An Introduction to Qualitative Research Barbara Pini John Curtin Institute for Public Policy Curtin University of Technology b.pini@curtin.edu.au Overview Part One: Research Process and Design Part Two: Sampling Part Three: Documentary Research Part Four: Participant Observation Part Five: Interviews Part Six: Focus Groups Part Seven: Mixed Methods (Case Studies, Ethnography and Action Research) Part Eight: Qualitative Data Analysis Part Nine: Trustworthiness in Qualitative Research Part Ten: Ethics and Qualitative Research PART ONE Research Process and Design What is Research? “A systematic process of critical enquiry leading to valid propositions and conclusions that are communicated to interested others” (McLeod, 1994). What are some of the key words in this definition and why are they important? Research Proposal Why do we need it? At Curtin University you will need to produce a written document outlining your research for the Faculty of Humanities. This is called an Application for Candidacy. Such proposals are very common in academia (for funding, ethics approval etc). Components of Research Proposal BACKGROUND AIMS/OBJECTIVES/RESEARCH QUESTIONS SIGNIFICANCE METHODS ETHICS FACILITIES AND RESOURCES TIMELINE Taking A Closer Look at Methods METHODS What methods will you use to address the research questions? How many and why this many? (sampling) How will these methods be designed? i.e. How will the study be conducted? Where? How will you gain access? What is the justification for these methods? What questions will be asked and why? What are the limitations of these methods and how will you address these limitations? How will analysis be undertaken? What are the ethical concerns related to these methods and how will these be addressed? All the methodological decisions you make – i.e. how you answer each of the above questions should be tied to the methodological literature and/or the literature in your subject area. Deciding on a methodological approach Ontology: What is the nature of the phenomena, or social reality, that you want to investigate? Epistemology: What might represent knowledge or evidence of the social reality that you want to investigate? Research area: What topic is the research concerned with? Research Question: What do you wish to explain or explore? Ontology What is the nature of things in the social world? For example, are you investigating: Bodies, subjects, objects Rationality, emotion, thought Feeling, memory, senses Motivations, ideas, perceptions Attitudes, beliefs, views Texts, discourses Cultures, society, groups Interactions, social relations Some ontologies are better matched to qualitative research methodology than others (e.g., social processes, interpretations, social relations, experiences etc.) Epistemology What is your theory of knowledge? What are your presuppositions about the nature of knowledge? Examples of epistemological perspectives Positivist Perspectives (also called empiricism) Fundamental claim is that reality is a fixed, measurable entity that is external to people. There exist “social facts.” Aims to to find true, precise and wide-ranging laws of human behaviour which we can generalise to the population as a whole “If it can’t be measured, it doesn’t exist.” Social Constructionism Reality is constructed socially so rejection of “social facts” Aim is to describe the subjective and consensual meanings that constitute social reality. Understanding of social world as “local truths” which cannot be evaluated by external criteria Qualitative and Quantitative Methods Qualitative Makes less use of mathematical techniques. Focus on interpretation by researcher Systematically arranging and presenting information to search for meaning in data collected “Words, not numbers” Usually involves a philosophical stance that human knowledge is, to some extent, contextualised or local. But some form of counting is almost always involved in qualitative analysis. Quantitative Employs statistics or other mathematical operations to analyse data Concepts are assigned numerical values Collects a small amount of data from a large number of people Allows generalisation to wider population Strengths of Quantitative Research It can deal with large numbers of cases It is capable of examining complex patterns of interactions between variables It can make possible the verification of the presence of cause and effect relationships between variables Weaknesses of Qualitative Research Lack of in-depth information Ignores individual perspectives and experiences Limited with topics we know little about Can be built on pre-existing biases of the researcher The case of questionnaires: Language used Ordering of questions Forced response formats; what if ‘it depends…’? Missing data Sampling issues Response rates Lies, lies and damn statistics; torturing your data until it confesses Strengths of Qualitative Research Research done in natural settings Emphasis on informant interpretations and meanings Seek deep understanding of informants world “Thick Description” (Clifford Geertz) Humanising research process by raising the role of the researched High levels of flexibility in research process Weaknesses of Qualitative Research Problems of reliability - The difficulty of replicating findings “Subjectivity” of nature of data collection and analysis Observations may be selectively reported making it impossible to gauge the extent to which they are typical Risk of collecting meaningless and useless information from participants. Problems of objectivity vs detachment (particularly in participant observation but also applies to other methods) Problems of ethics: Entering the personal world of the participant Very time consuming PART TWO Sampling Why Sample? 1. Generalisability To generalise the properties of the sample to the wider population To make conclusions regarding the wider population 2. Pragmatic reasons applied research in organisations often has resource constraints sampling reduces burden on resources (i.e. time and money) usually not feasible to contact the whole population 3. Destruction of test units some research projects (e.g. quality testing) require the destruction of the items being tested e.g. testing cars for safety Sampling Important issues: properly selected samples are sufficiently accurate in most cases to make statements about the population even when population has considerable heterogeneity, larger samples can provide data of sufficient precision upon which to base conclusions In quant research, the characteristics of the sample and actual sample size is more important than the relative size of the sample compared to the population In qualitative research, aim is not statistical representativeness, but representativeness in the sense of gaining access to the full range of views, themes or possibilities in the population Ensuring Representativeness Once the decision to sample has been made, the researcher must identify the target population Must carefully define the target population so that the proper source from which to collect the data can be identified Population any complete group sharing some common set of characteristics can be finite or infinite the group we wish our research to comment upon Types of Sampling Probability Sampling Probability sampling facilitates generalisability occurs when elements in the population