Summary Qualitative Research methods (Maruster & Gijsenberg) part 1

Deze samenvatting is gebaseerd op het studiejaar 2013-2014.

Chapter A : The Basics of Qualitative Research

Qualitative research is defined as the way in which people being studied understand and interpret their social reality is one of the central motifs.

Key characteristics of qualitative research:

• Aims that are directed at providing an in-depth and interpreted understanding of the social world of research participants.
• Small samples that are purposively selected based on a salient criterion.
• Data collection methods which involve close contact between the researcher and the research participants.
• Detailed and rick data.
• Analysis which is open to emergent concepts and ideas.
• Outputs which tend to focus on the interpretation of social meaning.

Historical development of qualitative research:

1. Development of empiricism and positivism

Empiricism = all knowledge about the word originates in our experiences and is derived through the senses.

Positivism = social world can be studied in terms of invariant laws. Characteristics of positivism are:

• Methods of the natural sciences are appropriate for the study of social phenomenon.
• Only those phenomena which are observable can be counted as knowledge.
• Knowledge is developed inductively through the accumulation of verified facts.
• Hypotheses are derived deductively from scientific theories to be tested empirically.
• Observations are the final arbiter in theoretical disputes.
• Facts and values are distinct, making it possible to conduct objective research.

2. Development of interpretivism

This concept states that knowledge about the world can be acquired in other ways than direct observation. Arguments in favor for this concept are:

• Perception relates not only to the senses but to human interpretations of what our senses tell us.
• Our knowledge of the world is based on ‘understanding’ which arises from thinking about what happens to us.
• Knowing and knowledge transcend basic empirical research.
• Distinctions exist between ‘scientific reason’ (causal determinism) and ‘practical reason’ (moral freedom).

Weber argued that the key difference between natural (positivism) and social science (empiricism) is that the purpose natural science is to produce law-like propositions whereas the aim of social sciences is to understand subjectively meaningful experiences. Interpretivism stresses the importance of both interpretation and observation in understanding the social world.

3. Development of qualitative research methods and challenges to the scientific method

Since positivism became dominant in quantitative research, qualitative research was criticized as ‘unscientific’. In response to that some qualitative researchers tried to formalize their methods, called the modernist phase. However, in 1970 positivism was questioned concerning:

• Whether it is possible to control variables in research involving human subjects.
• Whether the elimination of contextual variables in controlled experimental conditions is an appropriate way to study human behaviour.
• Whether it is appropriate to disregard the meaning and purpose of behaviour in controlled experimental studies.
• Whether overarching theories of the world have any relevance to human lives.
• Whether emphasis on hypothesis testing neglects the importance of discovery through alternative understandings.

These challenges encouraged the use of qualitative research. In the last decade of the twentieth century qualitative methods were more widely accepted within British psychological research practices. Since then, there was an explosion of interest in qualitative research.

Ontology = a philosophy that tries to categorize all existing things. Main ontological debate is about whether there is a captive social reality and how this should be constructed. Three distinct positions:

1. Realism: there is a distinction between the way the world is and the meaning and interpretation of that world held by individuals.
2. Materialism: there is a real world but only material features of that world hold reality. Thus, values, beliefs or experiences do not shape the material world.
3. Idealism: reality is only knowable through the human mind and through socially constructed meanings.

Epistemology = ways of knowing and learning about the social world. Three main issues are:

1. The relationship between the researcher and the people being researched. Natural science view, phenomena are independent and unaffected by the behaviour of the researcher. Thus, the researcher can be objective and the investigation is value free. Social science view, the relationship between researcher and the people being researched is interactive. Thus, the researcher cannot be objective and findings are value-mediated. ‘Empathic neutrality’ is a position that recognizes that research cannot be value free but which advocates that researchers should make their assumptions transparent.
2. Issues around ‘truth’. Natural science view, there is a match between observations or readings of the natural world and an independent reality. Social science view, if several reports confirm a statement then it can be considered true as a representation of a socially constructed reality. Pragmatic view, an interpretation is true if it leads to actions that produce the desired or predicted results.
3. The way in which knowledge is acquired. Inductive view, looking for patterns and association derived from observations of the world (using evidence for a conclusion). Deductive view, propositions or hypotheses are reached theoretically through a logically derived process (using evidence in support of a conclusion).

For the differences between ontological and epistemological stances see Box 1.3 (p. 18).

Pragmatism = qualitative and quantitative research should not be seen as competing and contradictory, but should instead be viewed as complementary strategies appropriate to different types of research questions or issues.

‘Multi-method, trans disciplinary’ research = employs a range of different methods and draws on expertise from a range of alternate disciplines, as appropriate to the research questions.

Chapter B: Social Research And Qualitative Methods

Theoretical research is concerned with the aim of testing, generating or enhancing thinking within a particular discipline. Applied research is concerned with using the knowledge acquired through research to contribute directly to the understanding or resolution of a contemporary issue.

First theoretical research was more popular. However, researchers began to see the benefits of applied research in the latter twentieth century when they searched for a more fully understanding of the nature of problems that social policies had to address.

Functions of qualitative research:

• Contextual: describing the nature of what exists.
• Explanatory: examining the reasons for what exists.
• Evaluative: appraising the effectiveness of what exists.
• Generative: aiding the development of theories, strategies or actions.

CONTEXTUAL research aids to:

1. Map the range of elements, dimensions, classes or positions within a social phenomenon (e.g. what perceptions of politics do young people hold?).
2. Display the nature or features of a phenomenon (e.g. how does racism manifest itself?).
3. Describe the meaning that people attach to an experience, event, circumstance or other phenomenon (e.g. what does it mean to be a grandparent?).
4. Identify and define typologies (e.g. what are the different models for organizing ‘self-help’ groups?).

EXPLANATORY research aids to identify:

1. The factors that underlie a particular attitude, belief or perception (e.g. what shapes people’s beliefs about poetry?)
2. The motivations that lead to decisions, actions or non-actions (e.g. why do people gamble?).
3. The origins of events, experiences or occurrences (e.g. why does homelessness occur?).
4. The contexts in which phenomena occur (e.g. in what environments does traffic blight occur?).

EVALUATIVE research aids to:

1. Identify the factors that contribute to the (un)successful delivery of a program, service or intervention.
2. Identify the effects of taking part in a program and how they occur (e.g. how do environmental conservation schemes change behaviours?).
3. Examine the nature of the requirements of different groups within a target population (e.g. how do different groups of older people respond to home security initiatives?).
4. Explore a range of organizational aspects surrounding the delivery of a program, service or intervention.
5. Explore the contexts in which interventions are received and their impact on effectiveness (e.g. how does the nature of relationship breakdown affect receipt of family mediation services?).

Patton’s list of evaluative functions for qualitative research:

• Goal free, looking at actual rather than intended effects.
• Responsive, to diverse stakeholder perspectives.
• Connoisseurship, use the evaluators’ perceptions to draw conclusions.
• Utilization-focused, intended use and users of the evaluation.

Formative evaluation = designed to provide information that will help to change or improve a program or policy, either as it is being introduced or where there are existing problems with its implementation.

Summative evaluation = the impact of an intervention or policy in terms of effectiveness and the different outcomes that have resulted.

GENERATIVE research aids to:

1. Develop new conceptions of social phenomena (e.g. the nature of ‘social exclusion’).
2. Develop hypotheses about the nature of the social world and how it operates.
3. Generate new solutions to persistent social problems.
4. Identify strategies to overcome newly defined phenomena or problems (e.g. ways of restoring declining rural economies).
5. Determine actions that are needed to make programs, policies or services more effective (e.g. changes that are needed to help reduce hospital waiting lists).

