An Outline for a Methodology of Qualitative Social Research (1)(1982)
[Author's note February 2001 - later additions and explanations are put into square brackets]:
This is a translation of the first presentation of the "qualitative-heuristic" methodology for the social sciences in 1982. For the benefit of the present reader I offer some comments on this quasi "historical" paper:
The text of the paper has been slightly shortened. For a full account of notes and literature (up to 1981) the reader is referred to the original publication in German (1982).
A short essay on an extensive theme could well be so sketchy that it fails in its objectives.
A methodology couched in generalities and without examples, is seldom more than a framework or scaffolding. Within the present context, however, too much detail is not possible. A brief sketch will allow the author to concentrate on points of view which he considers to be of prime importance.
The present essay has as its central premise the claim that a qualitative methodology is necessary, both as counterpart and complement to the quantitative methodology - even when it is obvious that the latter, in areas such as sampling theory, scaling techniques and electronic processing of data, has reached a high level of sophistication.(2) Comparable literature on qualitative methods in the social sciences does not exist.(3)
The preponderance of contributions on methods using quantitative thinking is such that many may have the impression that quantitative methods are the only ones, or the only "scientific" ones, in social science. It will be argued - and demonstrated - that this is not the case. The precise position of qualitative social research within the totality of social research, the rules governing this distinctive approach, as well as the particular type of data it yields must be clarified.
The many particular qualitative techniques cannot here be discussed in detail; the contention which should be made clear, however, is that every known quantitative procedure can be contrasted with a definite qualitative procedure; while taken together, the two categories of procedures show characteristic distinctions. Qualitative social research is applicable to the whole range of themes in empirical sociology; in fact, it exceeds quantitative method and involves art and literature - even beyond their "quantitative content".
The present article does not concern itself with recent views of literature and also does not take this as a point of departure. My essay is based upon my own practical experience of empirical research, both qualitative and quantitative.(4)
I. The relationship between qualitative and other research procedures
The social scientific methods developed out of techniques of perception and acting in everyday life. In these simple techniques personal experience and traditions of former generations are consolidated into strategies for everyday use. We recognize, evaluate and change the world around us according to rules which we have learnt, experienced and modified in application.
The everyday techniques form the reservoir for all social scientific methods. The latter are generated from the former both by widening their range of application and by taking them out of their everyday context through abstraction. We experiment with subjects and objects in everyday life and we construct situations in which persons or conditions are tested. Games, too, have the attraction of comprising a trial, a form of testing. On a scientific level we conduct tests under controlled conditions and with a limited number of points of view. The scientific experiment is an abstraction from everyday experimenting. The same happens in observation. Daily we experience and know the world around us both by distancing ourselves from it and by meditating on it. When this happens under controlled conditions with concentration on a limited number of aspects, we approach the procedures of science. Another example would be interviews and questionnaires. These originate in conversation, dialogue, the normal exchange of utterances - a complex from which the unit of question-answer is abstracted.
Techniques such as experiments, observation and interviews are already rather advanced action patterns which combine various acts and relations. Whether these techniques compare, measure or whatever, they all originate from the basic techniques used spontaneously and often unwittingly in everyday life. Spontaneous activity is the origin of every modification of the environment by the human beings and it follows also of the controlled experiment or the comparison of two forms or contents of experience is the origin of every interpretation of qualitative or quantitative character, not only of the "comparative method" but also for instance of measuring.
The social scientific procedures for the perception and understanding of the world around us are not in opposition to the "natural" techniques; they are not alien to them. Scientific procedures are not deduced from perfectly logical constructions free from all contradictions. They were developed from natural ways of orientation and knowing; they operate in terms of the same "rules". All strategies of perception share the same pragmatic basis. This situation allows us to better comprehend and apply scientific techniques: their origin and coherence are to be found in everyday coping with reality. Everyday techniques are the basis from which qualitative procedures develop as a first, and quantitative procedures as a second step in the process of abstraction.
This concept has three consequences: The first concerns the unity of all method. If all procedures originate in everyday techniques, they belong together and can be seen as a unit: their precise mutual relations can be ascertained, both on the two levels (qualitative and quantitative) and with regard to the different procedures/techniques of interviews, observation and experiments. If we concentrate on one level of abstraction or one technique, its connection with the other level or technique should be kept in mind. There is no special methodology for individual procedures; all are part of a comprehensive methodology. Also, there is no special manner of proof or validation tied to a single procedure or one of the two levels of abstraction. Every procedure is part and parcel of scientific logic as a whole.
Of course, the unity of all method does not imply that all procedures are equal, can be treated as equal or are subject to the same criteria of evaluation. Unity is not to be understood as uniformity. The unity of procedures implies systematic interrelations which can be ascertained and that we can expect to find a "system" of methods.
Specification of the interrelations of research techniques will show whether the known techniques are "complete" compared with the system of all methods. While there is a correspondence of most techniques at the levels of everyday experience, of qualitative and quantitative abstractions (observation, interview, text- or content analysis, analysis of non-verbal manifestations etc.) scientific knowledge of a "qualitative experiment" is missing. I hope to be able to show what its characteristics are and how it should be used in research(5).
