Md. Abdullah Al Muzahid khan
Department of Anthropology
University of Rajshahi
The core of qualitative analysis lies on three related processes: describing phenomena, classifying it and seeing how the concepts interconnect. Dey draws these as a circular process (p. 31) to show that they interconnect each other. But because qualitative analysis is iterative process, he also represents them by iterative spiral (p. 53).
Qualitative analysis is to develop thorough and comprehensive description of the phenomenon under study. Geerz (1973) and Denzin (1978) call this as ‘thick’ description. If ‘thin’ description merely states ‘facts’, a ‘thick’ description includes information about the context of an act, the intentions and meanings that organize action, and its subsequent evolution (Denzin, 1978). Thus description encompasses the contexts of action, the intentions of actor, and the process in which action is embedded.
Qualitative data are words rather than numbers. Words describe and explain. Words suggest new perspectives. Conclusions expressed as words seem more convincing than pages of numbers. But words are also ambiguous and difficult to compare objectively. It is never clear how much of a verbal description of one instance carries over to other instances. One observer's description, however precise, may not concur with another's. "It is easy for a qualitative researcher to jump to hasty, partial, unfounded conclusions" (Miles & Huberman [M&H], 1984, p. 15).
Alvin Toffler (in Coveney and Highfield, 1991) said that, we are so good at dissecting data that we often forget how to put the pieces back together again. This problem will not arise if description and classification are not ends in themselves but must serve an overriding purpose, that is to produce an account for analysis. For that purpose we need to make connections among building block of concepts of our analysis.
Specific analytic strategies of Qualitative Data
Although a description of the actual procedural details and nuances of every qualitative data analysis strategy is well beyond the scope of a short paper, a general appreciation of the theoretical assumptions underlying some of the more common approaches can be helpful in understanding what a researcher is trying to say about how data were sorted, organised, conceptualised, refined, and interpreted.
The strategy you use will depend on your research topic, your personal preferences and the time, equipment and finances available to you. Also, qualitative data analysis is a very personal process, with few rigid rules and procedures.
CONSTANT COMPARATIVE ANALYSIS
Many qualitative analytic strategies rely on a general approach called "constant comparative analysis". Originally developed for use in the grounded theory methodology of Glaser and Strauss,4 which itself evolved out of the social or anthropological theory of symbolic interactionism, this strategy involves taking one piece of data (one interview, one statement, one theme) and comparing it with all others that may be similar or different in order to develop conceptualisations of the possible relations between various pieces of data. For example, by comparing the accounts of 2 different people who had a similar experience, a researcher might pose analytical questions like: why is this different from that? and how are these 2 related? In many qualitative studies whose purpose it is to generate knowledge about common patterns and themes within human experience, this process continues with the comparison of each new interview or account until all have been compared with each other. A good example of this process is reported in a grounded theory study of how adults with brain injury cope with the social attitudes they face (see Evidence-Based Nursing, April 1999, p64).
Constant comparison analysis is well suited to grounded theory because this design is specifically used to study those human phenomena for which the researcher assumes that fundamental social processes explain something of human behaviour and experience, such as stages of grieving or processes of recovery. However, many other methodologies draw from this analytical strategy to create knowledge that is more generally descriptive or interpretive, such as coping with cancer, or living with illness. Naturalistic inquiry, thematic analysis, and interpretive description are methods that depend on constant comparative analysis processes to develop ways of understanding human phenomena within the context in which they are experienced.
When data is analysed by theme, it is called thematic analysis. This type of analysis is highly inductive, that is, the themes emerge from the data and are not imposed upon it by the researcher. In this type of analysis, the data collection and analysis take place simultaneously. Even background reading can form part of the analysis process, especially if it can help to explain an emerging theme. Closely connected to thematic analysis is comparative analysis. Using this method, data from different people is compared and contrasted and the process continues until the researcher is satisfied that no new issues are arising. Comparative and thematic analyses are often used in the same project, with the researcher moving backwards and forwards between transcripts, memos, notes and the research literature.
Constant comparative analysis is not the only approach in qualitative research. Some qualitative methods are not oriented toward finding patterns and commonalities within human experience, but instead seek to discover some of the underlying structure or essence of that experience through the intensive study of individual cases. For example, rather than explain the stages and transitions within grieving that are common to people in various circumstances, a phenomenological study might attempt to uncover and describe the essential nature of grieving and represent it in such a manner that a person who had not grieved might begin to appreciate the phenomenon. The analytic methods that would be employed in these studies explicitly avoid cross comparisons and instead orient the researcher toward the depth and detail that can be appreciated only through an exhaustive, systematic, and reflective study of experiences as they are lived.
Although constant comparative methods might well permit the analyst to use some pre-existing or emergent theory against which to test all new pieces of data that are collected, these more phenomenological approaches typically challenge the researcher to set aside or "bracket" all such preconceptions so that they can work inductively with the data to generate entirely new descriptions and conceptualisations. There are numerous forms of phenomenological research; however, many of the most popular approaches used by nurses derive from the philosophical work of Husserl on modes of awareness (epistemology) and the hermeneutic tradition of Heidegger, which emphasises modes of being (ontology).5 These approaches differ from one another in the degree to which interpretation is acceptable, but both represent strategies for immersing oneself in data, engaging with data reflectively, and generating a rich description that will enlighten a reader as to the deeper essential structures underlying a particular human experience. Examples of the kinds of human experience that are amenable to this type of inquiry are the suffering experienced by individuals who have a drinking problem (see Evidence-Based Nursing, October 1998, p134) and the emotional experiences of parents of terminally ill adolescents (see Evidence-Based Nursing, October 1999, p132). Sometimes authors explain their approaches not by the phenomenological position they have adopted, but by naming the theorist whose specific techniques they are borrowing. Colaizzi and Giorgi are phenomenologists who have rendered the phenomenological attitude into a set of manageable steps and processes for working with such data and have therefore become popular reference sources among phenomenological nurse researchers.