The factors that determine whether qualitative methods should be the sole method used are centrally related to the objectives of the research. Certain conditions or factors are:

• Ill-defined/not well understood: qualitative research is useful when the subject needs to be more clearly understood before it can be measured.
• Deeply rooted: the phenomena that need to be studied are deeply set within the participants’ personal knowledge. These might be related to the origins of long-standing values or beliefs.
• Complex: the subject is complex and there is a need to understand the phenomena which are conceptually difficult to relate. The complexity may lie in the nature of the subject itself, or in the level of unpacking that is needed to formulate a position, view or belief.
• Specialist: the collection of information from individuals or groups that have a highly specialized role in society.
• Delicate or intangible: certain subjects in social research are difficult to capture because they are so fragile in their manifestation.
• Sensitive: it is hard to predict the subject matters that might prove distressing or emotive to individual participants.

There exist two approaches for collecting qualitative data 1) naturally occurring data, and 2) generated data through the interventions of the research.

Naturally occurring data is a research method developed to allow investigation of phenomena in their natural settings. They provide data which is an enactment of social behaviour in its own social setting, rather than a recounting of it generated specifically for the research study.

Approaches:

1. Participant observation: researcher joins the study population to record actions, interactions or events that occur. The researcher experiences the phenomena for themselves.
2. Observation: record and analyse behaviour and interactions as they occur, although not as a member of the study population.
3. Documentary analysis: study of existing documents, either to understand their substantive content or to illuminate deeper meanings which may be revealed by their style and coverage. This is particularly useful when the history of events is relevant.
4. Discourse analysis: the construction of texts and verbal accounts to explore systems of social meaning. It examines ways in which versions of the world, of society, events and inner psychological worlds are produced in discourse.
5. Conversation analysis: a detailed examination of talk in interaction to determine how conversation is constructed and enacted. The aim is to describe people’s methods for producing orderly social interaction.

Generated data involves reconstruction and requires re-processing and re-telling of attitudes, beliefs, behaviour and other phenomena.

Ways in which data can be generated:

1. Biographical methods: use life stories, narratives and recounted biographies to understand the phenomena under study. It allows participants a high degree of freedom to shape and order the reconstructions in their own way.
2. Individual interviews: a detailed investigation of people’s personal perspectives, for in-depth understanding of the personal context within the research phenomena are located, and for very detailed subject coverage.
3. Paired (or traid) interviews: in-depth interviews but carried out with two people at the same time. This allows participants to reflect on and draw comparisons with what they hear form others.
4. Focus groups or group discussions: several respondents brought together to discuss the research topic as a group. It provides an opportunity for reflection and refinement, and for direct and explicit discussion of differences as it emerges in the group.

Mixing qualitative approaches can be useful, for example when using interviews in combination with observation methods. When mixing qualitative and quantitative methods it must be emphasized that the purpose of bringing the different approaches together is to yield different types of intelligence about the study subject rather than simply to fuse the outputs. In social policy research combining the methods can be very useful since many of the questions that need to be addressed require measurement of some kind but also greater understanding of the nature of an issue.

Three ways of using qualitative and quantitative research together are shown blow.

Preceding statistical research

Using a preliminary qualitative research before a quantitative research can be of value when the subject matter is complex and where some identification of the underlying constructs is needed before relevant questions can be structured. Besides that, it can be used to generate hypotheses for statistical testing. Due to the in-depth investigation, qualitative research can discover possible connections between phenomena. Lastly, preliminary qualitative research can define the dimensions of sample segmentation.

Alongside statistical research

Qualitative and quantitative methods can be used to study the same or different phenomena. There are also occasions where qualitative and quantitative research are brought together in the study of the same phenomenon but then divide in terms of what is explored. Furthermore, qualitative and quantitative research combined can be used to study different phenomena in the same field of research. The need to use both methods is particularly evident in evaluative studies.

As a follow-up to statistical research

Doing qualitative research after quantitative research is useful when the quantitative findings need further explanation, detail or in-depth information. When using both researches, it is important to recognize that each offers a different way of knowing about the world. Therefore, it should not be expected that the evidence will replicate each other.

Triangulation involves the use of different methods and sources to check the integrity of inferences drawn from the data. In this case there is a debate about the extent to which triangulation offers qualitative researchers a means of verifying their findings.

Chapter C : Problems Concerning Design

A good qualitative research study design is one which has a clearly defined purpose, in which there is coherence between the research questions and the methods or approaches proposed, and which generates data which is valid and reliable. Design in qualitative research is a continuing process which calls for constant review of decisions and approaches.

Five key aspects of research design are:

• The development of research questions
• Building design around research settings and populations
• Selecting the time frame for data collection
• Choice of data collection
• Negotiation of research relationships

The development of research questions

Requirements of research questions, they need to be:

• Clear, intelligible and unambiguous
• Focused, but not too narrow
• Capable of being researched through data collection
• Relevant and useful
• Informed by and connected to existing research or theory, but with the potential to make an original contribution or to fill a gap.
• Feasible concerning the resources
• Of interest to the researcher

Building design around research settings and populations

Selecting research settings and populations involves identifying those which are able to provide the most relevant, comprehensive and rich information. Two main issues related to the way in which design needs to be built around the research settings and populations selected are 1) comparisons in qualitative research, and 2) the role of case studies.

• Building comparison: comparison should inform the selection of research locales and populations, aid theory building, and enhance the solidity of research findings. An often discussed issue is control, a feature in which groups are constructed to differ in respect of one key variable so that the effect of that variable can be understood. Comparison in qualitative research can contribute by:
  • Identifying the absence or presence of particular phenomena in the accounts of different groups.
  • Exploring how the manifestations of phenomena vary between groups.
  • Exploring how the reasons for phenomena vary between groups.
  • Exploring the interaction between phenomena in different settings.
  • Exploring more broadly differences in the contexts in which phenomena arise or the research issue is experienced.

Control groups are more strongly associated with quantitative designs and with measurement of the effect of the intervention.

• Building case studies and structural linkage: the term case study is sometimes used as a synonym for qualitative research. Features associated with case studies are:
  • The fact that only one case is selected, although it is also accepted that several may be.
  • The fact that the study is detailed and intensive.
  • The fact that the phenomenon is studied in context.
  • The use of multiple data collection methods.

Case studies raise a number of questions at the design stage. Decisions should be made about which cases to select, the composition of the case, and the amount of consistency between cases.

Selecting the time frame for data collection

Determining the appropriate timing of research in relation to the event involves considering what perspectives on that event are implied by the study objectives.  In investigations of new initiatives, services or policies, the appropriate timing will depend on whether the focus of the study is.

• Single research episodes: many studies involve only one episode of fieldwork, even in a changing quality of what is being studied. This is often sufficient since qualitative research involves probing and clarification. However, if the process of change is an important aspect of what is being researched, involves complex processes or a substantial timespan, a single episode of data collection might not be enough.
• Longitudinal qualitative research studies: this involves more than one episode of data collection. There are two broad forms:
  1. Panel studies: people are interviewed more than once. Used to explore micro-level change, where the focus of change is the individual.
  2. Repeat cross-sectional studies: subsequent samples of new participants are interviewed.  Used to explore macro-level change, where the focus of change is on the wider context within which they are situated.