have an equal probability of being selected in the sample logic depends on selecting a truly random and statistically representative sample that will permit confident generalisation from the sample to a larger population best for quantitative research Types of sampling Probability sampling How do you select a sample that will look like the population in terms of its demographics? Simple random sampling Systematic sampling Stratified random sampling Cluster sampling Area sampling Double sampling Advantages and Disadvantages of Sampling Methods Simple random sampling high generalisability of findings not as efficient as stratified sampling Systematic sampling easy to use systematic biases are possible Stratified random sampling most efficient and precise would adequately represent strata with low numbers Advantages and Disadvantages of Sampling Methods Cluster sampling goal is to reduce costs of data collection the least reliable among all the probability designs Area sampling type of cluster sampling cost-effective. Useful for decisions regarding location Double sampling offers more detailed information on the topic of study original biases, if any, will be carried over Types of sampling Non-probability sampling Non-probability sampling occurs when elements in the population do not have a pre-determined probability of being selected in the sample ie non-random logic depends on selecting cases rich in information that will permit an in-depth understanding of the research question often used in qualitative research an un-representative sample might be a useful and more stringent test for a law-like hypothesis E.g. Does wealth = health?; sample the very rich who should be extremely healthy and the very poor who would be extremely unhealthy E.g. Helping behaviour and number of onlookers?; sample disaster situations and assess whether ‘law’ holds Convenience Sampling Convenience sampling e.g., interviews on the street; simply asking for volunteers; using clients in clinical or business setting quick, convenient, less expensive not generalisable at all Purposeful Sampling Purposeful sampling or Judgment sampling sometimes the only meaningful way to investigate Useful when you need a targeted sample Includes: Snowball sampling Starting with a small group and asking for further contacts Useful for sensitive topics Quota sampling Population is stratified and numbers within strata are decided Contacts are made until quotas are full Quotas can be proportional or non-proportional to the population Sampling Criteria The appropriate sample design will depend on the following criteria: degree of accuracy required local versus national project need for statistical analysis resources (ie time / money) Sample Size in Qualitative Studies Adequacy of sample depends not so much on the number of cases Depends on the proper specification of the cases to be analysed Redundancy in information is often a sign that the sample size is adequate PART THREE Documentary Research DOCUMENTARY RESEARCH “The good stuff of social science” (Ryan & Bernard 2003) Definition Types of documents Classifications Advantages and disadvantages Analysis Ethics Activities: Journal Article and Document analysis Definition A document in its most general sense is a written text…Writing is the making of symbols representing words, and involves the use of a pen, pencil, printing machine or other tool for inscribing the message on paper, parchment or some other material medium…Similarly, the invention of magnetic and electronic means of storying and displaying text should encourage us to regard ‘files’ and ‘documents’ contained in computers and word processors as true documents. From this point of view, therefore, documents may be regarded as physically embodied texts, where the containment of the text is the primary purpose of the physical medium (J. Scott 1990: 12-13) So…this includes… Photographs Videos / film footage Political speeches Minutes of meetings Plays, novels Media sources Personal documents such as diaries, oral histories Emails [Visual] Any other suggestions? Classification of Documents Primary Written or collected by those who actually witnessed the events they describe. Secondary These are written after an event which the author had not personally witnessed. Tertiary These enable us to locate other sources. They are indexes, abstracts, bibliographies etc. Other Typologies for classifying Documents John Scott (1990) divides documents into four categories according to the degree of their accessibility: Closed Restricted Open-archival Open-published Public and Private documents Solicited and unsolicited documents (Some documents are produced with the aim of research in mind, whereas others would have been produced for alternative uses). Advantages of Documentary Analysis Cost-effective Permanence – particularly for past events Access is usually relatively easy Provides understanding of certain phenomena that is rich in detail and meaning Methodological Problems Authenticity of documents Secondary data When researchers use documents as a source of data, they generally rely on something which has been produced for other purposes and not for the specific aims of the investigation. Credibility What purpose was the document written to? Who produced the document? What was the status of the author? When was the document produced? Representativeness and bias Is the document typical of its type? Does it represent a typical instance of the thing it portrays? Is it complete? Has it been edited? Understanding meaning Is the meaning of the words clear and unambiguous? Are their hidden meanings? Reading/legibility Incomplete sources Gaining access (restricted documents) Analysing Documents Quantitative approaches Content analysis Qualitative approaches “Reading” the text An understanding of the context in which it was produced Examining symbols, hidden meanings What is not contained in the text? What does this mean? Depictions of work in family-genre movies PART FOUR Participant Observation Definition By participant observation we mean the method in which the observer participates in the daily life of the people under study, either openly in the role of researcher or covertly in some disguised role, observing things that happen, listening to what is said, and questioning people, over some length of time. (Becker and Geer 1957: 28) Origins and Links Origins in ethnography Linked with epistomological orientations of ethnomethodology and grounded theory (these methods entail naturalistic investigations of culturally contexted social processes). Pseudo-objective stance of the researcher has largely been abandoned in favour of more personal and subjective accounts of the participant observation experience (see Tedlock, 2000) Traditionally this method has been paired with interviews and document analysis, and more recently with digital photography When to use participant observation Participant observation is especially appropriate for scholarly problems when: Little is known about the phenomenon (a newly formed movement/religion) There are important differences between the views of insiders as opposed to outsiders (e.g. labour unions and management) The phenomenon is somehow obscured from the view of outsiders (mental illness, family life, private interactions) The phenomenon is purposefully hidden from public