Ethnographic research methods derive from anthropology's tradition of interpreting the processes and products of cultural behaviour. Ethnographers documented such aspects of human experience as beliefs, kinship patterns and ways of living. In the healthcare field, nurses and others have used ethnographic methods to uncover and record variations in how different social and cultural groups understand and enact health and illness. An example of this kind of study is an investigation of how older adults adjust to living in a nursing home environment (see Evidence-Based Nursing, October 1999, p136). When a researcher claims to have used ethnographic methods, we can assume that he or she has come to know a culture or group through immersion and engagement in fieldwork or participant observation and has also undertaken to portray that culture through text.6 Ethnographic analysis uses an iterative process in which cultural ideas that arise during active involvement "in the field" are transformed, translated, or represented in a written document. It involves sifting and sorting through pieces of data to detect and interpret thematic categorisations, search for inconsistencies and contradictions, and generate conclusions about what is happening and why.
For those types of analyses at the other end of the qualitative data continuum, the process is much more mechanical with the analysis being left until the data has been collected. Perhaps the most common method of doing this is to code by content. This is called content analysis. Using this method the researcher systematically works through each transcript assigning codes, which may be numbers or words, to specific characteristics within the text. The researcher may already have a list of categories or she may read through each transcript and let the categories emerge from the data. Some researchers may adopt both approaches. This type of analysis can be used for open-ended questions which have been added to questionnaires in large quantitative surveys, thus enabling the researcher to quantify the answers.
NARRATIVE AND DISCOURSE ANALYSIS
Falling in the middle of the qualitative analysis continuum is discourse analysis, which some researchers have named conversational analysis, although others would argue that the two are quite different. These methods look at patterns of speech, such as how people talk about a particular subject, what metaphors they use, how they take turns in conversation, and so on. These analysts see speech as a performance; it performs an action rather than describes a specific state of affairs or specific state of mind. Much of this analysis is intuitive and reflective, but it may also involve some form of counting, such as counting instances of turn-taking and their influence on the conversation and the way in which people speak to others.
Cognitive processes inherent in qualitative analysis
The term "qualitative research" encompasses a wide range of philosophical positions, methodological strategies, and analytical procedures. Morse1 has summarised the cognitive processes involved in qualitative research in a way that can help us to better understand how the researcher's cognitive processes interact with qualitative data to bring about findings and generate new knowledge. Morse believes that all qualitative analysis, regardless of the specific approach, involves:
comprehending the phenomenon under study
synthesising a portrait of the phenomenon that accounts for relations and linkages within its aspects
theorising about how and why these relations appear as they do, and
recontextualising, or putting the new knowledge about phenomena and relations back into the context of how others have articulated the evolving knowledge.
Although the form that each of these steps will take may vary according to such factors as the research question, the researcher's orientation to the inquiry, or the setting and context of the study, this set of steps helps to depict a series of intellectual processes by which data in their raw form are considered, examined, and reformulated to become a research product.
Stapes of Qualitative Data Analysis
When we want to go insight into the problem deeply concerning a research, we use qualitative research techniques. It helps to analysis the problem or situation intensely as well as to achieve the objectives of a research. Qualitative techniques such as the use of loosely structured interviews with open-ended questions, (focus) group discussions, observations, projective and participatory approaches will therefore be appropriate in many studies, especially at the onset. For sensitive topics they may be the only reliable techniques. Irrespective of how and for what purpose the data has been collected, the researcher usually ends up with a substantial number of pages of written text that needs to be analyzed. Although procedures and outcomes of qualitative data analysis differ from those of quantitative data analysis, the principles are not so different. In both cases the researcher will have to:
describe the sample populations;
order and reduce/code the data (data processing);
display summaries of data in such a way that interpretation becomes easy, e.g., by preparing compilation sheets, flowcharts, diagrams or matrices;
draw conclusions, relate these to the other data sets of the study and decide how to integrate the data in the report; and
if required, develop strategies for further testing or confirming the (qualitative) data in order to prove their validity.
The Steps of Qualitative Data Analysis :
Producing an account
Using maps & matrices
Associating &linking 12-13 Connecting categories
Reading and annotating 10 Splitting & splicing
Finding a focus 8 creating categories
Miles M.B., Huberman A.M. (1984)
NATURE AND SOURSE OF QUALITATIVE DATA
Sources of QD
a. interviews b. focus Groups c. field observations d. survey comments e. historical records f. secondary data g. photos, paintings, songs ...