Questions concerning the design stage in longitudinal studies:

• The number of research episodes and their timing
• Initial sample selection
• Field work methods
• Selection for follow-up interviews: the follow-up stage can be designed to allow intensive study of particular groups or issues, returning to a purposively selected sub-sample of those interviewed at the first stage. Alternatively, the follow-up sample can be shaped around events or experiences that have occurred since the first stage fieldwork.
• Analysis: the analysis of the first stage fieldwork needs to be organized in a way which will make it possible to integrate later stages of data, to make comparisons and identify stages.

Choice of data collection

Three considerations should be made:

1. Choosing between naturally occurring and generated data

The research will need to consider the following:

• The importance of context: generated data allows participants to describe the personal context in which the research issue is located and how they relate to it. Naturally occurring data the context is more important since then observing the research phenomenon in its natural context is critical for understanding.
• Whether a recounting of the research phenomenon is likely to be sufficiently detailed, accurate or complete: if people are unable to give a truthful account since they are unwilling to talk frankly about something, or if they are bound up with social rules and expectations, naturally occurring data may be useful.
• Whose interpretation is paramount: naturally occurring data relies on the researcher’s interpretation of what is observed. Generated data collection gives participants a direct and explicit opportunity to convey their own meanings and interpretations through the explanations they provide.

2. Choosing between in-depth interviews and focus groups

The selection depends on three key factors:

• The type of data sought: a key feature of in-depth interviews is their depth of focus on the individual. Focus groups are used where the interaction between participants (group process) are important.
• The subject area: very complex systems, processes or experiences are best addressed in in-depth interviews as well as understanding motivations and decisions. Intangible or conceptual topics are well suited to focus groups as well as studies focusing on attitudes and views.
• The nature of the study group: in-depth interviews allow the participant’s to choose the location. Furthermore, in-depth interviews are more appropriate if the participants have nothing in common. Although focus groups benefit from diversity, it is helpful to have some commonality between the participants related to the research topic. Small groups might provide a good balance between the group and the individual context.

Combining in-depth interviews with focus group can be very useful. Focus groups might be used as an initial stage to begin to explore relevant issues which will then be taken forward though in-depth interviews. Furthermore, focus groups could be used after in-depth interviews to discuss the issues at a more strategic level.

3. Secondary data analysis

This can be a valuable resource, providing an opportunity to bring a new perspective to existing data, to use elements of the data that have not been fully analysed, or to form a base for comparison with newly collected data. However, the adequacy of the original data for the new research needs to be considered carefully.

Points of attention:

1. It may be that certain subject areas were not central to the original objectives, and that this is reflected in the data available.
2. The sample may not be ‘comprehensive’ for the purposes of the secondary analysis and may have important constituencies missing.
3. The original data needs to be of high quality in terms of the conduct of the original data collection.

Negotiation of research relationships

The way in which access (approval of research) is negotiated on the ground can be critical to the success of a study. Engaging in research settings can be aided in a number of ways:

1. Being sensitive to the hierarchy.
2. Providing clear information about the objectives of the study and why that setting has been chosen.
3. Being open and consistent about what is required.
4. Being clear about how the findings will be used.
5. Anticipating and being responsive to concerns and sensitivities raised.
6. Having a single point of contact within the organization.
7. Being flexible about shaping the study approach in response to the precise setting and accepting advice.
8. Considering how findings can be shared and at what stage.

Making studies accessible to the researched groups involved requires consideration of the appropriate language to use in approaching them, anticipation of the possible barriers to participation, and provision to help to overcome them. The particular questioning techniques that will make the study accessible to the participant should also be considered.

It is also useful, at the design stage, to think about some reward in return for the assistance, time and thought given by the participants. However, the requirement of being objective and neutral must stay important.

Researchers try to present themselves objectively and neutrally. However, this is not the only thing that is relevant: a broader cross-perception between participant and researcher also takes place. A solution might be a researcher and a participant that are matched on key socio-demographic criteria, such as gender, social class or ethnicity. A limitation to this approach is that insufficient explanation or clarification is sought by the researcher because of assumptions created by their shared experience. One must weigh up the relative risk of cultural collusion versus unhelpful power dynamics.

In doing research one should consider the amount of information provided to the participants. This must be balanced since giving too much information may deter the potential participants and giving too less leads to inadequately prepared participants. Furthermore, the proposed conditions for anonymity and confidentiality should be given particular though, and made very clear to participants. Confidentiality means avoiding the attribution of comments to identified participants.

It is important to give consideration to ways in which taking part may be harmful to sample members, and to take aversive action. This may be relevant in studies on sensitive topics. A particular difficult ethical dilemma arises when information is disclosed, which indicates that the participant is at risk of harm. An appropriate answer would be for the researcher to encourage the participant after the interview to report it themselves, or to seek help in some other way.

Researchers who conduct fieldwork also place themselves at risk. In public places, this will involve decisions about appropriate modes of transport. In private, this involves the interview content and dynamic which might spark anger or raise risk.

Most research benefits from teamwork as working in teams provides more opportunities for reflection and review. The time involved in collaboration needs to be considered realistically and built into the research budget. A final consideration is the importance of managing the timetable and budget once work begins.

Chapter D : Sample Issues

When sampling strategies for social research are described, a key distinction is made between probability and non-probability samples. In a probability sample elements in the population are chosen at random and have a known probability of selection. Often the probability of units being selected is equal. However, unequal probability samples exist as well and then the data has to be re-weighted during the analysis. In a non-probability sample units are deliberately selected to reflect particular features of the sampled population. The sample is not intended to be statistically representative but instead the characteristics of the population are used as the basis of selection.

Main sampling approaches:

1. Criterion based or purposive sampling: the sample units are chosen because they have particular features or characteristics which will enable detailed exploration and understanding of the central themes the researcher wishes to study (e.g. socio-demographic characteristics). This approach has two aims 1) to ensure that all the key constituencies of relevance to the subject matter are covered, and 2) to ensure that some diversity is included so that the impact of the characteristic concerned can be explored. Different types of purposive sampling are:
   • Homogeneous samples: to give a detailed picture of a particular phenomenon.
   • Heterogeneous samples: to include phenomena which vary widely from each other in order to identify the central themes this cut across the variety of cases.
   • Extreme case/deviant sampling: cases are chosen because they are unusual or special and therefore potentially enlightening.
   • Intensity sampling: extreme cases of specific phenomena of interest.
   • Typical case sampling: cases which characterize positions that are normal or average are selected to provide detailed profiling.
   • Stratified purposive sampling: to select groups that display variation on particular phenomena but each of which is fairly homogeneous, so that subgroups can be compared.
   • Critical case sampling: cases are chosen on the basis that they demonstrate a phenomenon, dramatic position or are central in the delivery of a process.
2. Theoretical sampling: a type of purposive sampling in which the researcher samples incidents, people or units on the basis of their potential contribution to the development and testing of theoretical constructs. It is an iterative process, which means that the researcher picks an initial sample, analyses the data, and then selects a further sample in order to refine the emerging categories and theories. This goes on until the ‘data saturation’ stage has come (no new data is obtained through expanding the sample).
3. Opportunistic sampling and convenience sampling: opportunistic sampling involves the researcher taking advantage of unforeseen opportunities as they arise during the course of fieldwork. Convenience sampling lacks any clear sampling strategy, the researcher chooses the sample according to ease of access.