If everyday strategies generate social scientific techniques, it follows, as a second generalization, that the former precede the latter conceptually. Scientific techniques develop through rearrangement and abstraction from the "earlier" everyday strategies(6). And when qualitative procedures show a lower level of abstraction in comparison to quantitative procedures, the former have clearly been around much longer. This sequence is imperative. The theses of the sequence of techniques claims that from everyday procedures qualitative techniques have been or can be abstracted and from those the quantitative methods and that this sequence is coercive. It follows that it makes little sense to insist that social scientific procedures are deduced from formal logic - or qualitative methods from quantitative methods. The more abstract methods are in no way the measure of the more concrete ones - which are really more complex; but they have been abstracted from the more concrete strategies.
The above implies that scientific study should start with the application of techniques with a low level of abstraction, not with the more abstract and sophisticated ones. Qualitative research should in practice be applied "earlier" or before the phenomenon is subjected to quantitative techniques. (Of course qualitative analysis does not have to be followed by quantitative analysis. If a phenomenon is explained by qualitative analysis, quantification is of little additional help afterwards; when qualitative analysis does not explain the phenomenon, quantitative research cannot solve the problem either). Qualitative analysis can do without quantification but not the other way around. For quantitative research qualitative reflection is a necessary first step.
There is a third consequence or generalization which follows from the premise that social-scientific procedures and perception develop from everyday techniques. It concerns the contrast between the two scientific modes of research among each other and of each of them towards everyday techniques. Qualitative and quantitative procedures are not only tied to two different levels of abstraction; they also differ in the manner of abstraction from everyday techniques. In everyday life we perceive and know our environment through the integration of various sense impressions and the establishment of correspondence or similarity, and of difference or variety. Put simply, what happens in qualitative social research is that the process of abstraction produces or focuses on similarities; in the same process quantitative social research produces or focuses on differences. While concentrating an the one aspect, a mode of research also implicates the other: Qualitative research analyses the similarities of two or more phenomena by overcoming the differences between them. Quantitative research establishes distinctions by using the similarities as a basis for comparison. The objectives of the two modes of research are clearly different: discovery of relations there and measuring of different expressions of already known relations here. Their techniques and their strategies differ accordingly: research planning, definition of the topic of research, conception of the role of the researcher, sampling, instruments of investigation, procedures of analysis and proof. In "pure" form the qualitative procedure is the precise opposite of the quantitative; distinguishing themselves from one another by contradiction, but it is this contradiction which also binds together the two modes of research.
II. The objective of qualitative social research
Qualitative research abstracts and systematizes techniques of discovery from everyday procedures. Heuristics is therefore its "auxiliary" science.
Up to now heuristics has found little acceptance and application in sociology - which in part explains much of the lack of understanding of the fundamentals of qualitative research. This is in contrast with empirical psychology, especially cognitive psychology which has a heuristic tradition in dealing with problem-solving. [The name also is not used but explorative techniques are]. The importance of the Würzburg school of cognitive psychology (Oswald Külpe, Karl Marbe, Karl Bühler, Narziss Ach, Otto Selz and others) for empirical analysis of thought and action has been stressed by Paul Lazarsfeld (1972). Also Gestalt psychology is important for its attention to problems of observation and decision making (Karl Duncker, 1974; Wolfgang Köhler, 1971; Kurt Lewin, 1955). The "cognitive turn" in psychology in the sixties prepared the way for empirical research of heuristic operations(7).
Psychological heuristics is relevant for sociology because it explores problem-solving processes and in this way provides us with some of the fundamentals of sociological heuristics. All these developments, however, have not led to the establishment of definite methods of qualitative research in psychology. An examination of publications shows that, apart from same trends in psychoanalysis and psychotherapy, there is still a preoccupation with quantitative methods. An outline or design for a qualitative methodology for psychology remains to he worked out; it could follow the logic of the present outline but would naturally concentrate on the individual and his or her personal relations rather than an social phenomena(8).
The connection between qualitative social research and hermeneutics is more differentiated. Philosophical hermeneutics, though not directly involved with methods of the Humanities (Gadamer, 1960; Habermas, 1971) does contribute to qualitative methodology. Its treatment of the openness of questions or the "hermeneutical circles" is useful in qualitative social research. On the other hand, those developments which, following Schleiermacher, present hermeneutics as a theory of art, and similarly the empathy ("Einfühlung") theorists of "Verstehen", are of little help in contemporary social science. The applied variations of hermeneutics - theological, juridical, philological and historical - operate, legitimately, as a method to restore falsified texts or interpret incomprehensible documents - all which are in need of exegesis. This is not really the problem of qualitative social research whose objective is not to construe, decide on or restore original coherence, but to discover new ones. It is concerned with social reality's undiscovered relations and references not an already but possibly wrongly interpreted topic. Qualitative social research basically is not an art of interpretation but a means of discovery(9).
It is maintained that qualitative method is the real mode of discovery in empirical social research. Of course, everyday techniques and quantitative research also make discoveries - all research procedures bring new information. But qualitative research establishes knowledge more systematically than everyday techniques and discovers in more basic form than quantitative research. Qualitative techniques should allow the researcher to discover unknown and unexpected relations. Quantitative research procedure, even when not limited to measurement of predetermined dimensions, but also extended to organizing data into multivariate categories (cluster and factor analysis), can never transcend the instrumentarium set up by the researcher;
it is restricted within these limits. In this sense quantitative research "describes" rather than "discovers"; it produces data in predetermined categories and not a system of categories itself(10).
The above contradicts the occasionally expressed view that exploration through qualitative research only suggest hypotheses and that these still have to be "proved": Findings which are the result of qualitative research are proved qualitatively, not quantitatively. They can be described however by measurements though at a higher abstraction level. Of course, quantified data can also serve as a point of departure for the discovery of regularities (under certain conditions). In fact, all types of data can be a point of departure in strategies of discovery.