Qualitative data is extremely varied in nature. It includes virtually any information that can be captured that is not numerical in nature. Here are some of the major categories or types:
In-Depth Interviews: In-Depth Interviews include both individual interviews (e.g., one-on-one) as well as "group" interviews (including focus groups). The data can be recorded in a wide variety of ways including stenography, audio recording, video recording or written notes. In depth interviews differ from direct observation primarily in the nature of the interaction. In interviews it is assumed that there is a questioner and one or more interviewees. The purpose of the interview is to probe the ideas of the interviewees about the phenomenon of interest.
Direct Observation: Direct observation is meant very broadly here. It differs from interviewing in that the observer does not actively query the respondent. It can include everything from field research where one lives in another context or culture for a period of time to photographs that illustrate some aspect of the phenomenon. The data can be recorded in many of the same ways as interviews (stenography, audio, video) and through pictures, photos or drawings (e.g., those courtroom drawings of witnesses are a form of direct observation).
Written Documents: Usually this refers to existing documents (as opposed transcripts of interviews conducted for the research). It can include newspapers, magazines, books, websites, memos, transcripts of conversations, annual reports, and so on. Usually written documents are analyzed with some form of content analysis.
PROCEDURES FOR PROCESSING AND DISPLAYING OF QUALITATIVE DATA
1. Description of the sample population in relation to sampling procedures:
A useful first step in data processing (as well as in the reporting of findings) is a description of the informants. If numbers allow, relevant background data may be tabulated, for example on age, sex, occupation, education or marital status, as is the practice in quantitative studies. However, as qualitative data originates from small samples (sometimes a handful of key informants or focus group discussions and observations) more information is required to place the data in its context.
For example, who were the key informants, what made you decide to choose them? Who took part in the focus group discussions? How were the participants of the groups selected and how representative are they for your study population? For observations: under what circumstances were they carried out? Who were observed, and by whom?
Unless this type of information is provided, interpretation of data may appear haphazard.
2. Ordering and coding of data:
We will discuss two types of qualitative data: answers to open questions, and more elaborate narratives from loosely structured interviews or FGDs.
§ Answers to open questions
· A first, basic step in the analysis of answers to open questions is to list the answers of a sample of 20-25 informants as they were provided (adding the questionnaire number in order to avoid losing the connection with the informant’s other data).
· Then read the answers carefully, remembering the purpose of the question. The question ‘why are you smoking’ was supposed to help nursing students to develop an intervention against smoking.
· Make rough categories of answers that seem to belong together and code them with a key word. For example, answer 3 (It gives me pleasure) and answer 14 (I like to blow smoke rings) could be labelled with the term ‘pleasure’, which could be abbreviated with the code pleas.
· Then list again all answers but now per code, so that you get some 5-7 short lists, for example:
· Then interpret each list, and end up with some 5-7 meaningful categories with a characteristic key word. For example: Pleasure, being sociable, giving status, giving self-confidence, addiction, defiance. There may be discussion on the need to split up some categories or combine others with few answers. Answers 17 and 18, for example could be put in a separate category reducing stress. In that case there would be seven categories. The category defiance may have two answers: 4. I do not see why I would give up smoking!! and 12. Why not?!! The exclamation marks indicate that defiance rather than lack of knowledge forms the motivation for the answer. Without this addition by the interviewer, these answers would have been difficult to code.Now you can make a tentative interpretation according to the assumed willingness of your informants to change their behaviour. For those who smoke for pleasure or to socialise it might be most easy to give up smoking. Those who are addicted but tried to stop and those who feel they derive status from smoking might form a middle category, whereas for those who smoke to enhance their self-confidence and reduce stress or who are very defiant at the question why they smoke, it might be most difficult to stop.
· Now try a next batch of 20-25 answers and check if the labels work. It is well possible that at this stage still some labels will be changed or that you decide to add new categories or combine others.
· Make a final list of labelled categories and code all data including the data you already processed with the abbreviated codes.
Then discuss whether you will stick to your tentative interpretation of the data and what this means for the content of the messages to address different reasons for smoking. This content analysis is the most important purpose of the analysis. By counting the answers under each label, however, the researcher will gain insight as well in how common the different reasons are.
Both by coding and analysing data the researcher uses his personal knowledge and experiences as tools to make sense of the material (McCRACKEN 1988). Therefore, some of these tools are the researcher's unique impressions, which might remain intangible and undocumented (STRAUSS & CORBIN 1990, McCRACKEN 1988). 
(2) Elaborate narratives
The data from interviews with key informants or focus group discussions (FGDs) are as a rule more bulky than answers to open questions. The carefully transcribed field notes and tapes may consist of pages of narrative text. When analyzing the texts we usually discover that, no matter how good our guidelines for the discussion were, the data contain valuable information but also a number of less essential details. In addition, the data is usually not presented in the order we need for our analysis, since informants may jump from one topic to the other.
To make the analysis easier, we have to order and reduce the data. Ordering is best done in relation to the objectives and the discussion topics. Again, it is best to systematically follow a number of steps.
· Reread your objectives and discussion topics
· Carefully read a number of the interviews, FGDs or narrative observations you want to process. Number the material according to the broad discussion topic it pertains to. Use a yellow marker to highlight particularly illustrative remarks. Use the margins to define sub-topics.