view (crime and deviance, secretive groups) Strengths of participant observation Natural/unobtrusive. Requires little more than self Can produce rich insights into complex realities Context specific and flexible Holistic. Can incorporate relationships between factors (people, settings, documents). Provides insight into actors’ meanings as they see them Offers advantage of serendipity (See Dennis 1993). Limitations of participant observation Access. Limited options open to the researcher about which roles to adopt or settings in which to participate Commitment. Demanding method and significant personal resources. Danger (potentially) Reliability Observer effects Representativeness of data. Difficulty of generalising from data Ethical issues Easy or Difficult? This method (participant observation) is one that those new to social research believe they can undertake with ease. On first glance it appears to be just about looking, listening, generally experiencing and writing it all down. However, it is more plausible to argue that participant observation is the most personally demanding and analytically difficult method of social research to undertake (May 2001: 153). Strategies to overcome limitations Use multiple observers or teams Search for negative cases Spend an extended time in the field Use insider checking Use outsider checking Repeat observations under varying conditions Be meticulous in recording observations (Alder and Alder 1994) Participant Roles Complete participant Enter the field under pretence or deception Engages fully in the activities of the group or organisation under investigation Advantages are that it can produce more accurate/authentic information and an understanding not otherwise available Problem of recording observations [Visual] Gold (1958) Types of Participant Observation Participant as observer Enter the field setting with an openly acknowledged investigative purpose. Develop relationships with subjects Problem of ‘going native’ but dismissed by some (e.g. May 2001) May encounter hostility – particularly in early stages of research Problem of disengagement from field [Visual] Observer as participant Strictly speaking this would not be regarded as participant observation No lasting contact with people Focus on observation, not on interaction with people Problem is that it does not utilise the strengths of the time in the field to deepen understanding Participant Roles Complete observer Also a non-participant role Role completely removes the researcher from observed interactions Epitomised by laboratory experiments [Visual] Stages in Participant Observation Denzin (1989) Before actual field contacts and observations begin, a general definition of research problem is identified. Select field setting. Make initial contact and establish access. Collect descriptive data on setting and participants. Field work progressing. Informants selected, approached etc. Early theoretical formulations tested. General categories for data analysis are developed. Refining observations. Complex set of propositions developed and tested. Conclusion of study. Role disengagement. Writing of report. Recording Observations Spradley (1980) and Jorgensen (1989) discuss initial observations as primarily descriptive, unfocused and general. After observers become more familiar with their setting and grasp the key aspects of this setting their observations will become more focused and selected. Recoding Observations The participants: Who are the participants? How many are there? How can they be characterised (gender, occupations etc) Where are they situated in relationship to each other? Are there any key groupings or relationships? The tasks: What are the functions of the various groups of people? How are they relating in this setting? What are they doing during the key events or observations? Are these functions formally defined? Do individuals and groups have a variety of purposes for being there? Are there conflicting goals of various groups or individuals? What are these conflicting goals? The setting: Each setting has unique features. What are these? Equipment? Resources? Facilities? Use your senses. The behaviour and the outputs: How do people actually behave during the event? Describe this behaviour in descriptive terms. What are the specific movements made and activities that are carried out? Timing: The timing of the behaviour is described by the time it occurred, the time it takes, and the frequency. Unique causes or consequences: What unique occurrences affected the people, tasks, setting, behaviours, output and timing? Cunningham (1993:141) Video recording A “privileged gaze” (Atkinson & Hammersley, 1994). Purposes (Paterson, Bottorff & Hewat, 2005): Allows decontextualised sequencing of minute behaviours, concurrent behaviours, nonverbal behaviours and conversational analysis that are difficult to observe in real time To document the research process and check for observer effects To direct methodological decisions To enhance the validity of the researcher’s interpretation of observations Compared to participant observation where videorecordings are not used, relationships less important for the collection of data, but more important for getting consent to participate Participant Observation and Ethics Disguised or covert research has come under significant attack. Boundaries between covert and open research are not necessarily clear cut. Protection of informants/respondents Why have people shared information? Ethical dilemmas do not cease when you leave the field. PART FIVE Interviews Interviews 1 Defining Interviews Types of Interviews Advantages and disadvantages Design questions Sampling issues Types of interview questions Interview skills Defining Interviews A conversation with a purpose (Kahn and Cannell 1957:149) Silverman (1993) talks about us living in an ‘interview society’ Estimated that 90 per cent of all social science investigations use interviews in one way or another (Briggs 1986) Types of Interviews - Structured Many are formally structured. Associated with questionnaire research (oral questionnaire); also used in some job interviews Each person asked the same question in the same way so that any differences between answers are held to be real ones and not the result of the interview situation itself. No deviation from question order or wording of questions. No adjusting for level of language. No clarifications or answering of questions about the interview Types of Interviews: Semi-structured Questions are normally specified, but the interviewer is freer to probe beyond the answers. Questions may be reordered during the interview. Level of language may be adjusted. Interviewer may add or delete probes. Allows people to answer more on their own terms, but still provides a structure for comparability. Sometimes called semi-standardised. Most typically used in qualitative studies (Rossman and Rallis 1998: 124) Types of Interviews: Unstructured Includes life-history, biographical and oral history interviews Sometimes called informal, non-standardised Provides qualitative depth in allowing subject to talk about topic within their own frame of reference Issues “Increasingly, qualitative researchers are realizing that interviews are not neutral tools of data gathering but active interactions between two (or more) people leading to