But what is it that qualitative research wants to discover? It discovers conjunctions, circumstances, connections, references, relations: terms which imply the concept of structure, which itself has to be specified for qualitative research.
Structures refer to what is relational. We seem to see our world as existing out of objects, people and situations among which there are connections: A father and son, master and servant are related through kinship, through authority or a person and an object through ownership which transforms a person into an owner, the object into property. The triad of at least two topics and a relation between them, which brings them into contact with each other is only apparently so.
What are seen as "objects", "people" or' situations are in reality relationships. If the kinship relation disappears, there is no longer a "father" or a "son", one doesn't exist without the other. That is to say people are the relations. If I take authority away the "master" as well as the "servant" disappear. The same happens to ownership. What are seen as "objects", "people" or situations are in reality relations. One also can say relations manifest themselves ore become tangible and concrete in persons and objects or still in another way formulated structure becomes objective. The aim of qualitative social research is to "dissolve" the seemingly hard and fast "objects" into relations.
The concept of structure, ties up with that of system - relations are limited, restricted both in time and space. The domain of range of a system cannot be defined a priori (as in "Grand Theories", C. Wright Mills, 1959) but must be determined through research. It is not ordained that social "facts" can only be understood in terms of (other) social facts, economic facts through economic or psychological facts through psychological ones. Also it is not laid down that ideology must always be traced back to economic realities. Research alone can establish the relation in the various cases.
If the aim of qualitative research is seen as systematic discovery of relations or structures, it can also be understood as social diagnostics. Diagnosis, and not the description, of symptoms is its real concern; and to achieve this, definite methods have to be developed. The diagnosis may be followed by "therapy", but this will be taking the qualitative method beyond its original objective.
III Four rules for qualitative research
In qualitative research the strategies for the discovery of relations are based on four rules. A simple sketch may help to explain them.
Most basic is the subject-object relation: the investigating researcher and the object of his research, to be found. To find the unknown the researcher applies two strategies which can be separated analytically: Action - the researcher does something; he takes a hand or intervenes in the time-space occurrence and changes it, ba acting generates reaction. He also judges or evaluates these reactions to give new direction to his actions. (Obviously a capacity to store information is also required, but that is of psychological relevance and does not concern us here.) The subject - who remains part of the socii-cultural environment - thus approaches the object through action and evaluation of information. All relevant domains are interrelated; they co-determine and are co-determined by one another in the process of action-reaction, the establishment and processing of information. Obviously the "object" searched for is not always found; if not we are at a loss as we cannot determine whether it does not "exist", or the manner of investigation is inadequate, in which case a new approach is necessary, etc.
In the following paragraphs the process of investigation will be analytically divided to allow for a better presentation of the four rules covering four instances and directions of action. These rules are abstracted from relations and activities which are part of everyday life.
1. Rule One, concerning the subject or researcher:
Conceptions or ideas about the topic to be investigated should be regarded as preliminary and overcome with new not congruent information.
All that the researcher knows and understands about the object of investigation should be considered as preliminary and should be corrected and superseded especially by incongruent information. Obviously we do not start with a tabula rasa, are never free of preconceptions and don't have to be. Preconceptions, however, must be overcome and the obvious way to do this is to accept from the start that whatever we know, think or believe about the object, is provisional and subject to new information; and that it can be altered and superseded by such information.
This is not as easy as it seems, as we usually have a psychological if not emotional investment in what we believe and take for granted, before it is modified by objective investigation - and more particularly by incongruent data. The researcher has to examine carefully differences and similarities between "old" and "new" information and above all be "open" even in his "basic" convictions. (regarding the principle of openness see Gadamer 1960, 344 f; Hoffman-Riem 1980).
To help overcome psychological resistance to new or modified views, a number of techniques can be applied. These are mostly techniques of "avoidance". It is not of much use simply to make a statement about one's point of view about the problem or object; furthermore it contributes little when only the literature supporting one's own viewpoint is read. The chances to discover something different and new would obviously be small. Glaser and Strauss (1979, 37) even advise that one should start without (theoretical) background reading! Personally, I consider it useful to acquaint oneself with the relevant literature at the outset when various positions and contradicting interpretations are known to exist which are bound to rattle the researcher! This should lead to an involvement with the problem or object of research itself rather than with literature on it. When a study of the well known literature leads one to the Socratian position of "I know that I don't know" (scio nescio) , this is still the best starting point - that of the "expert" is the worst!
This is a plea for openness with regard to the problem or phenomenon which is being investigated. It concerns not only the researcher's attitude and motivation, but also his choice of instruments of research. Questions addressed to the object of research should not be restricted in terms of categories of likely answers presupposed by the investigator and forced upon the object by him. If potential informants are only allowed to answer yes or no to questions or are restricted to a limited range of choices, the researcher cannot expect to experience anything beyond his own frame of reference. This is the rationale for the use of "open" [non-directed, unstructured] questions in qualitative social research. The person who is being questioned should be given the opportunity to respond to what he or she personally perceives to be an important aspect or emphasis - even if this has minimal reference to the original question. In many cases it is possible and useful to formulate the question in descriptive terms thus making it flexible enough for answers to be a response to a chosen part or aspect of it (11).