Although the two main approaches used in qualitative research, purposive and theoretical sampling, differ, they also have much in common. The characteristics are:

1. The use of prescribed section criteria: samples need to be selected to ensure the inclusion of relevant processes that can inform that understanding. Units are chosen because they typify a circumstance or hold a characteristic that is expected or known to have salience to the subject matter under study. Furthermore, the sample must be as diverse as possible within the boundaries of the defined population. This optimizes the changes of identifying the full range of factors associated with a phenomenon and allows some investigation of interdependency between variables.
2. Sample size: qualitative samples are usually small in size because if data is properly analysed very little new evidence is obtained from each additional fieldwork unit. Besides, statements about incidence or prevalence are not the concern of qualitative research and the type of information that qualitative studies yield is rich in detail. Lastly, qualitative research is highly intensive in term of the research resources it requires. Issues that need to be considered in determining the sample size:
   • The heterogeneity of the population: if the population is very diverse, the sample size must increase.
   • The number of selection criteria: the more criteria, the larger the sample.
   • The extent to which nesting of criteria is needed: if criteria need to be interlocked (nested), the sample size will increase.
   • Groups of special interest that require intensive study: if groups require intensive study, the sample size will increase.
   • Multiple samples within one study: for reasons of comparison or control, this increases the sample size.
   • Type of data collection methods: if the data collection methods involve single interviews, paired interviews or group discussions (in this order), the sample size will increase.
   • The budget and resources available
3. Additional and supplementary samples: this may occur when it is found that important constituencies are not sufficiently well represented to derive sound qualitative evidence.

Theoretical sampling is particularly appropriate for exploratory studies in unfamiliar areas, since it may be difficult to identify in advance the groups and characteristics that need to be included in the sample.

The first stage in sample design involves identifying exactly what it is that is to be sampled. In social research this usually involves people. Therefore, it is necessary to define the parent population (the population from which the sample is to be drawn).

Three key questions in defining the population for study:

• Which group or subpopulation is of central interest to the subject matter of the study? The population that will provide the richest and most relevant information.
• Are there subsets of the central population that should be excluded?
• Are there additional groups or subpopulations that should be included because their views, experiences would bring contrasting or complementary insights to the research?

Defining a population involves two stages, specifying the characteristics of the collective units required (1) and then specifying those of the individuals required within them (2).

Key criteria by which any potential sample frame will need to be judged:

1. Does the sample frame provide the details required to inform selection?
2. Does the sample frame provide a comprehensive and inclusive basis form which the research sample can be selected?
3. Will the sample frame provide a sufficient number of potential participants to allow for high quality selection, particularly given that not all will be eligible or willing to participate in the study? The sample frame will need to generate a group of eligible potential participants which is around three to four times the size of the required study sample to allow scope for selection.

Two key types of sample frames:

• Existing sources: most convenient type of sample frame. Key types are:
  • Administrative records: principal shortcoming is that, because they are not generally designed for research purposes, they are unlikely to contain all the information that qualitative research sampling requires.
  • Published lists: it is important to investigate the criteria for inclusion, and to consider whether the list is sufficiently comprehensive.
  • Survey samples: very useful if it is available since it will meet all the requirements for comprehensiveness, diversity and lack of bias. Offers the opportunity to know how certain variables are distributed within the study population before sample composition decisions are finalized. However, one should avoid over-complex categorizations or making assumptions about likely attitudes based on survey responses.
• Generated sampling frames: time-consuming
  • A household screen: conducting a short interview in households in order to identify whether the household contains an individual who belongs to the study group and if so to collect further information relevant to sample selection. Often the researchers are unable to do the interviews themselves and therefore hire others which need to be fully briefed and possessed with all documentation (p. 94).
  • Through an organization: useful when populations are too scattered or too small to be identified easily through a household screen. A difficulty is how to include people who are not in contact with organizations. Another one is that when the selection criteria are includes complex or involve information that individuals may not want to disclose to the organization.
  • Snowballing or chain sampling: this involves asking people who have already been interviewed to identify other people they know who fit the selection criteria. A shortcoming is that the diversity of the sample frame might be affected.
  • Flow populations: samples are generated by approaching people in a particular location or setting. This might be effective in identifying a specific population.

Choosing a sample frame:

• General population samples à household screen (less useful for very rare groups).
• Groups that have an administrative significance à administrative records.
• Groups which are rare or hard to find à survey sample (1), generating a sample through organizations (2), snowballing (3).
• Specific minority ethnic groups. E.g. household screen in high density of ethnic community, snowballing in low density areas.
• Samples of organizations or professional à published lists.

If the sample is being generated through someone else than you, it will be necessary to consider carefully what arrangements will be required for asking permission from potential sample members for their details to be passed on to the research team. A solution might be an ‘opt in’ approach which gives individuals an opportunity to indicate that they do not want their details to be passed on, but treats inaction as consent.

Steps in designing a purposive sample:

1. Identifying the population for study (p. 13 of summary)
2. The choice of purposive selection criteria
3. Prioritizing the selection criteria
4. Deciding on the locations for the study
5. Designing a sample matrix
6. Setting quotas for selection
7. Area allocations
8. Sample size
9. Purposive sampling for group discussions

The choice of purposive selection criteria

Within the parent population, which constituencies need to be represented and with what level of diversity?

Factors that need to be considered:

• A review of relevant literature will identify characteristics that are shown to have an impact on the subject being investigated.
• There will be variables which need to be covered simply to achieve a balanced sample.
• There may be hypotheses that the research is exploring that will require coverage of particular subgroups.
• There may be subgroups about which little is known and whose circumstances or views need to be explored.

Prioritizing the selection criteria

The selection criteria should be prioritized in some way because otherwise the sample scale would be beyond what is manageable. The first step is prioritizing the criteria based on which are the most important in terms of achieving a sample that is inclusive of the demographic structure of the population being studied, that contains the key constituencies, and that is sufficiently diverse for comparative analysis. Secondly, the extent to which the variables should interlock (nested) should be considered. Furthermore, it is useful to consider whether any of the criteria identified are highly correlated with each other.

Deciding on the locations for the study

Locations are usually selected because of their salience to the subject under research.

Designing a sample matrix

This is the most useful way to convert decisions about the remaining sampling criteria into a sample design. An example is shown at page 103. The matrix will include a number of items relating to the primary sampling criteria. These will yield a number of cells, each of which will be assigned a number of sample units to be selected (a quota). The secondary criteria will then be considered in relation to each of the dimensions or cells to identify ways in which these variables can be controlled. Before placing the selection criteria items in the matrix, each needs to be divided into categories that are meaningful to the subject.

Setting quotas for selection

Quotas specify the precise number of people that will be needed with each of the characteristics set out in the sample matrix. They are used to control the final selection of participants, so that the study sample matches the sample design set out in the sample matrix. Quotas can be specified as exact numbers but it is more usual to use ranges. Two general features of quota specification are 1) the order in which the variables are listed takes account of the way that data might be collected ruing screening, and 2)the ranges given in each of the categories for patterns of attendance are slightly different so that they add up to achieve the right numbers overall.

Area allocations

If there are features that vary between areas, quotas may need to be specified separately for each area.

Sample size

If having drawn up the sample matrix the sample size falls outside the manageable range, some questions arise. Have too many variables been included or too many given top priority? Is the level of nesting proposed necessary?

Purposive sampling for group discussions

The sample size will be larger, but all the steps described need to be followed as well. There is one further decision to make, which involves specifying the composition of each of the groups. The group composition needs to be translated into a specification of quotas for each of the discussion groups.