negotiated contextually based results” (Fontana and Frey 2003: 62). Impact of identities of researcher and participants should be considered Advantages of interviews One of the most flexible/responsive methods available as different types of interviews can be engaged for different research problems. Ability to explore additional research questions / issues if they arise (semi-structured / unstructured only) Ability to gain rich and descriptive data; ideally suited to examining topics in which different levels of meaning need to be explored. Most participants will accept an interview readily. They are likely to be familiar with interviews. Ability to follow up research participants for clarification or further exploration Disadvantages of interviews Bias and subjectivity which, in turn, affects validity and reliability of data Generalisation problem Process of data collection, transcribing and analysis from each participant time-consuming; thus, sample size generally not large In reporting results, tendency of researchers to focus on quotes which are dramatic, unusual or interesting, rather than typical Design Questions The recommended duration for an in-depth interview is one hour and a half, but may be varied according to the situation and respondent (Burgess 1984, p.120). A write-up of observations may be completed following each of the interviews (Burgess 1984, p.119). Increased rapport is likely to be facilitated through follow-up visits which also will improve the quality of data produced (Whyte 1984, p.114; Lee 1993, p.113). What Counts as Data? Utterances only? Non-verbal aspects of the interaction? Written notes / tape-recordings? My own memories and unwritten interpretations of the interview? Diagrams, pictures, drawings, charts and photographs produced during the interview? NB: Absolute objectivity is a myth!!! Researchers continually make judgements about what to write down or record, what they have observed, heard and experienced and what they think it means (Mason, 2002). Sampling (as per Part 2) Specifically, Minichiello et al. (1995, p. 162) describe the process as it applies to in-depth interviewing as ‘selecting informants on the basis of relevant issues, categories and themes which emerge in the course of conducting the studies’. Types of Interview Questions (Kvale 1996: 133) Introducing questions E.g. “Can you tell me about…”? Etc. Probing questions E.g. “That’s interesting. What else can you tell me about…”? Specifying questions E.g. “Can you give me an example of…”? Direct questions E.g. “Earlier you said… How does that relate to…”? These may need to come later in the interview; may be slightly confrontational or ask for clarification of discrepant information Indirect questions (useful when trying to avoid social desirability bias) E.g. “What should someone else in that situation do…”? Structuring questions E.g. “I would now like to introduce a new topic…” Silence – just a nod or a pause Interpreting questions Rephrasing an answer, more speculative questions E.g. “So does that mean…”?; “Are you saying…”?; “Would I be right in interpreting that as…?” Interview Skills The good interviewer needs to be attentive. The good interviewer is sensitive to the feelings of the informant. The good interviewer is able to tolerate silence. The good interviewer is adept at using prompts. The good interviewer is adept at using probes. The good interviewer is adept at using checks. The good interviewer is non-judgemental. (Denscombe 1999:135) PART SIX Focus Groups Overview Definitions History Common uses Advantages and limitations Interviews versus focus groups Recruiting for a focus group The role of moderator Analysing focus group data Ethics of focus group research The future? Activity Definitions ‘The hallmark of focus groups is the explicit use of the group interaction to produce data and insights that would be less accessible without the interaction’ (Morgan 1988: 12). Kitzinger (1994: 159) ‘group discussions organised to explore a specific set of issues’. History Originally called focussed interviews. Origins in the Office of Radio Research at Columbia University in 1941 when Paul Lazarsfeld invited Robert Merton to assist him in the evaluation of audience response to radio programs. Method increasingly used in social sciences and marketing (Catterall and Maclaran 1997; Green 1999). Morgan (1988) says most common form of marketing research. Common uses of focus groups Obtaining general background information about a topic of interest Generating research hypotheses that can be submitted to further research and testing using more quantitative approaches (Stimulating new ideas and creative concepts) Diagnosing the potential for problems with a new program, service or product Generating impressions of products, programs, services, institutions or other objects of interest. Learning how respondents talk about the phenomenon of interest. This, in turn, may facilitate the design of questionnaires, survey instruments or other research tools that might be employed in a research project. Interpreting previously obtained research results. Advantages of Focus Groups (Quible 1998, Albrecht, Johnson and Walther 1993; Stewart and Shamdasani 1990). time and cost efficient direct interaction between researcher and researched, respondents can qualify responses, researcher can observe non-verbals large amounts of rich data in the respondents’ own words. The researcher can obtain deeper levels of meaning etc. synergistic flexible especially useful for groups with limited literacy, results are readily understood. synergism snowballing stimulation security Spontaneity Disadvantages of focus groups (Quible 1998, Albrecht, Johnson and Walther 1993; Stewart and Shamdasani 1990). Small number of respondents and convenience recruiting limit generalisability. Responses may be subject to group-think, especially if there are dominated or opinionated members. More reserved members may be overlooked (see MacDougall and Baum 1997). The open-ended nature of responses may make summarisation and interpretation difficult. Potential for moderator bias Cost (moderator fee, facility rental, recording and transcribing, data analysis, participant incentives) Subjects’ conformity Designing and Conducting Focus Groups “The experience of using the focus group as a qualitative research method can be compared with that of the tightrope walker: when things go well there is a feeling of exhilaration, when they go badly….it’s a long drop!” Pugsley 1996:126 Interviews and Focus Groups Focus groups are not appropriate when: 1. Detailed probing of an individual’s behaviour, attitudes or needs is required 2. The subject matter under discussion is likely to be of a highly confidential nature 3. The subject matter is of an emotionally charged or embarrassing nature 4. Certain strong, socially acceptable norms exist and the need to conform in a group discussion may influence response 5. A highly detailed (step-by-step) understanding of complicated behaviour or decision-making patterns is required 6. The interviews are with professional people or with people on the subject of their jobs.