The focus and strategies of qualitative research are clearly contrary to the quantitative research procedure of hypothesis testing. In the latter case, the researcher's role is that of the expert who knows the literature and is certain of his own position; he has formulated precise and unambiguous hypotheses which are operationalized in clear, preferably "closed" [direct, structured] questions if not definite metric scale values. There is no uncertainty as to what is being researched. The expert tests the hypothesis, and decides on the basis of definite criteria whether it can be falsified or not falsified. If, in the process, the researcher has bypassed the qualitative phase, he cannot know the structure of the object; it is possible that in spite of great expense only the researcher's original perception and understanding of the object has been tested - and that nothing has been learnt.
This insistence on overcoming incipient perception and understanding through an attitude of "openness" toward the object is in accordance with the concept of Verstehen - as it is appropriate for qualitative research in general. When the research objective, as previously explained, is the discovery of structures, Verstehen becomes the adaptation of the researcher's epistemic structure to findings about the structure of the object. The "aha-experience" (Karl Bühler, corresponding to the classical "heureca!"), well-known in psychology, signifies the reconciliation between the initial perception of the object - changed through the process of research - and its newly discovered structure. In the process of understanding, old and new perceptions are reconciled. This change is the process of Verstehen, prejudice is turned into comprehension (12).
2. Rule Two, concerning the object of research
The definition of the object or topic is preliminary and it is only fully known after being successfully explored.
This rule is a consequence of the fact that when starting a search for a topic or an object we do not know this object or its space-time location at the outset. The object of research must be defined, but only provisionally, and the more "open" the demarcation, the better as we do not yet know the "real" object: "We want to study Y", "we assume a special characteristic with Z" etc.
The provisional and often vague determination of the object corresponds to the uncertainty which the researcher has towards his initial perception at the start of his work. This can only enhance "openness" and sensitivity on the part of the researcher - two qualities which must be maintained up to the very end of the research process when the true structure becomes clear. What changes during the investigation, is the object as seen and understood by the researcher.
It is possible that the object under investigation changes thematically - that the original theme is completely replaced by a new one. It can also be modified in range and breadth, or redefine itself as only a component of a more comprehensive constellation which proves to be the "real" theme. Also the phenomenon can be divorced from the context in which it was first located, even discovered and described for the first time. (The pioneer studies carried out by the Chicago school would be examples). As the structure of the phenomenon is discovered its boundaries become apparent, the "system" reveals itself and is comprehended vis-a-vis the original and provisional picture of the object.
While the phenomenon or object originates or is constructed during the process of qualitative research, it must have been found at the start of quantitative research, before it can be tested, measured and quantitatively described. Quantitative research, contrary to its qualitative counterpart, proceeds from a precise definition of the phenomenon.
3. Rule Three concerning action:
The object should be approached from "all" different [distinctive? or only "all"?] sides - this is the rule of structural maximization of perspectives.
One-sided perceptions and perspectives must be avoided by trying to observe the object from different angles, technically speaking from maximally varied perspectives. Investigating and trying out objects from different angles is what children do with new toys, young animals with their own bodies, we all do with new things which we inspect, etc. etc. In serious journalism one-sided reports are held in low esteem - even if satire and controversy feed upon partiality. For journalism contra-investigation and "giving the other side" are part of professional concern and equity. The same is true for law suits pro and contra have been institutionalized since Roman Civil Law. The scientific treatment of a theme must be "comprehensive"; it must state and evaluate contrary views. Diagnosis in medicine and clinical psychology always rely on various procedures. All this is in accordance with Rule Three.
The alternative to a one-sided approach is not a single contrary one, but different approaches. Just as at the outset the phenomenon is not known or only known provisionally, the possible approaches or views are also unknown. This means that this general rule has to be applied operationally to different "parts" or aspects of the phenomenon: clearly differentiated information about the object, relating to different structural aspects, must be gathered. Structurally different are different perspectives which result in different data. Operationalized Rule Three reads: When it is presumed that a factor has an influence upon results, that factor must be varied. Only the induced variations can prove the presumption to have an effect on data; without structural variations the researcher is left with known results only.
Different approaches are not chosen at random in empirical research; they are decided upon in accordance with preliminary results and tests. The choice and application of various approaches requires skill but is not an art. Qualitative research is not an art but can be learned as systematically as quantitative research - when definite rules are observed. Experience in gathering and analysis of data is, as in all learning processes, obviously useful. Both experience and dexterity in applying various perspectives can be gained from widely different areas. Glaser & Strauss (1979, 37) ?169? suggest that sociologists should give preference to belles-lettres and creative writing on the themes concerned rather than work through sociological writings only.
Applied to observation and interviews, as an example, Rule Three would imply the following:
It is taken for granted that the observer influences the results, therefore more than one observer is necessary. If there is any suggestion that sex, age, attitudes, etc. may play a role in the observation, these attributes have to be varied among observers. Observational categories may exert an influence - therefore observation modes have to be varied. If local, time or seasonal conditions are relevant, these too have to be varied. On the assumption that the researcher who is responsible for the analysis can also influence the findings, it becomes necessary to have more than one investigator or arrange for experienced researchers to discuss the project with colleagues.
Qualitative interviews, using questionnaires and "open" questions follow the same lines - being a form of observation where the interviewee as informant acts as the observer of the phenomenon being investigated. Also to be varied maximally here are: interviewer, program, time and place of interview; sex, age, position, social class and interviewee-contact with the matter concerned. Again it is advisable to have two or three researchers working on the analysis of the data, always assumed, influences on data are expected from these characteristics.