For example, in a gender and leprosy study carried out in different countries, it appeared that the discussion topic stigma had to be differentiated according to different social settings in which it occurred: among close relatives (parents-children), spouses, in-laws, and community members. Further, a distinction had to be made between self-stigmatisation (e.g., a wife diagnosed as a leprosy patient encouraging her husband to marry a second wife in order to prevent divorce, or a patient not attending community meetings for fear of being avoided) and stigmatisation by others. Different degrees of severity in stigmatisation could also be distinguished, varying from slight avoidance to complete expulsion. If stigma would be topic (11) in your discussion list, you would mark everything related to stigma with an (11) in the margin, and add key words such as self-stigm., spouse, in-laws, comm., in the margin, as well as key words such as sleep(ing) sep(arately) or divorce indicating the severity of the stigma.
· List all key words that belong to a certain topic in the sub-categories that have been developed under (2). E.g., everything belonging to stigma could be subdivided and listed in the four major social settings in which stigma was found to manifest itself.
· Interpret the data, e.g., distinguish the major forms in which stigma manifests itself in these different social settings, try to make a ranking order of severity and link it to other variables (such as degree of deformity, socio-economic status) in order to understand differences in stigma.
· Then code all your qualitative data in this way. If necessary, adapt your coding scheme as you order, code and interpret more data. In that case, you should again read and possibly re-code the material you have already processed.
However, instead of developing a very detailed coding system on your rough data, you may also refine your interpretation as you record your roughly coded, summarised data in
Problem of Coding: The difficulty, of course, is in the coding of texts and in finding the patterns. Coding turns qualitative data (texts) into quantitative data (codes), and those codes can be just as arbitrary as the codes we make up in the construction of questionnaires.
I don't remember how many descriptors we came up with, but there were dozens. Some were pretty lame (pour the contents into a beaker and see if the boiling point was higher or lower than that of sea water) and some were pretty imaginative (let's just say that they involved anatomically painful maneuvers), but the point was to show us that there was no end to the number of things we could measure (describe) about that Coke bottle, and the point sunk in. I remember it every time I try to code a text
3. Summarizing data in compilation sheets: After ordering the data we will have to summarize them. A useful first step is summarizing all data of each study unit per study population on separate compilation sheets. Like the master sheets for quantitative data, compilation sheets for qualitative data consist of a number of columns with the topics covered by the study as headings. These may be further sub-divided in smaller themes that you identified and coded when ordering the data (see Annex 23.1). Each interview, FGD or observation gets a number and is successively entered in that sequence on the relevant compilation sheet. If there are different categories of informants within one study population, for example, young mothers and an older generation of mothers, or male and female patients, the data for these groups are entered on separate sheets. If the topics covered in those sub-groups are not completely identical, it is important to be systematic and follow roughly the same sequence of topics for each category of informants. The information inserted is summarized in key words and key sentences, clear enough to remember the statements informants made. (As the number of each study unit is entered in the compilation sheet, it is always possible to go back to the original data and present the full statement, for example in a presentation or in the research report).
Now you have an overview of all data per study population on one or more big sheet(s). If you read the columns, you have a list of answers of all group members on a certain (sub-)topic. If you read horizontally, you can per informant relate different topics to each other or to personal characteristics of the informant. It becomes also easy to compare the answers of different groups on specific issues by comparing compilation sheets.
For example, the personal data of leprosy patients (recently declared cured) and a number of topics and sub-topics discussed with them are presented. Stigma actually experienced, which originally was one topic, has in the compilation sheet been subdivided in the four major social settings in which stigmatisation may occur: close blood relatives, marriage, wider circle of spouse’s relatives and community. In each of those still finer distinctions can be made (e.g., community can be neighbours, friends, work mates, school mates or distant community members). As samples are small, these may all be inserted under the heading ‘community’. Codes (italics) can be added to the statements presented in key words, for example big fear and worried under the heading ‘first reaction’. From the three examples presented, it already appears (confirmed by the analysis of all data in all four countries) that in general the stigma feared when patients hear the diagnosis of leprosy is bigger than the stigma in reality experienced. Patient (12) is in this respect an exception. Ironically, the husband who divorced her had already died from another disease at the moment she was declared cured from leprosy. Horizontal comparison of the data of patient (1) teaches us that it is highly unlikely that the man’s friends do not know about the disease, as even after he has been declared cured he has visible signs. Here the researchers had to interview the friends to find out if indeed this man was (or had not been) stigmatised at all by the community. You may notice that interpretation of data and labelling becomes indeed easy when using compilation sheets, as a researcher can visualize all aspects of his/her informants even if (s)he looks at one aspect at a time for the whole study population. A next step in summarizing may be the combination, contrasting or further analysis of important topics through graphical displays such as matrices, diagrams, flow charts and tables.
4. Further summarizing of data in matrices, figures and tables
Matrices can be used for quantitative as well as qualitative data comparison. In qualitative data we may compare different groups or data sets on important variables, presented in key words. A MATRIX is a chart that looks like a cross-table, but contains words (as well as, sometimes, numbers). In a focus group discussion on changing weaning practices, the researchers listed the answers of young mothers concerning the introduction of soft foods and those of mothers above childbearing age. They then summarized these answers in a matrix:
Figure ---: Matrix on introduction of soft baby foods among mothers of different age groups
This type of display made it easy for the researchers to conclude that:
Younger mothers start giving soft foods, on average, 2.5 months earlier than the generation of their own mothers;
Younger mothers use a larger variety of soft weaning foods than women in the preceding generations; and
Younger mothers give soft foods to their babies more frequently, but for the same reasons as their mothers did.