Steps for implementing the sample design:

1. Selection to meet quota requirements: unless the sample frame provides all the information relevant to selection and assignment to quotas, a screening exercise will be needed with a short screening questionnaire. People that fall out need to be replaced by others with as similar as possible characteristics.
2. Documenting outcomes: when recording the outcomes of approaches to potential participants one will see whether there is attrition among specific groups or constituencies in the sample frame. Furthermore, it must be recorded in categories why people fall out:
   • Ineligible or out of scope: fall outside the detailed definition of the study sample.
   • Non-contacts
   • Not meeting quota requirements
   • Refusals to participate
   • Agreement to participate

Chapter E : Fieldwork

All qualitative data collection will have some structure, however, the extent to which the structure of data collection can be usefully depends on the specific purposes of the study. Although qualitative research often follows the direction taken by participants, there will be a strong sense in advance of the issues that needs to be explored.

Data collection is more structured in an evaluative or investigative study and less structured in a very exploratory study.  In general, too, focus group data collection is less structured than in-depth interviewing since data emerges though interaction within the group.

Two main types of qualitative interviews are discovered:

• Unstructured, non-standardized/in-depth interviews (exploratory), more probing
• Semi-structured/semi-standardized interviews (evaluative), less probing

An important aspect of an interview is the order in which issues and topics might be approached. This involves mentally picturing the interview or group discussion and working out the most natural way to structure it. An order is useful for several reasons:

• Interviews are processes with their own dynamic, which means that different issues are best addressed at different stages of the process.
• Understanding something of the personal context early in the interview will be important to make sense of what they later say, and to probe effectively.
• It will be easier for the research team to become familiar with the topic guide.
• In general, a chronologically order will aid in-depth exploration.

Useful general principles in ordering an interview, Box 5.1 (p.116):

1. The opening topics should be easy and unthreatening.
2. The opening topics is an opportunity to collect information that will provide important information for alter stages of the interview.
3. Move from general to more specific topics to create an unthreatening atmosphere.
4. Questions about experiences, circumstances and behaviours should precede motivational or attitudinal information since people talk easier about the first.
5. Introduce a discussion of definitions at an early stage in order to hear participant’s initial reflections rather than asking these questions later when their definitions has been influenced by the discussion.
6. The interview should have a clear and expected end and lead to a positive note (e.g. questions about the future).
7. Towards the end, it can be helpful to include questions which summarize somebody’s attitudes or experiences.

A topic guide provides documentation of subjects to investigate that serves as an interview agenda. A well-designed topic guide will provide flexible direction to field-work process and essential documentation of a central aspect of the research. This ensures that the relevant issues are covered systematically.

Designing topic guides consists of several processes:

1. Establishing subject coverage
2. The structure and length of the guide
3. Language and terminology
4. Specification of follow-up questions and probes
5. Making the guide easy to use

Establishing subject coverage

Which subjects will be covered in the data collection is often clear to the researcher from the stated objectives of the research and the exiting literature in the field. It is important to maintain clarity about the central objectives of the study and not to allow specific questions to shift the focus too far. Therefore, part of this process involves ruling certain topics as outside the scope of the research.

The structure and length of the guide

Concerning follow-up issues, Rubin and Rubin distinguishes two types of interviews 1) tree and branch model (the branches are issues being pre-specified for follow up), and 2) rivers and channel model (researcher follows channels wherever they lead).

A very detailed topic guide may include:

• Suggested wording for opening and closing
• Specific subjects to be covered
• Suggestions for prompts and direction for probing
• Suggested wording for questions addressing sensitive topics.

The amount of detail of the topic guides reflects the extent of pre-structuring that is desirable. Difference in length is also determined by individual styles of creating topic guides. Shorter guides generally encourage more in-depth data collection. However, more detailed topic guides are preferred for advisers and research commissioners. For focus groups less key topics are desirable compared to an interview in order to allow time for all group members to be drawn into each topic.

Language and terminology

In general, topic guides work best when items are not worded as actual questions, but instead use single words to indicate the issues and leave the formulation to the researchers. Since the researcher needs to be responsive to the language used by participants, the items should be phrased in language which is as neutral as possible.

Specification of follow-up questions and probes

Topic guides usually include some indication of issues for follow-up questions and probing. These follow-up questions are vital to ensure full exploration of the issues. One way in which topic guides can vary considerably is the degree to which follow-up questions are included in the guide or left to the researcher.

Making the guide easy to use

1. Objectives: begin with a brief statement of the objectives as a reminder of the underlying purpose of the study.
2. Introduction: to remind the interviewers what to say at the start. This may include details of the research team or reason of conduction.
3. Summary of topics: provides a quick and easy-to-read reminder.
4. Layout: a lot of space makes it easier to read and to annotate if needed.
5. Instructions: the following can be useful to note on the guide:
   • Suggested wording for questions that are difficult to introduce.
   • How to carry out a task or when to show a visual aid.
   • Description of the rationale behind a particular question.
   • Indication of the relative priority of different topics (e.g. ‘key topic’ and ‘briefly’).
   • Instructions for where certain sections are relevant to subgroups only.
   • Instructions to explore particular dimensions
6. Ending: reminder for talking about confidentiality and how the data will be used.
7. More than one topic guide: creation of more than one topic guide is easier if a study involves subgroups whose circumstances differ and therefore need to be asked a separate set of questions. One must ensure that both guides cover the common areas in the same way.

In the early stage of the topic guide structure one must consider incorporating other, additional research instruments. Other research instruments are:

1. Collecting structured data: this can be important when relatively complex and detailed background information is needed in order to have a clear understanding of participant’s situations. Furthermore, using calendars or diaries can enhance data collection since it acts as a memory and supports greater precision in the dating of events.
2. Using case illustrations and examples: rooting discussion in specific examples can add depth and richness to data collection. It ensures that information obtained is not general, but a description of actual behaviour.
3. Enabling and projective techniques: techniques that are used to aid expression and refinement of views. They can help to focus the discussion, enabling people to consolidate their views, or promote further thought.
   • Vignettes: these are short descriptions of particular circumstances, persons or events, which might be described verbally by a researcher or a written version shown. They introduce an element of consistency which can be useful, allowing comparison between the reactions of different participants to the same hypothetical example.
   • Card-sorting: participants are shown a number of written or visual examples of an issue, and asked to order them. This facilitates discussion of the reasons for choices and priorities.
   • Giving information or showing written material: when reflections on different proposals are required, discussion must be stimulated further, or if the topic is one about which knowledge is likely to be limited among participants, there is a need to introduce information.
   • Mapping emergent issues: this is more appropriate for focus groups as it displays what is generated already and encourage them to take ownership of it and to move it forward.
   • Projective techniques: a range of strategies designed to stimulate discussion and communication, and to access thinking or beliefs that are less conscious or that may be difficult to speak about. Five types of projective techniques are distinguished.
     • Association: word or brand association
     • Completion: compete sentences or stories
     • Construction
     • Expressive: drawing and role enactment
     • Choice-ordering: selecting or ranking items

Projective techniques provide a means of cutting through self-consciousness and can draw out views that are otherwise less acceptable, less conscious or are based on strong underlying emotions. Disadvantages are that it is time-consuming, disruptive to the flow of discussion and open to misinterpretation.

4. Field notes: field notes provide an opportunity to record what researchers see and hear outside the immediate context of the interview.

A full briefing for the whole research team is one of the most critical elements for successful data collection. It is a very good opportunity to discuss how the topic guide will work in practice, to identify any potentially difficult areas, and to think about different ways in which questions might be phrased or issues approached. Written information will also be an important aspect of the briefing of team members. It may also be necessary to obtain comments on the topic guide from a research commissioner or advisory group.