Leaving aside the qualitative analysis of documents and the qualitative experiment, something should be said about the narrative interview - which at present by some is regarded as the most successful technique in qualitative analysis. This is the interview in which a person is allowed to tell his or her "story" with a minimum of direction or suggestions form the interviewer. The prominence of the narrative interview among qualitative techniques is not altogether deserved. Its advantage lies in its openness - the interviewee is free to relate whatever information he or she deems appropriate. This is also the disadvantage: it encompasses one-sidedness and strengthens a particular view of the matter. There is the danger that the narrative results in a very subjective reproduction of an experience or situation and the researcher is left with more information about the person than about what is being researched. The person being interviewed gives a protective protocol which may be interesting and which can be analyzed qualitatively, but which tells us less about the relevant social situation than other modes of investigation would have.
The application of the rule makes requirements concerning variations, two of which should be named as particularly important: The first is the necessity of the variation of method itself. If it is accepted that methods of research have an influence on the results, then, logically, the methods have to be varied: in interviewing the experiment may be considered as an additional method; in observation text analysis may be appropriate and useful. Rule Three is the only one that rationally justifies the full use of the multitude of methods The second requirement also suggested by this rule of variation, is the involvement of historical and geographical dimensions. Where it is known or suspected that the social phenomenon being investigated has changed over time and that its social environment determines or co-determines that phenomenon, the latter has to be investigated under varying historical conditions. The same applies for influences from the particular cultural space or context.
It may seem that there is no end to the required and possible variations and that the questions being asked about the object are limitless. It should be remembered, however, that research proceeds step by step - all possible variations are not considered at the same time - the more important ones are given priority. In any case the rule does not require the involvement of all possible variations of all possible perspectives. It requires maximally structural variation of perspectives, a variation of structurally important ones.
In practice for qualitative interviews 60-80 persons are ample; 20-25 questions in each questionnaire are sufficient - in preliminary interviews even less. The qualitative sample used in observation, interviews and text analysis can either be a "total" sample (universe) or one of "extreme" groups (13).
Both these kinds of sample - total or extreme groups - differ from those used in quantitative research. The sample associated with quantitative survey research [viz. representative, random, or probability sample] is defined in advance: the universe must be known or be determined. In the case of qualitative research the universe emerges in the course of the investigation - only through the process of research do the relevant interrelations and thus the structure become explicit and definite. Qualitative research finds the method of drawing of samples by quantitative researchers unsatisfactory: it consolidates the very restricted approach and the resulting one-sided view of the phenomenon. To a qualitative researcher, the premise that random samples are representative of social structures, is unproven and prejudicial. He takes it as unproven and as a prejudice that a random sample represents the structure of a social framework. He doesn't have the opinion that human beings with lively social relations can be compared or equated with objects without contact to each other, equipped with a limited number of properties and additionally thoroughly mixed as imputed to the model of mixing red and black balls in a container which is the basis of a random sampling.
The qualitative researcher joins those anthropologists who believe that a culture can better be studied and understood through its most marked occurrences and contours - festivals and catastrophes, marginal groups, privileged and outcast, crises, wars and crime, cultural and scientific achievements - and the normality of everyday life - rather than by interviewing or observing a representative cross-section. Results of such representative surveys can, in accordance with Rule Three, also be counted as data on the matter being investigated; they should not, however, be considered the only asset or contribution to our knowledge.
The representative sample is suitable in the investigation of many phenomena including the study of human beings, when they can be reduced to certain characteristics, such as enjoying equality of choice, or the capacity to buy and consume goods, and so forth. In qualitative research, however, the representative sample is seldom useful.
4. Rule Four concerning evaluation:
Analysis of data on the basis of similarities or homologies.
Structure is reference or relation. It is operationalized as similarity between two or more given. To discover the structure of an object, the relevant data, maximally varied, must be explored for similarities. While Rule Three insists upon varying perspectives in the research design and sample choice, Rule Four concerns itself with the analysis of data - the strategy to be followed in interpretation(14).
Rule Four sometimes creates difficulties for social scientists: because of their scientific training they are inclined to accord precedence to the observation of differences rather than to the establishment of similarities. The wood is not seen for the trees. But everybody else usually sees the "wood", the whole, the greater unit, sharing the same characteristics in the same milieu. In everyday life as well, similarities and stabilities play an important role and are easily recognized and understood. It would not be possible to recognize people, things or situations, if we did not have the ability to see the similarities in our radically different and ever changing experiences and appreciate the stable and the constant.
When we attempt to establish the underlying structure in the course of the scientific analysis of data, we accept that at first glance the phenomenon seems partly homogeneous/coherent, partly heterogeneous/incoherent. We group the coherent "parts" together and try to grasp what it is which makes them so. We do the same with other data, which can show up different similarities. In this way 4-6 "batches" of data produce supplementary information - for the time being separate and not yet "integrated". It is possible that the researcher may be already able at this stage, to formulate, provisionally, the essence of similar characteristics shown up by the different batches of information. It will become clear that a number of statements can not be contracted into a single term but have to be transformed into a longer description to characterize the particularity of its homologies. The next step is to explore the "common" similarities in the various batches of data. One possibility here is to change the original clusters of information - i.e. allow data brought forward by a particular varied approach to become part of another cluster to allow for a better "fitting" of clusters. One can of course move backwards and forwards in analyzing these clusters of information. Those variations and corrections are particularly important because they may lead to the discovery of new structural relationships. In the end all of the material, inclusive of apparently deviant cases, was brought into a set of total coherence. Phrased more precisely: we found the homologies within the varied phenomena and fragments of data. Proceeding in this manner we have discovered the structure of the topic for which we were searching.