Rasch analysis helps qualitative researchers by: 1) aiding in the clear conceptualization and construction of unidimensional variables, 2) identifying useful rating scale categorizations, and 3) enabling qualitative results to be reduced (summarized) into simple metric forms for plotting and further analysis.
A Rasch analysis of M&H's Chart 13a (M&H p. 97) illustrates how Rasch analysis can simplify qualitative research. Four types of subjects are rated on 12 conditions with respect to preparedness. M&H's published chart is crammed with anecdotal information. An abbreviated Chart 13a is shown here. Even in this compressed version the reader is forced to pick and choose which cells to regard as important in order to form an overall impression of the situation.
Matrices facilitate data analysis considerably. They are the most common form of graphic display of qualitative data. They can be used to order and compare information in many ways, for example, according to:
Time sequence (of procedures being investigated in different periods, for example),
Type of informants (as in the example above), or
Tocation of data collection (to visualize differences between rural and urban populations).
A DIAGRAM is a figure with boxes containing variables and arrows indicating the relationships between these variables. When analyzing the problems you wanted to investigate during the development of your protocols, most groups developed a diagram. In a similar way diagrams can be developed to summarize findings of a study. (See Figures 23.2 and 23.3).
You might use a diagram to illustrate a crucial issue in your study, combining all available qualitative and quantitative data collected.
Figure 23.2: Reasons for early introduction of soft foods by young mothers
Diagrams, like matrices, can be of great assistance in providing an overview of the data collected and in guiding data analysis.
Figure 23.3: Reasons for late introduction of soft foods by young mothers
FLOW CHARTS are special types of diagrams that express the logical sequence of actions or decisions. The figure preceding Modules 1-18, indicating the successive steps in protocol development, is an example of a flow chart.
Flow charts are especially useful to summarise different flows of events that are mutually connected. A counselling team in Bulawayo, Zimbabwe, for example, which interviewed some 95 HIV positive persons in-depth over a period of two years, summarised the roughly 100 pages of interview material for each informant by drawing five lines (see Figure 23.4). One central line presented the development of the disease over time, with
crises and periods of relative well-being. Another line presented different forms of medical care sought, a third the flaws in economic status connected to the disease (e.g., loss of job, seeking employment elsewhere), a fourth the possible changes in social status such as divorce or (re)marriage, whereas a fifth line presented the patient’s emotional status linked to events occurring in the four other fields (e.g., positive coping, depression). These flow charts were extremely useful for comparison of data, per
informant and between different groups of informants (e.g. males/females, single/married). They highlighted the impact of the disease on the lives of different groups of patients and their way of coping with it.*
*Meursing (1997) A world of silence.
Figure 23.4: Flowchart on coping of HIV+ persons with their condition over time
… in the next page.
A TABLE is a chart with rows and columns that has numbers in the various cells or boxes. Qualitative data can also be categorized, coded, inserted in master sheets or computer and counted, together with other quantitative data, and displayed in tables. Answers to open-ended questions in questionnaires will usually be categorized and summarised in this way. However, you will in the first place want to analyze the content of the individual answers in each category. (See section II-2 and section III in this module.)
DRAWING AND VERIFYING CONCLUSIONS
Drawing and verifying conclusions is the essence of data analysis. It is not an isolated activity, however. When we start summarizing our data in compilation sheets, flowcharts, matrices or diagrams, we continuously draw conclusions, and modify or reject quite a number of them as we proceed. Writing helps generate new ideas as well. Therefore writing should start as early as possible, right from the onset of data processing and analysis, if only for ourselves. No creative insights should get lost!
Collection, processing, analysis and reporting of qualitative data are closely intertwined, and not (as is the case with quantitative data) distinct successive steps. It may often be necessary to go back to the original field notes and verify conclusions, collect additional data if available data appear controversial, and get feedback from all parties concerned.
Identifying variables and associations between variables:
During or at the end of the study it will be possible to define certain variables and search for associations with other variables, without having the prior aim of measuring them. Many studies have qualitative parts with open questions, key informant interviews, focus group discussions or observations for the purpose of identifying these variables. The researcher who uses such a qualitative approach should be like a detective who searches for evidence, accounts for countervailing evidence, and verifies the findings by looking for independent, supporting evidence, until (s)he is confident about possible associations among certain variables which shed light on the problem under investigation.
For example, if we find among the mothers who wean their children early that quite a number have jobs, we may assume that having a job contributes to early weaning. Similar studies carried out elsewhere with similar findings support this assumption (independent evidence). Only if there are very few employed women who wean their children late, however, can we be more certain that our assumption is true, and for each of those exceptions we should try to find an explanation. Do the mothers take their children with them (crÃ¨che at place of work) or do they work near their homes so that they can feed the baby during breaks? Or do they successfully combine breast-milk with alternatives? If yes, why don’t more mothers try this combination? etc. etc.
Finding confounding or intervening variables
Sometimes variables appear to be related but the association cannot easily be explained. Other times it seems that variables should logically go together, but you cannot find a relationship. In cases such as these there may be another variable (‘Q’) influencing the association between the two variables concerned, that has to be identified.