After briefing one should study the guide and become familiar with its structure, detailed contents, thinking about how different issues might be addressed, the type of responses they might yield and how they will need to be followed up.

Initial interviews and focus groups will be an important test of the scope of the topic guide. It is important to review whether the topic guide allows participants to give a full and coherent account of the central issue and incorporate issues. A useful aid in the refinement of fieldwork strategies and topic guides is for members of the team to conduct initial interviews. This is helpful for discussion of how well the guide is working. Furthermore, it is very valuable to review the topic guide after several interviews since it provides researchers the chance to refine the guide.

Chapter F : Interviews

In-depth interviews can be regarded as being ‘conversations with a purpose’, appearing naturalistic yet bearing little resemblance with ordinary conversations. Two metaphors can describe the phenomenon: miner (knowledge is given and interviewer must ‘mine’ this information to surface). This resembles a mine worker who must extract the raw materials. Secondly, a traveller metaphor (knowledge is created and interviewer must ‘accompany’ the interviewee on its route) is used. The stories of the interviewee are interpreted by the researcher can may lead to new insights.

There are six key features of in-depth interviewing. Firstly, it is intended to combine structure with flexibility. This way, there is space for structured questions but also for spontaneous probing. Secondly, the interview is interactive in nature.  Based on the interventions of the researcher, the interview can be guided. Thirdly, the researcher uses a range of techniques to acquire depth.  This generally refers to follow-up and probing questions. Fourthly, the interview is generative so that new knowledge is created. Both the researcher and the participant can discover original avenues of thoughts. Fifthly, interview data must be captured in its natural form. Therefore, the interview is ordinarily taped since note-taking would alter the form of data.  Lastly, qualitative interviews are generally conducted face-to-face.  Because of its intensity, a physical encounter is essential for it is flexible, interactive and generative.

Researches must be skilled to execute these kinds of interviews. They must be able to listen very well in order to be able to probe; they must have hat the researcher can come back to a previous point at any time. During the interview, however, the researcher must avoid analysing the information immediately and must focus solely on interviewing.

The staging of the interview must meet the purposes of the research. The researcher must be aware of these stages, must understand the terms of contract between the interviewee and himself, and can decide on the role of the participant during the interview. Generally, the stages are as follows:

1. Arrival: establish immediately a relationship and comfortable atmosphere
2. Introducing the research: what is the research topic about?
3. Beginning the interview: start neutrally to distill contextual information about the participant
4. During the interview: guide the participant through key themes
5. Ending the interview: introduce the end and make sure everything has been said
6. After the interview: thank the participant and reassure confidentiality issues

The contract between both parties is vital. Interviewees must have given their consent to be interviewed within the terms both agreed upon. Those terms generally consist of the length, the venue, the topic, and confidentiality conditions. Participants have the right to change their mind at any time. Terms are continuously re-negotiable even during the interview.

The researcher and the participant have different roles. Both roles should be clear prior to the interview. The interviewer can be seen as a facilitator or enabler, who participates actively by choosing questions and structuring the interview. The interviewees role is to give fulsome answers, reflect and think, answer probing questions and raise other issues they are unsure about. 

In order to achieve breadth and depth, the research can asked content mapping questions and content mining questions. The content mapping questions entail the following types:

• Ground mapping questions: open up the subject
  • ‘Have you ever applied for a benefit?’
• Dimension mapping questions: signpost, structure, and direct the interview. It usually refers to the process or experience the participant underwent.
  • ‘What happened next?’
• Perspective-widening questions: let interviewees give more than their first thoughts and widens their initial perspectives.
  • ‘Are there any other factors that would influence your decision?’

On the other hand, the content mining questions deal with the following types:

• Amplificatory probes: enable the interviewee to provide a full description of a subject
  • ‘Can you tell me a bit more about…?’
• Exploratory probes: exploring impacts, effects and consequences
  • ‘How did you respond when…?’
• Explanatory probes: understand the reasoning of the participant
  • ‘What makes you say that?’
• Clarification probes: achieving high degree of clarity and precision
  • Clarify terms and language
  • Clarify details and sequences
  • Clarify through expressing a position
  • Challenge inconsistency

With respect to the question formulation, the researcher should seek to use both broad and narrow questions, avoid leading questions, and ask clear questions. With respect to the content mapping and content mining questions, the amount of questions varies. There are both broad (how did you…) and narrow (yes/no) questions that constitute an in-depth interview. The researcher should seek to avoid asking questions that suggest a possible answer, like ‘Were you furious when he said that?’ The last criterion refers to not prefacing a question, not asking double questions and avoiding abstract or theorized questions.

Obviously, listening is very important and can serve to increase depth, together with remembering the answers. To further enhance that depth, the researchers should facilitate a relationship with the interviewee by expressing interest and attention; emphasizing there cannot be right or wrong answers; being sensitive to non-verbal characteristics; allowing the participant time to reply; and handling extraneous information. A researcher should furthermore turn assumptions and interventions into questions. That is, he must not assume; refrain from commenting on, summarizing, and finishing off an answer; and avoid extraneous remarks. Lastly, the interviewer must be neutral.

From page 163 onwards, a number of scenarios is presented in which is explained how the interviewer can respond to different situations. Among them are sensitivity, emotion, and anxiety.

Practical considerations prior to and during the interview are appointment scheduling, venues, recording, and audience. Generally, an interview should last for at least an hour. However, this duration can change so that the interviewer and interviewee must be flexible regarding appointments. Moreover, the participant generally chooses the venue of the interview and the venue should be conducive to concentration. It is highly desirable to audiotape the interview so the interviewer does not have to take many notes and can be entirely devoted to the answers given. Because of the clarity, the interview should preferably be carried out by one researcher primarily.

Chapter G : Focus Groups

A focus group is not just a collection of individual interviews with comments directed solely through the researcher. This is better described as a ‘group interview’, and lacks both the depth of individual interviews and the richness that comes with using the group process. Instead, Focus groups are synergetic in the sense that the group works together: the group interaction is explicitly used to generate data and insights.

A further feature of focus groups is the spontaneity that arises from their stronger social context. In responding to each other, participants reveal more of their own frame of reference on the subject of study. All this emerges from discussion within the group, the perspective is less influenced by interaction with the researcher than it might be in a one-to-one interview. In a sense, the group participants take over some of the interviewing role, and the researcher is at times more in the position of listening in.

Typically, focus groups involve around six to eight people who meet once, for a period of around an hour and a half to two hours. Since the last decade of the twentieth century, there has been an emphasis on using research for consultative purposes. This led to some innovations in the application of research methods, and particularly of group discussion methods. Consultative panels have been conducted in different forms and involve drawing people together in a series of sessions to deliberate and contribute to decision making. Focus group research is an interesting application to decision-making, particularly useful in more unfamiliar, technical or complex areas where information provision is more important.

Although group based research usually involves a physical coming together of participants, this is not always the case. Nominal groups have been used for some time. Here, views are gathered from group members individually and collated and circulated for comment. The Delphi technique is a particular application of this. A panel of experts is asked individually to provide forecasts in a technical field, with their views summarized and circulated for iterative forecasting until consensus is reached.

Group process

Based on examination of studies of small groups, Tuckman and Jenson identified five stages in small group development which demonstrate a sequence that groups tend to pass through. The model was based on examination of studies of small groups which were mainly therapy and training groups.