It can happen that the analysis does not "jell" - that there are data which cannot be integrated. This implies that the phenomenon can be comprehended from particular vantage points but does not "make sense" from other perspectives. On no account can we dismiss the deviant voice - even if it is only one view. Rule Three about maximum perspectives means that an "unusual" approach (and data) cannot be dismissed but must be (re)considered seriously. Perhaps the overall analysis must be modified on decisive points to accommodate "deviant" observations. A qualitative analysis, i.e. one which establishes social structure, is inconclusive and largely immaterial when it does not integrate or accommodate all the available maximally diversified data. Statistical thinking based upon frequency distributions is different; deviant data are ignored if in fact they are infrequent.
Rule Four demands the full one hundred percent: all data must have a place in the structural coherence and be comprehensible as part of the whole. The Rule is not weakened when data are incomplete or fragmentary, or when not all questions are answered, then no information may contradict the analysis.
How are likeness homology, resemblance or similarity established from data in empirical social research? Two relations are indicative of similarities: 1. direct or symbolic identity, and 2. complete disagreement, opposition or negation. In terms of formal logic, the second position is not regarded as suggesting a relation. Both, however, are well known in everyday life - the proverbial birds of a feather flock together, and also opposites attract one another. Hegel's dialectics and the "unity of contradictions" are relevant here. Apart from these expressions, many a researcher has noted ways in which antitheses go together. The work of Sigmund Freud and Georg Simmel contains good examples - the first in the field of symbolization of the unconscious, the second in the field of social conflict (Freud, 1910, 214-221; 1925, 11-30; Simmel 1904, 305-312; 1958, 186-255).
Contradictions serve as an excellent starting point for qualitative analysis. If an object is described by one person as "cold" and by another as "warm" the common or shared "dimension" clearly concerns a certain form of experience of temperature, expressed in extremes of effect. "Beautiful" and "repugnant" though contradictions establish the dimension (or similarity) of aesthetics; "active", "passive", "tired" that of the use of energy, etc. etc. That is to say that antagonisms are particularly indicative of the dimensions, from which they deduce their opposition and which nevertheless and simultaneously fuses them to unity.
Quantitative social research lays hold of differences rather than similarities, but differentiates on the basis of "belonging together ". Also in this respect it is the opposite of qualitative analysis and at the same time exceeds it. It first has to establish the respects in which objects are similar or equal and only then can proceed to measure the differences.
IV. The Dynamics of the Process of Qualitative Research
To search, to do research and to arrive at findings - these are all processes. The preceding paragraphs, demonstrating the "Rules", concentrated on the more static elements. The more dynamic aspects of the research process should now be considered; these involve the concepts of dialogue, circularity, totality and objectivity.
1. The concept of dialogue
Here also we have a basis in everyday life. The organism relates to the surrounding world through its senses and actions. During early childhood human beings experience the division of an initially unified world: the lifeworld differentiates into an I and the "Umwelt"(15) (environment). The latter shows itself to be different from the I, often opposed to it and not "behaving according to the expectations" of the I. Herein lies the (psychological) origin of man's constant endeavor to know and change the world.
This rather complex process has been reduced by behaviorism to the stimulus-response scheme. For qualitative analysis this is inadequate; the mechanistic model which is basic here, focuses on detail and cannot cope with relations, wholes and structures. As a strategy for research the stimulus-response model can only end up with trial-and-error which is inappropriate either and for the same reason.
Strategies of investigation relevant to qualitative research develop from the general modes of reaction to and understanding of the environment. They are characterized by a particular intent: something definite is searched for. The environment reacts when the intention is translated into action. This reaction is noted and valued by the subject, and influences further action. The interchange of information required, received, accepted and assimilated and the need for new information which then repeats the whole process, can be seen as a question-answer sequence.
The gathering of data is a process in which adjacent question answer-units are linked up. These units can follow one another - both in everyday and scientific situations - in one of two ways: Questions and answers can be "laced" together relatively loosely; or one can insist that the "new unit" originates from the "old unit". In the first case questions allow for answers relevant to various aspects of the object or matter being questioned. In the second case the answer provokes a further question. Cross-examination would be an extreme example of the first case; dialogue is an example of the second.
The relation between questioner and respondent differs widely in the two instances. In cross-examination the questioner is active and follows a definite plan; the respondent is passive, only reacts to questions and is the object of the examination. In a dialogue (ideally), on the other hand, both partners are on an equal footing; the roles continuously vary as an answer is simultaneously a question and a question an answer. The first form of question-answer is frequently used in quantitative research, particularly in questionnaires with a large number of questions not necessarily logically connected or naturalistic experiments dealing with participants as test-objects. The second form of question-answer is a model for qualitative research.
Therefore we imagine to start a dialogue with the object of research, in which the object does not only produce answers to questions but also communicates itself, play an active role which causes impressions or reactions with os., the subjects. Actually we also "react" to our everyday environment, we do not only act.
The topic of the research in this manner will be brought to the level of the researcher. To keep a dialogue going, answers to specific questions must be allowed to range beyond the narrow limits of the formal question; the respondent must be allowed to "digress" and raise issues or aspects which, strictly speaking, fall outside the limits of the original question; he must be allowed to comment upon what is not asked. This "freedom" is the essence of the "open" question in qualitative research. In addition: from a question must not only an answer emerge, but from the answer again a question. The researcher must determine analytically not only the manner in which a question is answered, but also which question elicited which information as answer.