For example, one expects a relationship between the quality of drinking water and the incidence of diarrhea. It is assumed that the incidence of diarrhoea would decrease as the number of water faucets in a village increased. If there is no change over time, there might be a confounding variable. People, for example, may dislike the taste of tap-water so much that they use it for everything, except for drinking.
Such unexplained associations may appear in any study. The essential characteristic of a qualitative research approach is that it purposively looks for such associations during the fieldwork, and that additional questions and tools may be developed to highlight such relationships. In quantitative surveys that attempt to objectively measure the strength of a presupposed association between two variables, the tools should not be changed once the fieldwork is ongoing.
Integrating qualitative and quantitative data:
Thus far we have discussed the analysis of qualitative data as a separate activity. However, if a research team has collected qualitative as well as quantitative data, which is the case in most qualitative studies, it would be foolish not to look at them in combination, as this can inspire to deeper and more rewarding analysis.
For example, the Indonesian ‘gender and leprosy’ research team found, when analyzing the registration data of 4500 new leprosy patients who had registered over the past five years, that the M/F ratio was most unfavourable in the age group of 15-44 years. This was a puzzling finding, as in Nepal women in this age group were reporting much better (though still less than men). In-depth interviews with staff revealed that they suspected adolescent girls and young women to hide their skin patches, because of shameful associations with dirt, ugliness. This provided the incentive for a further break down of the quantitative data, which revealed that the M/F difference in reporting was indeed most pronounced in the 15-34 age group, and leveled off above 35. The reason(s) for this relatively large gender difference in the younger age groups were then further explored.
Content analysis of qualitative data for action:
Content Analysis: Content analysis is the analysis of text documents. The analysis can be quantitative, qualitative or both. Typically, the major purpose of content analysis is to identify patterns in text. Content analysis is an extremely broad area of research. It includes: Thematic analysis of text :(The identification of themes or major ideas in a document or set of documents. The documents can be any kind of text including field notes, newspaper articles, technical papers or organizational memos).
Indexing: There are a wide variety of automated methods for rapidly indexing text documents. For instance, Key Words in Context (KWIC) analysis is a computer analysis of text data. A computer program scans the text and indexes all key words. A key word is any term in the text that is not included in an exception dictionary. Typically you would set up an exception dictionary that includes all non-essential words like "is", "and", and "of". All key words are alphabetized and are listed with the text that precedes and follows it so the researcher can see the word in the context in which it occurred in the text. In an analysis of interview text, for instance, one could easily identify all uses of the term "abuse" and the context in which they were used.
Quantitative data serve in the first place to convince health authorities that there is indeed a serious, sizeable problem; qualitative data help to provide ideas on how to solve it. The FGDs on weaning foods with young mothers and mothers who had surpassed the childbearing age, for example, will yield many suggestions on how to develop interventions with the mothers which they are likely to consider useful and will be able to implement. Likewise, the in-depth interviews with leprosy and ex-leprosy patients will provide new insights into how best to counsel new patients and their close relatives/spouses in order to reduce unnecessary fears.
Computer analysis of qualitative data:
For the past 10 years or so, the majority of qualitative researchers have used word-processors to transcribe their interviews. Furthermore, since around 1996 in the UK we have witnessed a huge growth in the use of computer-assisted qualitative data analysis software (CAQDAS) packages in qualitative research. CAQDAS software, such as ATLAS-ti, NUDIST and WINMAX are rapidly becoming the accepted tool for handling the description and interpretation of qualitative data. For qualitative data, issues about preservation of data from these packages are something we have had to address with some urgency. These are proprietary software packages and in the past it has not been possible to import and export data from one package to another. qualitative data has developed guidelines on what to keep for archival purposes—i.e. extracting or reducing the data to its simplest form—ASCII text or rtf. As expected, in the past year we have seen software developers taking steps to encourage sharing between packages, for example adding export and import facilities to their programmes. Qualitative data is liaising with the developers to discuss extending the functionality of the packages to include archiving features.
With the ever-increasing importance of computers in research, strategies for analyzing qualitative data by computer have been/are being developed. There are several possibilities, ranging from simple word processing programs to highly sophisticated Qualitative Data Management Software including possibilities for statistical testing of associations. As numbers are usually small in Health System research and content analysis, which can be done by hand, is most likely more important than testing of associations, we will not elaborate these techniques here. Rather we refer the interested students to Anthropology or Psychology Departments at universities that have experience with programs such as Qualitan or SPSS for qualitative data processing.
REPORTING THE DATA
"The report is itself a social construction in which the author's choice of writing style and literary devices provide a specific view on the subjects' lived world." (KVALE 1996, p.253) 
Basically, there are two ways of reporting qualitative data that form part of a study in which different research techniques were used. One way is summarizing the major qualitative results in a separate section of the findings, with examples and quotations, following the objectives that guided the collection of this particular data. The results would then be discussed in the chapter ‘Discussion’, together with the results of other, more quantitative data collection tools and would subsequently be reflected in the summary of the findings and the recommendations.
Another possibility is to fully integrate different data sets in the chapter of findings, ordered according to the objectives of the entire study. If quantitative and qualitative data have been analyzed and sometimes even collected in an integrated way, it would also be logical to present them in an integrated fashion. Attention should be paid that no valuable data get lost. Therefore a rough draft of all important findings is required in any case, after which can be decided to present the data either in separate sections or chopped up for integration with other data.
