Alan bryman quantity and quality in social research pdf




















Books, images, historic newspapers, maps, archives and more. First published in Routledge is an imprint of. Campbell and Fiske, However, a very real problem is that path analysis is agnostic in regard to equally plausible models. Causality Quantitative research is often highly preoccupied with establishing the causal relationships between concepts.

The literature review about previous research in relation to a particular concept or cluster of concepts, which is a standard precursor to the presentation of the results of the report of a piece of quantitative research, is often used as a substitute for a prior annd of theory. Much of the time he is concerned to extricate the causal relationships among his variables.

In general, techniques of participant observation are extremely use- ful in providing initial insights and hunches that can lead to more careful formulations of the problem and explicit hypotheses. Few quantitative researchers subscribe to a view that research xnd be value-free; therefore, replication can act as a check on any excesses. Therefore many social scientists prefer to think of participant observation as being useful at a certain stage in the research process rather than being an approach that yields a finished piece of research.

In short, such linguistic devices act as signals which forewarn the reader about the material to come. Apan participant observation is probably the method of data collec- tion with which qualitative research is most closely associated, it is slan no means the only one. A further difficulty is that even among more sophisticated treat- ments of positivism a wide range of meanings is likely to be discerned.

These activities cast me in anx types of research roles: Skip to main content. The problem, then, is to demonstrate whether the concept actually exists and qusntity classify people, organizations, or whatever, in relation to it. By contrast, the generative account of causality permits, even invites, causal explanations in terms of unobserved entities. These discussions have the advantage of bringing to the surface the differ- ences among the participants and the contradictions within and between their replies.

The fact that unstructured interviewing is often used as an adjunct to participant observation though it is frequently employed on its own as well is indicative of the tendency for participant observers to bring into play a number of data gathering methods. First, a great deal of social research is conducted in such a way that steps in the flow pro- posed by Lazarsfeld are bypassed.

The growth of interest in qualitative research is often viewed as indicative of a reaction against the application of a natural scientific model to the study of society.

Unstructured interviewing, in which the researcher provides minimal guidance and allows considerable latitude for interviewees, is also a favoured technique. This procedure entails examining what appears to be common to the items which make up each dimension. Write a customer review. Phenomenology The study of phenomenology is a vast field which can be addressed here only in a highly summarized form and with specific relevance to the topic of qualitative research.

The issue of whether positivism is an ade- quate account of the natural sciences has tended to figure much less directly in the various critiques offered by qualitative researchers. Alterna- tive interpretations have been proffered by Fiedler and others. Amazon Restaurants Fesearch delivery from local restaurants. This possibility may prompt other researchers to establish generality. Such rhetorical devices run counter to the positivist account of the modus operandi of the scientist, since they are frequently not amenable to observation.

They are then asked to evaluate their LPC in terms of a series of at least the number varies sixteen pairs of adjectival opposites, each pairing being on an eight-point scale. The bulk of the data was collected by a self-administered ques- tionnaire which was completed by the students.

His deci- sion to concentrate on fiddling was not made until a considerable pro- portion of the research had already been conducted. The authors of textbooks on social research methods give an account of the logic of quantitative research that bears a striking similarity to the positivist position e.

It also seems that there may be aspects of the general approach of quanti- tative researchers which are not directly attributable to either posi- tivism or to the practices of the natural sciences, for example, the aforementioned elevation of concepts as focal points of empirical inquiry. Theoretical consider- ations are given lip-service rather than constituting major foci in their own right in much research.

Get fast, free shipping with Amazon Prime. There are a number of points about these tenets which are worth registering. Unstructured interviewing, in which the researcher provides minimal guidance and allows considerable latitude for interviewees, is also a favoured technique. Kuhn encourages people who have no idea suantity a stone falls to the ground to talk with assurance about scientific method. The best-known of these methods is participant observation, which entails the sustained immersion of the researcher among those whom he or she seeks to study with a view to generating a rounded, in-depth account of the group, organization, or whatever.

Read more Read less. Giddens, ; Cohen, ; Bryant, Through their rejection of a scientific idiom and their recourse to the style of qualitative research they signal their adop- tion of a different framework and expect their work to be read and judged within the confines of that framework. This notion can be discerned in explications of posi- tivism in two senses. The first is the more obvious sense of need- ing to purge the scientist of values which may impair his or her objectivity and so undermine the validity of knowledge.

Clearly, within the domain of the social sciences, in which moral or politi- cal predispositions may exert a greater influence than in the natural sciences, this aspect of positivism has special relevance. Positivism denies the appropriate- ness of the sphere of the normative to its purview because norma- tive statements cannot be verified in relation to experience. While positivists recognize that they can investigate the implications of a particular normative position, they cannot verify or falsify the posi- tion itself.