In the forming phase, individuals may be guarded, tense and anxious and concerned about inclusion and acceptance. They tend to address comments solely to the moderator, not yet engaging with other group members. This is the stage at which background information is usefully collected so that participants are on familiar ground.

Storming is a period of tension or criticism that may be shown up in a number of ways. It may  be typified by dominance or one-upmanship from some individuals, by silent aloofness from others, or by the adoption of particular roles as a defensive position. Strong differences may emerge in this phase of the group which may provide useful material to return to, but these differences may diminish later as people express themselves with more complexity and subtlety.

This is followed by the group settling down to a calmer phase of sharing, similarity and agreement, or norming, in which the norms of the group are established. The group begins to work cooperatively and may be particularly keen to find common ground, to agree with each other and to reinforce what others say. Participants may in this phase begin to put into practice the ‘ground rules’. This is the stage at which social norms will be most influential, revealing what are seen as socially acceptable views or behaviours.

The performing phase which follows finds the group working interactively in open discussion on the research issues. This is likely to be with energy, concentration, enjoyment and a less guarded stance, allowing both agreement and disagreement between participants. This is the most productive phase of the group process, but it takes time to reach it.

Finally, in the adjourning phase, the group works towards ending. Participants may take the opportunity to reinforce something they have said earlier or to give their final thoughts. There may be a circular process, with the group dynamic perhaps reverting back from the performing stage to the storming stage.

Stages of a focus group :

1. Scene setting and ground rules
2. Individual introductions
3. The opening topic
4. Discussion
5. Ending the discussion

The researchers aim is to allow as much relevant discussion as possible to be generated from within the group while at the same time ensuring that the aims of the research are met. The researcher will remain as non-directive as possible but will nevertheless be pacing the debate to ensure that all the key issues are covered as fully as possible.

How to control the balance between individual contributions :

• create space for everyone to contribute
• Address the dominant participants, reduce their influence in the discussion
• Try to actively let the ‘’silent’’ participants join the discussion
• Avoiding simultaneous dialogue

A good focus group is more than the sum of its parts. The researcher harnesses the group process, encouraging the group to work together to generate more in-depth data based on interaction.

To encourage the group to build on what they have generated, we can apply several approaches :

• If a potentially interesting issue has been raised by one group member, the researcher may allow the discussion to continue, seeing whether others will pick up on it
• The researcher may decide to draw attention more directly to the point, asking for more comments on it or asking a specific question about it of the group.
• They may encourage the group to reflect on the links or relationships between what individual participants are saying.
• If divergent views are being expressed, the researcher may ask whether these are in conflict with each other or can be reconciled
• They may encourage respondents to focus on the implications or consequences of what has been raised in individual examples

Group composition

The size and composition of a group will be critical in shaping the group dynamic and determining how, and how well, the group process works. As a general rule, some diversity in the composition of the group aids discussion, but too much can inhibit it. Very heterogeneous groups can feel threatening to participants and can inhibit disclosure. If the group is too disparate, it is difficult to cover key topics in depth. The ideal is therefore usually a point of balance between the two extremes of heterogeneity and homogeneity, with as much diversity as the group can take but no more.

Three further issues need to be considered in weighing up the extent of diversity to build into group composition.

• First, it is usually necessary for respondents in each group to have broadly the same proximity to the research subject.
• Second, the socio-demographic makeup of the group can influence how frank and fulsome discussions will be.
• A third consideration is that it may be a specific requirement of the research to look at differences between subgroups within the sample.

Group size

Focus groups typically involve around six to eight participants, but the optimum group size will depend on a number of issues :

• The amount of group participants are likely to have to say on the research topic
• The sensitivity or complexity of the issue
• The extent to which the researcher requires breadth or depth of data
• The population group involved
• The structure and tasks involved in the session

If the group is larger, above about eight participants, not everyone will be able to have their say to the same extent. In groups that are small than about five or six, the researcher may similarly need to be more active, but in the sense of energizing or challenging the group. If the group is very small, with fewer than four participants, it can lose some of the qualities of being a group, particularly if there is a lot of difference between respondents.

• Practical arrangements are also key to the success of group discussions: the time, the venue, the layout of the room and the quality of recording equipment are all important.

Chapter H : Focus on Investigation

Observation in research content is more systematic and formal. Ethnocentric research is predicated on the regular and repeated observation of people and situation. Observation is defined as ‘the act of noting a phenomenon, often with instruments and recoding it for scientific purposes.’ In everyday usage, observation is restricted to the visual but a good ethnographer must be aware of information coming in from all sources. Ethnocentric observation is conducted in the field of natural settings.

According to Gold (1958) there are four categories of roles adopted by the ethnographer:

• Complete observer role: observers are neither seen nor noticed
• Observer-as-participant role: the researcher is known and recognized but relates to the ‘subjects’ of study solely as a researcher.
• Participant-as-observer: his or her activities as a researcher are still acknowledged by being fully integrated into the life of group under study
• Complete participant: he or she disappears completely into the settings and is fully engaged with the people and their activities

Most ethnographers position themselves somewhere within the second two roles.

Roles can also be discussed in terms of membership:

• Adopting peripheral membership: researcher observes and interacts closely with the people under study and thereby establish identities as insiders but do not participate in those activities constituting the core of group membership.
• Adopting an active membership role: researcher engages in core activities although they try to refrain from committing themselves to the group’s values, goals and attitudes.
• Adopting complete membership: study settings in which researchers are active and engaged members. They are often advocates for the positions adopted by the group.

Observational techniques suitable for research dealing with

• Specific settings
• Events
• Demographic factors

It is necessary to have the following qualities: language skills, explicit awareness, a good memory, cultivated naiveté and writing skills.

The process of observational research:

First step is site selection. As the site is selected, it is necessary to gain entrée into the community. In less inviting settings, added preparations must be made and gatekeepers must be approached and their approval and support gained. Then, researcher my begin observing immediately. The more exotic the locale, the more likely will it be that the researcher suffers from culture shock. An understanding of what is and is not central comes only after repeated observations. The method is best that helps the individual researcher retrieve and analyse whatever has been collected is the bottom line. With the progress, observations will gradually fall into discernible patterns.

James Spradley (1980) has referred to the stages of observation as a ‘funnel’ because the progress gradually narrows and directs researchers’ attention more deeply into the element of the setting. Observations continue until a point of theoretical saturation is achieved.

Reliability is the measure of the degree to which any given observation is consistent with a general pattern and not the result of random change. There are some ways in which observation-based researchers can achieve something approaching criteria of scientific reliability. For instance, observations that are conducted in a systematic fashion or that are repeated regularly over time.

Validity is the measure of the degree to which an observation actually demonstrates what is appears to demonstrate. Qualitative ethnographic researchers are not usually concerned with reliability because there is not expectation that one researcher observing a community one time will exactly duplicate the findings of a different researcher. If there is no basis for trusting the observation the research is meaningless. Observations are susceptible to bias from subjective interpretations. Observational findings are rarely confirmable. The most deployed means of achieving validity are:

• Work with multiple observers or teams (represent various viewpoints)
• Methodology of analytic induction (emergent propositions are tested in a search for negative cases. Goal is to achieve assertions that can be taken as universal.
• Techniques of verisimilitude (style of writing that draws the reader into the world that has been studied to evoke a mood of recognition (use of descriptive language) à more than other types of scientific ‘data’ ethnographic observations only become ‘valid’ when they have been rendered into some sort of coherent, consistent narrative

There are five categories that help us to judge the quality of research:

• Objectivity/confirmability (‘external reliability): degree to which conclusions flow from the information that has been collected
• Reliability/dependability/auditability: degree to which the process of research has been consistent and reasonably stable over time
• Internal validity/credibility/authenticity (‘true value’): degree to which the conclusions of a study make sense
• External validity/transferability/fittingness: degree to which the conclusions of a study have relevance to matters beyond the study itself
• Utilization/applications/action orientation (‘pragmatic validity’): degree to which programs or actions result from a study’s findings and/or the degree to which ethical issues are forthrightly dealt with

Observer bias

‘Observer effects’= tendency of people to change their behaviour because they know they are being observed. Ways to minimize the bias:

• Naturalness of observation provides some inoculation against bias
• Observation research is emergent: to avoid predetermined categories, the research can shift the question(s) he or she is pursuing.
• Combination with other techniques for the collections of information. The process of triangulations is a good hedge against the biases that result from ‘pure’ observation.