Quality measures in qualitative analysis
It used to be a tradition among qualitative researchers to claim that such issues as reliability and validity were irrelevant to the qualitative enterprise. Instead, they might say that the proof of the quality of the work rested entirely on the reader's acceptance or rejection of the claims that were made. If the findings "rang true" to the intended audience, then the qualitative study was considered successful. More recently, nurse researchers have taken a lead among their colleagues in other disciplines in trying to work out more formally how the quality of a piece of qualitative research might be judged. Many of these researchers have concluded that systematic, rigorous, and auditable analytical processes are among the most significant factors distinguishing good from poor quality research. Researchers are therefore encouraged to articulate their findings in such a manner that the logical processes by which they were developed are accessible to a critical reader, the relation between the actual data and the conclusions about data is explicit, and the claims made in relation to the data set are rendered credible and believable. Through this short description of analytical approaches, readers will be in a better position to critically evaluate individual qualitative studies, and decide whether and when to apply the findings of such studies to their respective sectors
FURTHER STRATEGIES FOR TESTING OR CONFIRMING QUALITATIVE FINDINGS TO PROVE VALIDITY
Researchers who use quantitative research designs reduce their data to numbers and apply statistical tests. This does not necessarily insure that their research results are valid: something may have gone wrong during sampling or collection of data or even in the earlier design of the study (overlooking possible confounding variables). The following strategies will therefore be of use to any researcher. They are particularly relevant, however, to qualitative research, since the small numbers of qualitative data often generate questions concerning its validity.
1. Check for representativeness of data.
Although in qualitative research informants have usually not been selected randomly, they must have been selected systematically, according to previously established rules. Check whether you have indeed interviewed all categories of informants needed to get a complete picture of your topic (not relying excessively on talkative authorities). Make sure that you do not generalize from unrepresentative events.
2. Check for bias due to observer bias or the influence of the researcher on the research situation.
3. Cross-check data with evidence from other, independent sources (Triangulation).
These sources may be different independent informants, different research techniques employed to investigate the same topic, or results from other, similar studies. The data should confirm or at least not contradict each other. Actively cross-checking data, looking for independent evidence or counter-evidence, is one of the most important ways to enhance the validity of research data. For example, answers of husbands and wives (and other informants concerned) should confirm each other on such issues as who decides whether and what family planning methods should be used, who decides whether daughters should be circumcised, or what has changed in husband-wife relationships after the diagnosis of leprosy or another feared disease in one of the spouses.
4. Compare and contrast data.
Comparison will often have been built into the research design through including different categories of informants. If we want to be sure, for example, that variable A (high level of education) influences variable B (use of family planning methods) we have to compare a group of mothers with high education to a group of mothers with low education on their use of family planning methods. Comparing and contrasting data is important if you are attempting to identify your variables as well as to confirm associations among variables.
5. Use extreme (groups of) informants to the maximum.
In the discussion of study design and sampling we stated that it may be useful to look for categories of informants that represent the extremes on a certain variable. For example, you may find it most useful to study ‘drop-outs’ and regular attendees of TB services, leaving out the category of irregular attendees. This may be the most efficient way of identifying the key variables that influence the attendance behaviour of TB patients.
6. Do additional research to test the findings of your study.
The results of your study may be so intriguing that you decide to do a follow-up study afterwards. Such a study may be undertaken for several reasons: to replicate certain findings, to rule out (or identify) possible intervening variables, to rule out rival explanations by investigating them, or to look for negative evidence. Additional studies undertaken for one or more of these reasons may serve to make the results of your original study more convincing.
7. Get feedback from your informants.
You need to involve all parties concerned in the various stages of the research. This is important not only for ethical reasons or because it will improve the chances that the results will be implemented, but also because it will improve the quality of your study design, of your data, and of the conclusions drawn from these data. Suggestions and additional information collected during feedback sessions will invariably increase the quality of your research report
The Qualitative data analysis ( QDA) Problem
Coding is one of the steps in what is often called "qualitative data analysis," or QDA. Deciding on themes or codes is an unmitigated, qualitative act of analysis in the conduct of a particular study, guided by intuition and experience about what is important and what is unimportant. Once data are coded, statistical treatment is a matter of data processing, followed by further acts of data analysis.
When it comes right down to it, qualitative data (text) and quantitative data (numbers) can be analyzed by quantitative and qualitative methods. In fact, in the phrases "qualitative data analysis" and "quantitative data analysis," it is impossible to tell if the adjectives "qualitative" and "quantitative" modify the simple noun "data" or the compound noun "data analysis." It turns out, of course, that both QDA phrases get used in both ways. Consider the following table:
Cell a is the qualitative analysis of qualitative data. Interpretive studies of texts are of this kind. At the other extreme, studies of the cell d variety involve, for example, the statistical analysis of questionnaire data, as well as more mathematical kinds of analysis.
Cell b is the qualitative analysis of quantitative data. It's the search for, and the presentation of, meaning in the results of quantitative data processing. It's what quantitative analysts do after they get through doing the work in cell d. Without the work in cell b, cell d studies are puerile.