In a sense, this standpoint is a special instance of the doctrine of phenomenalism, but it has been taken to have a particu- lar relevance in the context of the social sciences Keat, , though it figures in more general treatments too Kolakowski, A number of liberties have been taken in this exposition: there is no single treatment of positivism which entails all of these principles and not all positivists living or dead would subscribe to all of them.

Some points have been treated in a fairly cavalier manner in order to cut a swath through a very dense undergrowth of debate. The first two ingre- dients probably come closest to what most people mean by positivism and are also the ones which recur most strikingly in the various exposi- tions of it.

There are a number of points about these tenets which are worth registering. Principles 2 and 4 together imply a belief that there is a sharp difference between theory and observation. Such a view is suggested by T.

A major implication of his account of the history of science is that, as one paradigm is sup- planted by another, the image of the world held by ensuing scientists also changes, so that observations are interpreted within a different context.

An example which gives a flavour of this line of reasoning can be cited: During the seventeenth century, when their research was guided by one or another effluvium theory, electricians repeatedly saw chaff particles rebound from, or fall off, the electrified bodies that had attracted them… Placed before the same apparatus, a modern observer would see electrostatic repulsion rather than mechanical or gravitational rebounding.

Kuhn, , p. This view suggests a circu- lar process whereby hypotheses are deduced from general theories and submitted to empirical test; the subsequent results are then absorbed into the general theories. This portrayal often underlies the accounts by social scientists of the way in which scientists proceed see Wallace, , and below. Thirdly, the importance accorded the rule of phe- nomenalism implies that observations are the final arbiters of theoreti- cal disputes, and therefore generates a view which substantially rele- gates theoretical reasoning to a relatively minor role Alexander, This tendency is further underlined by the doctrine of opera- tionalism, which is generally associated with a positivist position and in particular can be viewed as a ramification of phenomenalism.

Sim- ply stated, operationalism seeks to remove the ambiguity in the con- cepts that are typically embedded in scientific theories by specifying the operations by which they are to be measured. Once concepts have been operationalized, we would conceive of them almost exclusively in terms of the procedures developed for their measurement.

Further, the doctrine of operationalism implies that concepts for which opera- tional definitions cannot be devised should have little or no place in the subsequent development of scientific theories in a particular field of inquiry. It is precisely this celebration of the domain of empirically observable and verifiable phenomena that has caused positivism to be the butt of much criticism.

For example, such writers often draw attention to the frequent use by scientists of analogies and metaphors to facilitate their understand- ing of the causal mechanisms which underpin the phenomena being observed. Such rhetorical devices run counter to the positivist account of the modus operandi of the scientist, since they are frequently not amenable to observation.

It seems likely that positivism is an accu- rate description of some scientific fields at certain junctures; for instance, certain aspects of physics seem to conform to the tenets of positivism, and it is no coincidence that the doctrine of operationalism was largely formulated within the context of that discipline Bridgman, If it is the case that positivism does not adequately describe the nature of the natural sciences, two related questions present themselves in the light of the chapter thus far.

Why treat positivism as the central focus of a discussion of the nature of science, and why not give much more space to apparently more accurate accounts? In fact, although the positivist account has been questioned by some philosophers of sci- ence, it is misguided to believe that there is some absolutely definitive version of the nature of science. Philosophers of science disagree widely over what science comprises. Even when they share apparently similar positions, they are not necessarily in accord over certain issues.

Further, the chief rea- son for dealing with the nature of positivism is that quantitative research has been heavily influenced by an account of scientific method which has typically been construed in positivist terms. In other words, quantitative research is conventionally believed to be positivist in conception and orientation. The authors of textbooks on social research methods give an account of the logic of quantitative research that bears a striking similarity to the positivist position e.

Goode and Hatt, ; Phillips, Further, the critics of quantitative research have invariably depicted it as inherently positivistic and have criticized its slavish endorsement of an approach which they deem inappropriate to the study of people e. Filmer et al. More recently, Guba , who writes from the viewpoint of qualitative research, has noted the arguments against viewing the sciences as positivistic. However, the key points to note are that: science has invariably been believed to operate according to the tenets of positivism; quantitative researchers have typically sought to conform to the methods and proce- dures of the natural sciences and consequently have been considerably influenced by positivism; the critics of quantitative research have viewed it as seeking to follow the precepts of the scientific method and thereby positivism.

The next step is to investigate more systematically the influence of positivism on quantitative research. Positivism and Quantitative Research Quantitative research is often conceptualized by its practitioners as having a logical structure in which theories determine the problems to which researchers address themselves in the form of hypotheses derived from general theories.