(Hawkins et al, 1994; 554-444). Steps in Design and Use of Focus Groups Problem definition/formulation of research question Identification of sampling frame Identification of moderator Generation and pre-testing of interview guide Recruiting the sample Conducting the group Analysis and interpretation of data Writing report Designing Focus Groups: How do you recruit participants? Time-consuming Krueger (1988: 94) refers to ‘recruiting on location’ Morgan (1995) says recruitment is the single most common source of failure he has encountered in focus group research. How to avoid problems: repeated contacts, offering incentives, over-recruiting. (Morgan 1988 suggests over-recruiting by 20%). May be recruited by existing social networks, word of mouth or advertising. Designing Focus Groups: How many people in a group? Literature differs but researchers highlight that size should be related to research topic/purpose Group sizes of 4 to 12 are recommended with an ideal group the size of 8 (Morrison & Peoples, 1999; Diloria 1994 et al.) 6 to 9 (Garrison et al. 1999) Generally 8-12 individuals (Stewart and Shamdasani 1990) 6 to 10 (MacIntosh 1993) Up to 15 (Goss and Leinbach 1996) Smaller groups may be dominated by one or two members Larger groups may be difficult to manage, obtain the perspectives of all members Designing Focus Groups: Who should make up your focus groups? (Sampling as Part 2) The issue is sample bias not generalisability (Morgan 1988) Typically use purposefully selected samples. Heterogeneous or homogenous. Do you want to make comparisons between different groups? Key question: Would these groups normally discuss the topic in day-to-day interaction? (Morgan 1988). Designing Focus Groups: How many groups? Depends on approach; research questions; time and budget constraints Some use only one meeting with each of several focus groups (e.g. Burgess 1996) Others use follow-up meetings (e.g. Pini 2002) Multiple groups of similar participants are usually necessary for data to be valid Most questions can be answered by 6 to 8 groups, although 4 groups are adequate for some studies and 50 are needed for more extensive studies (Krueger 1994) One important determinant is the number of different population subgroups required (Morgan 1988) Designing Focus Groups: How long should they last? One and a half to two and a half hours (Stewart and Shamdasani 1990) Consider moderator as well as participant fatigue. Keim et al (1999) study used one hour groups for children and found this was too long. Designing Focus Groups: Developing a focus group guide In general, keep the number of broad concepts examined in a focus group moderate so that each can be examined in detail. Tend to be general in nature and open-ended. Moderator will be improvising comments and questions within the framework. Opening question is one that everyone answers at the beginning of the focus group. Designing Focus Groups: What is the role of the moderator? Smith (1995) recommends two moderators for better control of group cohesion and more thorough observation of group dynamics. Morgan (1988: 49) favours approach he calls ‘highly nondirective focus groups’ or what he says are ‘self-managed groups’. Moderator needs to have both strong interviewing and observational skills (McDonald 1993). What is the role of the moderator? Consider advantages of high moderator involvement: Can cut off unproductive discussion Ability to ensure all topics covered Can adjust discussion Consider disadvantages of high moderator involvement A biased moderator will produce data that reproduces these biases Does not allow new / unanticipated issues to emerge Consider advantages of low moderator involvement Can assess participants’ own interests Participants can bring up controversial topics/topics not considered by moderator Consider disadvantages of low moderator involvement Relatively disorganised in content and so more difficult to analyse Some topics may never come up Analysing Focus Group Data A typical two hour session yields an average of 40 to 50 transcript pages. Morgan (1988: 64): The group is the fundamental unit of analysis and the analysis should begin in a group-bygroup progression. Hyden and Bulow (2003) stress the need to examine not only pure content, but who is saying what Krueger (1993) says read transcripts/summaries and: Consider the words Consider the context Consider the internal consistency Consider the specificity of responses Find the big ideas Consider the purpose of the research Krueger (1993) Quality control in focus group research. . In D. L. Morgan, Successful Focus Groups, Sage, Newbury Park, 65-85 Ten quality factors in focus group research: Clarity of purpose Appropriate environment Sufficient resources Appropriate participants Skilful moderator Effective questions Systematic and verifiable analysis Appropriate presentation The future? Emerging literature on virtual focus groups. Who and what are being researched in online focus groups? Are online groups going to replace traditional focus groups? Are respondents who they say they are? Do respondents in online groups really interact with each other? (See Sweet, 2001; Murray 1997; Walston and Lissitz 2000. ) PART SEVEN Mixed Method Approaches to Qualitative Research Integrative Models Model 1 Quantitative or qualitative approach is used independently of the other Model 2 Qualitative approach is used to develop quantitative measurement scales Model 3 Qualitative approach is used to interpret quantitative findings Model 4 Quantitative findings approach is used to interpret qualitative Specific ‘Mixed-Methods’ Approaches to Research We have covered the 4 major qualitative methods, but some recognised methods combine these approaches Case study method Ethnography Action research Case Study Method Distinct from “a case” (the object of study) Features (Yin 2002) Single example of a phenomenon of interest (organisation, part of an organisation) May involve more than one ‘case’ (multi-site study) but comparisons between them are a feature of the research (separately identifiable) May also only involve a single case (within-site study) Used in law (illustrative cases), health (unusual or interesting illnesses), psychology (Freud), political science (case reports) and business (organisations with defined features) Case Study Method (cont.) May be qualitative or quantitative or both, but relies on multiple sources of evidence where data triangulates in a converging fashion For qualitative case studies, Yin (1989) suggests 6 types of information: Observations, interviews, audio-visual material, documents, archival material, physical artifacts Case Study Method (cont.) Challenges in case study method: number of cases selected – the more cases, the more diluted the overall analysis. Typically no more than 4 The issue of Single case study research Choosing the case(s) – strong rationale for purposeful sampling strategy is important Deciding the ‘boundaries’ of a case – how it might be constrained in terms of time, events and processes Presents general propositions but not broadly generalisable, but should it be? The Debates on Generalisability Critique of the importance and goal of generalisability (the discovery of laws, Lincoln & Guba) Attribute the belief to positivism Critique the view that we can produce knowledge that is free of time and context Argue that the choice is not about searching for general laws OR studying the unique, but something in between i.e. stating conclusions from studying one context that might hold in another context, ‘working