Which leaves cell c, the quantitative analysis of qualitative data. This involves turning the data from words or images into numbers. Scholars in communications, for example, might tag a set of television ads from Mexico and the U.S. in order test whether consumers are portrayed as older in one country than in the other. Political scientists might code the rhetoric of a presidential debate to look for patterns and predictors. Archeologists might code a set of artifacts to produce emergent categories or styles, or to test whether some intrusive artifacts can be traced to a source. Cultural anthropologists might test hypotheses across cultures by coding data from the million-pages of ethnography in the Human Relations Area Files and then doing a statistical analysis on the set of codes. Strictly speaking, then, there is no such thing as a quantitative analysis of qualitative data. The qualitative data (artifacts, speeches, ethnographies, TV ads) have to be turned first into a matrix, where the rows are units of analysis (artifacts, speeches, cultures, TV ads), the columns are variables, and the cells are values for each unit of analysis on each variable.
On the other hand, the idea of a qualitative analysis of qualitative data is not so clear-cut, either. It's tempting to think that qualitative analysis of text (analysis of text without any recourse to coding and counting) keeps you somehow "close to the data." I've heard a lot of this kind of talk, especially on e-mail lists about working with qualitative data.
Morse JM. "Emerging from the data": the cognitive processes of analysis in qualitative inquiry. In: JM Morse, editor. Critical issues in qualitative research methods. Thousand Oaks, CA: Sage, 1994:23–43.
Holloway I. Basic concepts for qualitative research. Oxford: Blackwell Science, 1997.
Schwandt TA. Qualitative inquiry: a dictionary of terms. Thousand Oaks, CA: Sage, 1997.
Glaser BG, Strauss AL. The discovery of grounded theory. Hawthorne, NY: Aldine, 1967.
Ray MA. The richness of phenomenology: philosophic, theoretic, and methodologic concerns. In: J M Morse, editor. Critical issues in qualitative research methods. Thousand Oaks, CA: Sage, 1994:117–33.
Boyle JS. Styles of ethnography. In: JM Morse, editor. Critical issues in qualitative research methods.. Thousand Oaks, CA: Sage, 1994:159–85.
Sandelowski M. We are the stories we tell: narrative knowing in nursing practice. J Holist Nurs 1994;12:23–33.[Abstract]
Boutain DM. Critical language and discourse study: their transformative relevance for critical nursing inquiry. ANS Adv Nurs Sci 1999;21:1–8.
Thorne S. The art (and science) of critiquing qualitative research. In: JM Morse, editor. Completing a qualitative project: details and dialogue. Thousand Oaks, CA: Sage, 1997:117–32.
Alvesson, Mats & Sköldberg, Kaj (2000). Reflexive Methodology—New Vistas for Qualitative Research. London: Sage.
Burgess, Robert G. (1984). In the Field—An Introduction to Field Research. London: Routledge.
Corti, Louise (2000). Progress and Problems of Preserving and Providing Access to Qualitative Data for Social Research—The International Picture of an Emerging Culture [58 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3).
Denzin, Norman K. & Lincoln, Yvonna S. (1994). Handbook of Qualitative Research. Thousand Oaks: Sage.
Jensen, Mogens Kjaer (1991). Kvalitative metoder i anvendt samfundsforskning, Rapport 91-1, København, Socialforskningsinstituttet.
Kluge, Susann & Opitz, Diane (2000, December). Computer-aided Archiving of Qualitative Data with the Database System "QBiQ" [36 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3).
Kuula, Arja (2000, December). Making Qualitative Data Fit the "Data Documentation Initiative" or Vice Versa? [28 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3).
Kvale, Steinar (1996). InterViews—An Introduction to Qualitative Research Interviewing. Thousand Oaks: Sage.
Lave, Jean & Kvale, Steinar (1995). What is anthropological research? An interview with Jean Lave by Steinar Kvale. Qualitative Studies in Education, 8(3), 219-228.
McCracken, D. Grants (1988). The Long Interview (Qualitative Research Methods, Vol. 13). Newbury Park: Sage.
Pedersen, Søren Hviid (1999). Qualitative and Quantitative research strategies: towards a possible convergence. Conference paper at the IASSIST (International Association for Social Science Information Service and Technology) Conference 17-21 May 1999.
Strauss, Anselm & Corbin, Juliet (1990). Basics of Qualitative Research. Newbury Park: Sage.
Wallerstein, Immanuel; Juma, Calestous; Keller, Evelyn Fox; Kocka, Jürgen; Lecourt, Dominique; Mudimbe, v.y.; Kenan, William R.; Mushakoji, Kinhide; Prigogine Ilya; Taylor, Peter J. & Trouillot, Michel-Rolph (1998). Luk samfundsvidenskaberne op!. Report from "The Gulbenkian Commission of the Restructuring of the Social Sciences". Copenhagen: Roskilde Universitetsforlag.
Fink, Anne Sofia (2000). The Role of the Researcher in Qualitative Research [69 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3).
Glaser, Barney G. & Strauss, Anselm L. (1967). The Discovery of Grounded Theory: Strategies for Qualitative Research. Chicago: Aldine.
James, Jacqueline & Sorensen, Annemette (2000). Archiving Longitudinal Data for Future Research. Why Qualitative Data Add to a Study's Usefulness [59 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3.
Thompson, Paul (2000). Experiences of Re-analysing Data in Qualitative Research [48 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 1(3).
Miles M.B., Huberman A.M. (1984) Qualitative Data Analysis: A Sourcebook of New Methods. Newbury Park, CA: Sage