Observations in public spaces

Nature of setting is almost always the preferred technique. Some public spaces are fairly clearly delineated (e.g. airport) others less so, but all provide the context for studies involving moral order, interpersonal relations, and norms for dealing with different categories for individuals. Observational studies in public spaces allow researchers to gather data on large groups of people and thereby to identify patterns of group behaviour.

Most famous example of space observation: Humphreys (1975) who adopted a covert observer-as-participant role in a public bathroom. He observed men engaging in impersonal homosexual encounters and concluded that men in this setting adopt one of several possible roles.

Ethics: a researcher can be guilty by entering into places that can be construed as private even though they have a public character. The usual answer is that studying sensitive subjects is not taboo but doing so without the express permission of the participants is ethically wrong. Generally, it applies:

• Unethical to deliberately misrepresent his or her identity for the purpose of entering a private domain
• Unethical to deliberately misrepresent the character of the research in which he or she is engaged.

Chapter I :  The Concept of Analysis

There will be two chapters devoted to analysis. This chapter will cover the different approaches and practices of analysis. Also the features analytic methods need to hold for effective and penetrative investigation and the stages and processes involved in analysis will be covered.

Approaches to qualitative analysis are compared according to the way they address a number of different issues such as:

• The status of the data: data are referred to phenomena such as feelings, perceptions and events, which exist apart from the setting in which the data were captured.
• The primary focus of analysis: analyses are different in what they focus on. Discourse analysis, conversation analysis and narrative analysis focus on the language and the construction and structure of talk, text and interaction. Content analysis, grounded theory and policy analysis are concerned with interpreting common sense and substantive meaning in the data.
• The way data are reduced: for example paring down statements to their core meaning, thematic summaries, collective analytical categorization (categorizing the data), identification of the overall structure in the data or graphic displays of synthesized data.
• The kinds of concepts generated: identification of key themes, concepts or categories. The nature of these concepts and the way in which they are generated varies between approaches.
• The way concepts are applied to the data: two main ways; cross-sectional ‘code and retrieve’ methods and non-cross-sectional analysis. The first means devising a common system of categories and applying it to the data. The second means looking at different parts of the data separately and conceptualizing categories on these different parts.
• The extent to which data are retained in context: the way in which approaches treat the context and retain links to the original data. Code and retrieve approaches, for example, group and compare data chunks outside the context. This is often criticized as it does not provide the full picture.
• The way ‘analysed’ data are accessed and displayed: researchers take different approaches to accessing their already analysed data. Code and retrieve methods view the data in textual chunks, sorted by category. Other methods work directly from raw data.
• The explicit level of abstraction: the way in which the researcher describes the data and patterns. The extent to which this is detailed varies among approaches.
• The status of categories and the logic of explanation: the way in which categories are explained. Some researchers treat categories as variables explain a causal relationship with those variables (X leads to Y). Others say this is not possible with qualitative data.
• The place of the researcher in the analytical account: the way in which the researcher is considered in the analysis. Some approaches treat the researcher as not being of influence to the evidence while others see the researcher as an integral part of the interpretation.

Miles (1995) categorized five different types of computer-assisted qualitative data analysis software (CAQDAS).

1. Text retrievers: searches large amounts of data for words or phrases.
2. Text base managers: provide a structure to the data and are searchable like text retrievers.
3. Code and retrieve programs: allow you to label or tag passages to make them easier to find.
4. Code based theory builders: allow one to create links between different aspects of the data.
5. Conceptual network builders: facilitate graphic display of the data set concepts.

It is important to notice that there is not one best computer program. The researcher should see how well each program fits to their needs. However, the usage of CAQDAS methods offers the researcher more speed, consistency and eases the linking of data. A downside is that these programs take the segments of text out of their context and encourage the researcher to take shortcuts.

Researchers need certain tools to carry out their analysis as the data collected is likely to be very raw and difficult to interpret at first. These tools make the task of managing the evidence in a proper way easier for the investigators. It is important to choose a tool or ‘analytical support’ that maximizes the full potential of a qualitative analysis. Therefore, there are certain crucial characteristics of tools in any method used to investigate qualitative data:

• Remains grounded in the data: concepts should be rooted within the data, rather than simply being imposed on the data. Therefore, a structure that allows concepts or patterns to emerge from the data should be present in every method. The tools should thus allow patterns to be shown rather than a researcher devising patterns himself and forming them to the data.
• Permits captured synthesis: reducing the raw data to usable concepts still need to be done carefully in order to capture the original views of the participants. Therefore, the synthesis should be captured so that it can always be checked back against the original material.
• Facilitates and displays ordering: ordering the evidence should be done in such a way that it can be inspected in related blocks at any time during the study.

• Permits within and between case searches: the researcher should be able to easily search through the data to find relationships. This requires tools for three different types of search, namely (1) thematic categories and patterns across different cases, (2) associations between phenomena within one case, and (3) associations in phenomena between groups of cases.
• Allows systematic and comprehensive coverage of the data set: the analysis needs to be systematically applied to all cases. This means all cases are analysed in the same way.
• Permits flexibility: the method should allow the model to be amendable if new ideas emerge.
• Allows transparency to others: the content should be accessible to others in order to perform a follow-up study. Others should be able to review how the data was analysed.

In order to make the data set more manageable for the researcher, an analytical hierarchy was developed. It is made up out of different parts which involve various analytical tasks. This makes it easier for the researcher to gain an overview and make sense of the data. He or she can move between the stages of the hierarchy which helps to produce more reliable results of the analysis. There are three main stages, namely data management, descriptive accounts and explanatory accounts.

At the start of the analytic process is data management. Here, the researcher sorts the data to make it more manageable. It involves generating concepts to which data are labelled and sorted. It may be carried out manually or through CAQDAS programs.

Secondly, the researcher makes use of this ordered data to identify key dimensions. Two features of qualitative data are essential to the analysis, namely the language participants use (which shows how strongly they feel about a phenomenon) and the substantive content of people’s accounts (the importance of the content of each case). After the data have been classified into refined categories, typologies can be developed. These are specific forms of classification and can be divided into two major forms; indigenous and analyst constructed. The first are classification systems devised by the participants themselves. The latter are created during the analytical process and classify patterns emerging from the data.

Finally, explanatory accounts are developed. The analyst must try to find patterns of associations within the data to move from descriptive to explanatory accounts. This means that, after finding the main concepts described in the data, the researcher will investigate the patterns and relationships among concepts and try to explain why these patterns occurs. Some researchers find that qualitative analysis cannot be used to find causal relationships (X leads to Y). However, it is an important research method in order to clarify the nature of different factors and the interrelationship between concepts.