The dialogue "model" works well in verbal questioning when the researcher has the respondent in front of him. Two other techniques of observation, the analysis of documents and qualitative experiments, use essentially the same logic of inquiry. They apply fundamentally the same process of progressive self-correction and increasing understanding by purposefully allowing the interplay of questions and answers. Eventual insight into the structure of the phenomenon is only gained if the process of dialogue is allowed to run its full course. It is in this process that the "epistemic structure of the researcher" is itself brought "into line" with the structure of the phenomenon being investigated. The concept of dialogue has the effect of adjustment, it serves to adapt the epidemic structure of the researcher to the structure of the object in a step by step procedure.
The concept of dialogue relates naturally to problem-oriented social research. In the latter, "answers" are conceived of repeatedly as "symptoms" of "causes" or relevant factors, which, when fully explicated, show up the constellation of the structure concerned.
In quantitative research the relation subject-object is very different from the one basic to qualitative analysis. The dialogue "model" is replaced by an insistence of precision and on the restriction of the range of questioning (and answer). Essentially quantitative research is associated with the testing of hypotheses. An hypothesis is formulated by the researcher and operationalized in such a manner that it can be tested. The equality between subject and object which obtains in the dialogue model, does not obtain here. It is imperative that the hypothesis is formulated precisely otherwise the testing and the whole process of research cannot continue. Clearly the questions put to the object are not "open". The flow of information in the two modes of thinking is different. The qualitative researcher subjects his understanding and knowledge, his conceptions and preconceptions of the phenomenon to the vicissitudes of the dialogue in an effort to overcome this preconception; the quantitative researcher puts all of his earlier accumulated information and knowledge as a whole, and in an operationalized form, to the test - expecting to have it validated or invalidated "intact".
Thinking and acting are circular when the realization of an end result presupposes this result. A strategy is circular when it returns to its point of departure. Qualitative social research is circular in both these respects.
Every process of investigation, if it is to be successful, anticipates eventual results. This applies to both qualitative and quantitative social research. In qualitative research there is a progressive and systematic restructuring of elements and part relations. The rule for the establishment of the eventual structure is the maximum structural variation of perspectives (Rule Three). Here also there is circular reasoning: the maximum structural variation of perspectives presupposes knowledge of the structure which is being investigated.
What distinguishes qualitative research from quantitative research is circular thinking and acting in the second sense - the return to the point of departure. The strategy for quantitative research is linear; it starts out from conditions, defined as precisely as possible, and proves or disproves them through investigation. Frequency distributions of previously defined characteristics are derived and the confirmation or rejection of original hypotheses ensues. The qualitative approach uncovers a model which has structural elements as its constituent parts, leading to the discovery of the whole. The return to the point of departure is not a step backward but a step forward; the starting point is reached but now with the advantage of knowing other structural elements.
Beginning and end of a qualitative analysis therefore are operationalized as follows: the starting point is as its suits the researcher - he or she starts wherever it seems convenient. It may be a field, already familiar to the researcher or which seems to be peculiar, curious or contradicting and causes attention. As the procedure is circular and as all data anyhow have to be dealt with the start is open. The end of the analysis is reached, if further variation of perspectives will not result in data different from those which are already available and all information are structurally integrated.
"Totality" refers to the relation of parts to the whole. We experience our environment by perceiving parts of that world through our senses. The "parts" we "see" are naturally limited - as are our senses. But the environment is also experienced as a "whole", a unity put together from information offered by the various senses. And the whole is more than the mere sum of the parts, but also the parts are more than mere elements or fragments; in a very real sense they are parts of the totality.
With qualitative analysis the totality is "processual" (??): the eventual structure is established through a process of integrating and reintegrating parts in a developing totality. At the end, the researcher has the "correct" components of the structure. In this respect qualitative analysis corresponds to the Marxian dialectic method, which was used in his work and particularly discussed (1974, 21). Qualitative analysis does differ from Marxist analysis, however, with regard to the problem of where to start; the point of departure being largely one of choice. Qualitative analysis does show similarities to studies in Gestalt psychology where the concept of totality (of structure) or "Ganzheit" is central. However, this similarity does not cover the more specific research techniques.
The concept of totality refers not only to the relation of parts to whole in the investigation process, but particularly includes the linkage and unity of procedure of action and valuation, of subject and object. In contrast to this, in quantitative analysis research procedures and research findings are additive: they can be united to further knowledge or precision to a chosen level or previously determined range.
Quantitative thinking in contrast to a totality based on relations uses methods and results additively: they con be accumulated and increase knowledge or precision up to the desired level or the amount previously defined.
4. The concept of objectivity
In qualitative analysis objectivity is a contentious issue and a matter that is often misunderstood. Qualitative research is often presented as being intuitive, an art, invalid because of its subjectivity, uncontrollable because it cannot be repeated, suggesting only hypotheses which must be tested quantitatively.