hypotheses’, the ‘fit’ between one case study and another, generalising not about what is, but what may be or what could be Ethnography Genesis in cultural anthropology Argued to be not one particular method but a style of research that is distinguished by its objectives Some overlap with participant observation as this is the predominant technique used. However, interviews and documentary methods also often utilised. Definition: To understand social meanings and activities of people in a given setting “a description and interpretation of a cultural or social group or system” (Cresswell 1998) Mostly used in anthropology and sociology, but also health sciences, education, rarely in business Ethnography (cont.) Has a number of features distinct from other methods: Sees the world through the eyes of those being researched, allowing them to speak for themselves Researchers immerse themselves in the setting and become part of the group in which they are interested Aims to provide understanding of the meaning and importance that members of the group impart to their own behaviour and that of others Ethnography (cont.) Key terms Fieldwork – collecting data in a particular setting Gatekeepers Key informants Reciprocity Reactivity or reflexivity Action Research – a ‘participatory approach to enquiry’ Definition: “Disciplined enquiry (research) which seeks focused efforts to improve the quality of people’s organisational, community and family lives” (Calhoun 1993). Key tenets: Processes are rigorously empirical and reflective (research is self-conscious) Research engages people who have traditionally been called “subjects” as active participants in the research process Research results in some practical outcome related to the lives or work of the participants Democratic, equitable, liberating, life enhancing Operates at intellectual, as well as social, cultural, political and emotional levels Action Research (cont.) Has much in common with community development and practitioner research Routine is look, think, act… or observation, reflection, action… However, not linear, neat or orderly, rather routine can work backwards, in repetition and revision, can leap frog stages and sometimes make radical changes in direction PART EIGHT Qualitative Analysis Qualitative Analysis “If the sociologist or the biographer is like a detective, and collecting data is like detection, then analysing data is akin to the culminating stages of the criminal justice process. It has the same potential for abuse, and therefore requires similar safeguards. Unfortunately, whereas in criminal justice the adversarial roles of prosecution and defence can be allocated to different people, in qualitative analysis the analyst often has to play both roles” Dey, 1993 Your approach to qualitative data analysis will be informed by your epistemological stance (Chua, 1986) Three major philosophical positions in qualitative analysis: 1. Positivist Evidence of formal propositions, quantifiable measures of variables, drawing of inferences about a phenomenon from a representative sample to a stated population Eg Content analysis – simply counting words / phrases (e.g., political speeches; media articles) Relational content analysis – more in-depth; considers meanings of excerpts and the relationship between them 2. Critical Main task is social critique; helps to eliminate the causes of unwarranted alientation Approaches to qualitative data analysis 3. Interpretive Knowledge is gained through social constructions such as language, consciousness, shared meanings etc. Does not predefine dependent and independent variables; seeks to understand phenomena through the meanings that people assign to them E.g. Grounded theory – inductive, theory-building approach Narrative analysis – preserves the story rather than fragmenting data Phenomenological approaches – particularly concerned with generating meanings and gaining insights into phenomena Discourse analysis – assumptions and meanings underlying spoken language, ‘main line story’ Conversation analysis – highly specialised, based on linguistics (Atkinson & Delamont argue divorced from wider issues such as identity, interactions and social encounters) Interpretivist Approaches “Interpretivist positions are concerned with how the social world is interpreted, understood, experienced, produced or constituted. While different versions of qualitative research might understand or approach these elements in different ways (e.g. focus on social means, or interpretations, or practices, or discourses, or processes, or constructions), all will see at least some of these as meaningful elements in a complex – possibly multi-layered and textured – social world” Mason 2002 How to ‘Read’ Data 1. Literally actual words and language used – the literal content of the data The sequence of interaction – in the case of interviews, who speaks when? In the case of visual data – style, layout, literal form Although these categories may be important, few researchers will stop here. Some argue that purely objective description is not possible because the social world is always interpreted and what we see is shaped by how we see it! The How to ‘Read’ Data 2. Interpretively Constructing or documenting a version of what you think the data mean or represent Reading through or beyond the data E.g. implicit norms or rules with which an interviewee is operating Discourses that influence people Versions or accounts of how people make sense of social phenomena How to ‘Read’ Data 3. Reflexively Locates the researcher as part of the data generated Seeks to explore the role and perspective of the researcher in the process of generation and interpretation of data Captures or expresses the relationships between researcher and data E.g. response to a certain situation in fieldnotes (empathy, shock, agreement, amusement) Stages in the Analysis of Qualitative Data Stage 1: Immersion The researcher intensively reads or listens to material, assimilating as much of the explicit and implicit meaning as possible Stage 2: Categorisation Systematically working through the data, assigning coding categories or identifying meanings within the various segments / units of the ’text’ Stage 3: Reduction questioning or interrogating the meanings or categories that have been developed? Are there other ways of looking at the data? Do some codes mean the same thing? Stage 4: Triangulation sorting through the categories. Deciding which categories are recurring and central and which are less significant or are invalid or mistaken Stage 5: Interpretation making sense of the data from a wider perspective. Constructing a model or using an established theory to explicate the findings of the study Making a Convincing Argument about your Data (Mason, 2002) Making a convincing argument will be influenced by the research questions you originally posed, the focus of the research and the kinds of data generated Major categories of arguments: 1. Arguments about how something has developed – a meaningful process of development or a story or an ‘archaeology’ 2. Arguments about how something works or is constituted – how and why social phenomena work (but not cause and effect) 3. Arguments about how social phenomena compare – meaningful points of comparison in different contexts 4. Arguments about causation and prediction – the effects of variables