Friendly Guide for Social Scientists. Routledge, London.
Eyles, J. and Smith, D.M. (1988) Qualitative Methods in Fluman Geography. Polity, Cambridge.
Marshall, C. and Rossman, G.B. (1995) Designing Qualitative Research, 2nd edition. Sage, London.
Robson, C. (1993) Real World Research: A Resource for Social Scientists and Practitioner-Researchers. Blackwell, Oxford.
Scott, J. (1990) A Matter of Record: Documentary Sources in Social Research. Polity, Cambridge.