Subjectivity is present at the start of research. Right at the beginning the researcher "lays hold of" aspects or parts of reality. These aspects or areas are, however, freely chosen and determined by him - his choice being influenced by personal motivation, interests and knowledge and by social demands to which he responds, and other influences. The first involvement with the theme, the first "cut" from reality bringing the phenomenon into the field of vision of the researcher, is personal, subjective. This subjective position is soon challenged by the necessity to take account of new points of view, the application of "maximization" of perspectives, the consolidation of data to achieve an overall picture of the structure of the object. At the successful completion of the analysis, the structure of the object is revealed. Starting out from a subjective point of view the researcher, through exploration and analysis, eventually attains objectivity. In the case of qualitative research the concept of objectivity is an emergent one: objectivity emerges from subjectivity in the process of analysis. This is its first characteristic.
The second is that not only does objectivity ultimately emerge from an initial subjectivity, but it also detaches itself from that subjectivity. The structure of an object, known after a successful analysis, is independent of the subjective opinion of the researcher. It is not determined through a consensus omnium, not by what all or most of the experts believe. It is intersubjective and not tied to the views of particular subjects. On this independence from subject(s) rests the whole process of research and eventual discovery. In this manner objectivity however does not become disconnected from subjectivity. All social facts remain social and the method does not lend itself to transcendence.
Thirdly, the concept of objectivity is at the same time dialectical. The process leading from subjectivity to objectivity revokes the subjectivity in an Hegelian sense: subjectivity is overcome but also retained. Objective characteristics of the structure, which are present in the subjective view of the reality, are "filtered out" in the analysis. The subjective part-picture is absorbed in the objective structure. It is still recognizable for its subjectivity but is now present in the total context an a higher level. Said in other words: objectivity already has been part of a subjective view, as partial knowledge of the object but disappearing in the total though part of its constitution. It is the "objective" part of the total picture.
In the case of quantitative research objectivity is attained through the testing of hypotheses - the latter being originally subjective or originating from qualitative analysis. If one agrees with Karl Popper (1966), an hypothesis must be "formulated in such a way that it can be tested, i.e. be falsifiable. An hypothesis can only be considered true (or as an objective description of a matter, independent of the opinion of subjects) until it is refuted. The concept of objectivity which applies here, is one which suggests a provisional validity until the Opposite is proven. The qualitative concept of objectivity, in contrast, is final. It is only provisional or tentative if the structure of an object has not yet been (fully) discovered, i.e. during the course of the research process or when the analysis ends in failure.
V. Verification procedures in qualitative research
Like quantitative research, qualitative analysis also demands the verification of findings. Three criteria are of importance: reliability, validity and range of applicability.
Reliability concerns the question whether a method, procedure or technique will lead to the same results if applied repeatedly. In the case of qualitative research one has to look at the status of the process of investigation. In earlier phases one-sided views are overcome by the variation of perspectives. Renewed consideration of the data after gaining additional information leads to progressive coverage of the structure, considering the future results to a better analysis. At the start of the research process the test-retest-procedure is, therefore, not reliable. But it gains increasing objectivity. The research design is supposed to correct the errors and inadequacies of early one-sided approaches. At the completion of the research and the establishment of the structure of the object, the rule must be followed that all data which point to the same structure, are objective with regard to the object. The same applies to the evaluation by different observers. The Rule demands that every factor which may influence the outcome should be maximally varied. Directions already exist in the research process for the checking and correction of subjectively-toned, provisional analyses. The different evaluations of data by various researchers at the outset of the investigation are treated in the same way, as would separate data collected from different respondents or gathered by various observers. All this happens in the earlier stages of the investigation and is "superseded" in the eventual "collective" outcome.
Reliability in the case of qualitative analysis can thus only be verified at the close of the investigation. Then the criterion is very precise: all data brought to bear on the object or theme, which earlier did not "fit", must now confirm the structure under the 100% criterion.
Validity concerns the question whether a method, procedure or technique lays hold of what it professes to inform - whether it really leads to knowledge of the structure of the object being investigated.
Quantitative research verifies validity by taking into account internal and external data. In qualitative analysis comparison is only possible with "internal" data. The research process requires for the analysis different approaches than the obvious ones and different information than that which is immediately to hand. Again: the new positions must be maximally varied structurally. extreme cases have to be taken into account. The range of the research is thus broadened in which homologies have to be explored. Outside this range there are no corresponding data - if there are, it would mean that the data collection and the analysis have not been completed. Data can only be compared in terms of internal criteria - in qualitative research only internal validity is possible.
3. Range of applicability.
Quantitative social research gives no information on the range of application of their findings, except that they are applicable to the (statistical) universes which has been predefined. Popper's reservation that falsification always might occur is therefore applicable, even within the universe itself. Therefore a number of speculations exist which claim "universals" without being able to prove them, as Davis & Moore, 1945 or Parsons, 1970).
Empirical qualitative research determines the range or applicability of its findings through the process of investigation which demarcates the field in which the structural relations obtain. The findings are only valid within the "system" established in the investigation and not for other structures. The Rule on maximum variation of structural perspectives prescribes the space-time verification of the limits of the validity of statements. It follows that all findings of qualitative research are spatially and historically specified and demarcated; they are only applicable in particular societies or parts thereof for particular historical periods. The need for a space-time variation of perspectives therefore always exists independent of a possibilities to realize it.
Testing the range of applicability is prognosis. If the structure of a phenomenon is investigated and discovered and with it its range of applicability a prognosis about the existence of those structural characteristics is possible also in those areas which have not been investigated if within the "system"(16).
[All translations are the author's]
VIII. Literature on qualitative-heuristic methodology avaiable in English
Preliminary version of this translation - minor corrections (less than 1%) still to be made. A final version will replace this text after March 15, 2001.