on each other; not widely used by qualitative researchers Techniques to Ensure Qualitative Data is Credible (Cresswell, 1998) Triangulation – checking one source of data against another Leaving an audit trail – clear records about how the analysis was conducted Member checking – have more than 1 researcher conduct analysis and compare interpretations Checking for researcher effects – do results differ across researchers (e.g. focus groups with managers) Checking the meaning of outliers – find explanations for ‘extreme cases’ Searching for contradictory or negative evidence Replicating your findings (more difficult in qual research) Getting feedback from participants Seeking feedback from peers Then and Now Coding historically done by hand – marker pens, cutting and pasting (scissors and glue), sorting and shuffling file cards Early to mid 1980s marked the emergence of basic data programs for storing and accessing text Now at least 25 different programs – some specifically QDA, some more general purpose What software can and cannot do Software is a tool to help analyse qualitative data. It can: Store transcripts / other text Store codes Search and retrieve segments of text Link data segments to each other, forming categories, clusters or networks of information Make notes Edit Conduct content analysis (count frequencies, sequences or location of words) Graphically map concepts It cannot read the text and decide what it means Similarly, it cannot substitute for learning data analysis methods Advantages of Software for Qualitative Data Analysis Consistency all the places where a code or combination of codes applied, therefore not missing data that contradicts incorrect hypothesis Speed Once program is learned and data is set up, much quicker than manual coding (particularly re-sorting, re-defining codes and creating matrices of codes) Graphic maps Helps visualising and therefore thinking and theorising about possibilities and alternatives Advantages of Software for Qualitative Data Analysis Disadvantages Speed and ease of use can make us lazy Autocoding (searching key words) may encourage shortcuts May encourage ‘quick and dirty’ research with premature theoretical closure Direct representation of hierarchical relationships (as opposed to ‘circular loops or unstructured networks) encourages hierarchical thinking May tempt researchers to skip over the process of ‘proper’ learning NB As with every methodological decision, if you DO decide to use software you should justify it in terms of the literature, acknowledge its limitations (again referring to the literature) etc. PART NINE Trustworthiness in Qualitative Research Definitions Reliability – generally understood to concern the replicability of research findings and whether or not they would be repeated if another study, using the same or similar methods, was undertaken. Validity – traditionally understood to refer to the ‘correctness’ or ‘precision’ of research. Internal validity: ‘investigating what you claim to be investigating’ (Arksey and Knight 1999) External validity: ‘the abstract constructs or postulates generated, refined or tested’ are applicable to other groups within the population (LeCompte and Goetz 1982) Examining the terms: validity and reliability Validity, reliability and generalisability have been called the ‘holy trinity’ of the sciences (Kvale 1996). What assumptions are inherent in emphasising the importance of validity and reliability? To use or not use the terms validity/reliability Some qualitative researchers (e.g. critical theorists, feminist theorists, post-structural theorists) have criticised the use of these terms in qualitative research. On what basis? A range of qualitative researchers have denied the relevance of validity/reliability to qualitative research and argued that qualitative research has its own procedures and processes for judging and attaining validity/reliability. There is still ongoing debate about this in the literature. However, many qualitative researchers utilise the terms validity and reliability, and describe a range of strategies to enhance both in their work. Your view about reliability/validity will depend upon your own epistemological, theoretical and methodological position. Examples of moving away from the terms validity/reliability Smith (1996) argues that internal coherence (or lack of it) would be the most appropriate way of assessing qualitative research. Rather than being concerned, for example, with the representativeness of the sample, you should concentrate on whether it was internally consistent and coherent. Does it present a coherent argument? Does it deal with loose ends and possible contradictions in data? Are the interpretations that the researcher makes warranted by the data presented? Does the report deal with alternative readings? Leininger 1994 Credibility Confirmability Using repeated experiences, events to identify patterns etc Saturation Understanding data within holistic contexts (participants’ contexts) Recurrent patterning Repeated direct participatory and documented evidence observed or obtained from primary sources Meaning-in-context Ensuring that the researcher uses active listening, reflection and empathic understanding to grasp what is ‘true’ to informants in their lived environment Full immersion by the researcher in the phenomena being studied; getting ‘thick’ data to know fully what is being studied Transferability Examining general similarities of findings in similar environmental situations Popay et al 1998 The privileging of ‘subjective meaning’ – the research illuminates the subjective meaning, actions, and context of those being researched. Responsiveness to social context – the research design is adaptable/responsive to real-life situations. Purposive sampling – the sample produces the knowledge necessary to understand participants’ location in structures and processes. Adequate description – the reader can interpret the meaning and context of what is researched. Data quality – different sources of knowledge about the same issues are compared. Theoretical and conceptual adequacy – the research describes the process of moving from the data to intepretation. Typicality – claims are made for logical rather than probabilistic generalisations. Lincoln and Guba (1990) credibility applicability consistency neutrality Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Triangulation Denzin (1970) identifies four types of triangulation to confirm emerging findings: Multiple investigators, Multiple sources of data (time, space, person) Multiple data collection methods to confirm emerging findings (observations, interviews, focus groups etc). Multiple theoretical perspectives. This involves using several perspectives to examine the same set of data. Few investigators use this technique. Janesick (1994: 214) adds interdisciplinary triangulation. The greater the convergence attained through triangulation the greater confidence in findings. First argued by Foreman (1948) ‘to establish validity through pooled judgement’. Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Member checks Taking data and tentative interpretations back to the people from whom they were derived and asking if they were plausible. Also called member tests of validity and host verification Can be conducted continuously and both formally and informally (e.g. at the end of an interview by summarising data and allowing respondent to immediately correct errors of fact, by giving copies of various parts of your report to different stakeholders and asking for comment etc). Not necessarily free of bias or problems (see St. Pierre 1999 and Sandelowski 1993 ). Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Peer review/examination Discussions with colleagues regarding the process of study, the congruency of emerging findings with the raw data, and tentative interpretations. This peer is outside the context but has some general understanding of the study and methods etc. Criticised by some e.g. Morse (1994) Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Researcher’s position/reflexivity Critical self-reflection by the researcher regarding assumptions, worldview, biases, theoretical orientation, and relationship to the study that may affect the investigation. Keeping a journal/type of diary. Includes information about yourself as a researcher, your schedule, insights, decisions and justifications for decisions Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Adequate engagement in data collection Adequate time spent collecting data so that data become ‘saturated’. This may involve seeking discrepant or negative cases. Negative case analysis (or analytic induction) involves addressing and considering alternative interpretations of the data, particularly noting pieces of data that would tend to refute the researcher’s reconstructions of reality. Prolonged engagement will build trust and develop rapport and the impact of your presence may diminish. Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Attention to sampling Maximum variation: Purposefully seeking variation or diversity in sample selection to allow for a greater range of application of the findings. Do not suppress/ignore the deviant/the different. Allows for and opens up a range of realities and perspectives. Strategies for Promoting Validity and Reliability in Qualitative Research (Merriam 2002). Audit trail A detailed account of the methods, procedures and decision points in carrying out the study. Lincoln and Guba (1985) give six categories of audit trail materials: (1) raw data (interview guides, notes, documents) (2) data reduction and analysis products (3) data reconstruction and synthesis products (e.g. data analysis sheets) (4) process notes (journal) (5) materials relating to intentions and dispositions (inquiry proposal, journal, peer debriefing notes) (6) information relevant to any instrument development. Strategies for Promoting Validity and Reliability in Qualitative Research Cont. (Merriam 2002). Rick, thick descriptions Providing enough description to contextualise the study such that readers will be able to determine the extent to which their situation matches the research context, and hence, whether the findings can be transferred. Lincoln and Guba (1985, 125) state that ‘The description must specify everything that a reader may need to know in order to understand the findings (findings are NOT part of the thick description). PART TEN Ethics and Qualitative Research Principles of Research Involving Human Subjects 1. Respect for persons 2. Beneficence treating others as autonomous agents having rights and freedom not a means to an end free, voluntary and informed consent privacy and confidentiality research should be for the good of the subject either directly or indirectly through benefiting society possible benefits are maximised and risks minimised impasse often develops between social good and individual rights 3. Justice benefits and harms are to be distributed fairly vulnerable groups such as cognitively impaired and mentally ill, their above average rates of institutionalisation and their dependency on others, have made them a convenient subject pool for research who should participate in research poses significant challenges to policy formation Key Ethical Concepts Protection of participant Informed consent Use of deception Debriefing participants Right to withdraw Privacy and confidentiality Protection of Participants 1. Ensure minimal risk must apply the cost-benefit-ratio risks unlikely to be greater than any encountered in normal lifestyle must minimise negative outcomes 2. Strategies obtain advice from professionals screen vulnerable participants monitor unforeseen negative events debrief participants about research conduct long-term follow-ups have counselling or support available Informed Consent Rests on 4 elements competence, information, understanding of that information and voluntariness but… cannot be established in many important areas of research e.g. critically ill, demented, minors Social contract rests must on a mutually agreed contract reveal all aspects that might influence the decision to participate Strategies inform of the general aims of the associated costs and benefits consent forms project Use of Deception Subjects are not given an opportunity to provide their informed consent to participation before data collection. Examples include covert observation or subject knows they are participating in research but not the nature of the research. Problems deprives participant of the right to informed consent but… providing all information is likely to influence behaviour and therefore results should be avoided if possible Guidelines governing deception in research: no more than minimal risk to subjects rights and welfare of the subjects will not be affected research cannot practicably be carried out without the deception where appropriate, subjects are provided pertinent information about the research after participation (debriefing) Debriefing Participants Rationale traditional solution to deception problems participation considered an educational experience Strategies give all information needed and requested discuss their experience of the research provide contact details Right to Withdraw Rights can withdraw consent without any penalty can request data be destroyed Controversy use of captive audiences (e.g. students, military, prisoners, employees) use of incentives Privacy & Confidentiality Avoid the use of sensitive questions Do not record names if possible Code questionnaires Warn prior to data collection what identifying information will be kept Explain confidentiality procedures Research ethics and the Internet Dilemma based on three facts: Informed consent is not required for data to be collected from the public domain (naturalistic observation). The internet is a public domain Many online communications (email; discussion groups, chat rooms, newsgroups etc) cultivates an expectation of privacy The ease and attractiveness of Internet research renders the medium vulnerable to misuse. Guidelines: When subjects are recruited online, need secure server, secure protection of information during the study and removal of the records upon study completion When using data from online discussion groups, removal of any references to identity, web site or group, location and time of post is necessary for confidentiality Research with vulnerable populations – An EXTENSIVE methodological literature exists on undertaking research with specific populations and the ethics and practice of research with these populations. For example: Indigenous people Youth People with disabilities Migrants The aged