These hypotheses are invariably assumed to take the form of expectations about likely causal connec- tions between the concepts which are the constituent elements of the hypotheses. Because concepts in the social sciences are frequently believed to be abstract, there is seen to be a need to provide operational definitions whereby their degrees of variation and co-variation may be measured.

Data are collected by a social survey, experiment, or possi- bly one of the other methods mentioned above. Once the survey or experimental data have been collected, they are then analysed so that the causal connection specified by the hypothesis can be verified or rejected. The resulting findings then feed back into, and are absorbed by, the theory that set the whole process going in the first place. This account is, of course, a somewhat idealized account of the research process offered by many writers and is particularly prevalent in text- books on social research methods.

It conceives of quantitative research as a rational, linear process. Figure 2. However, although this view of the research process is commonly encountered in accounts of the logic of quantitative research, it has a number of defects. First and foremost, it almost certainly overstates the centrality of theory in much quantitative research. Of course, one needs to draw a distinction between grand theories and theories of the middle range.

Since grand theories were so abstract they offered few clues as to how they might offer guides to empirical research; by contrast, much research in sociology seemed to offer little prospect of absorption into wider theoretical schemas.

Thus one ends up with theories of juvenile delinquency, racial prejudice, bureaucracy in organizations, and so on. If it were the case that theory had the kind of priority that is implied by Figure 2.

In fact, it is difficult to sustain such a connection. For example, Platt has examined the often expressed assumption that there is an affinity between functionalism and the social survey, and has found the contention wanting. She finds that noted functionalists do not seem to have been especially predisposed to the survey technique and that, vice versa, survey researchers have not necessarily been strongly influenced by functionalism.

Platt draws these conclusions from an examination of the work of notable functionalists and survey researchers in Ameri- can sociology, as well as from interviews with some particularly influ- ential survey researchers. However, the low level of input of theory into the quantitative research enterprise is not confined, as one might expect, to grand the- ory.

Theory large or small is given lip service at best or is treated with hostility or disdain as unfounded, scientifically dangerous speculation. The role of theory is seen to follow inductively as its product or summary rather than preceding research as its subject or organiser. Warshay, , pp. Referring to the research pro- cess in psychology, Martin , p. A further problem with this idealized model derives from its appar- ent linearity and orderliness. Quantitative research is invariably much more messy.

It tends to involve false trails, blind alleys, serendipity and hunches to a much greater degree than the idealization implies. Nor does the idealized model take sufficient account of the importance of resource constraints on decisions about how research should be car- ried out. The idealized model implied by Figure 2. When researchers are asked to reflect upon the nature of their research, the image they project is of a much more untidy enterprise e.

Bell and Newby, Further evidence of the lack of a clearly ordered sequence of steps in quantitative research will emerge in the subsequent discussion. The impact of a general commitment to the scientific method, and to positivism in particular, on quantitative research has been to create a cluster of preoccupations which can be gleaned from both reports of investigations and various writings on matters of method.

The follow- ing discussion draws attention to some particularly prominent features. While inves- tigations displaying such a process exist, the problem that much quanti- tative research is relatively unconcerned with theory to which atten- tion has already been drawn implies that it is a weak account of how concepts come into being and also how they come to be subject to a measurement process.

In fact, concepts provide a central focus for much social science research but they are only loosely or tangentially related to theoretical considerations. Writing about quantitative research in American sociology, Warshay has argued that it tends to comprise the examination of concepts which are hardly at all derived from some prior theory. Thus the social world tends to be broken down into manageable packages: social class, racial prejudice, religiosity, leadership style, aggression, and so on.

The body of research relating to a particular concept, or to connections between concepts, forms the backcloth and justification for carrying out an investigation into a particular topic relating to that concept. The literature review about previous research in relation to a particular concept or cluster of concepts, which is a standard precursor to the presentation of the results of the report of a piece of quantitative research, is often used as a substitute for a prior body of theory.

Hypotheses, when constructed, are often not derived from a theory as such but from a body of literature relating to a concept. Bulmer and Burgess , p. Theoretical consider- ations are given lip-service rather than constituting major foci in their own right in much research. The emphasis on the concept as a focus for investigation is also evident in social psychology Armistead, , p.

Concepts, then, are seen as a major focus—and in many instances the point of departure—for social research. The positivist leanings of quantitative research strongly reveal themselves in the insistence, which is patently clear in the quotation at the end of the previous para- graph, that they have to be rendered observable, i.

This emphasis can be seen as the transportation into social research of the principle of phenomenalism and the doctrine of operationalism in par- ticular. In fact, the strict doctrine of operationalism—that concepts should be viewed as synonymous with the measuring devices associ- ated with them—has found few adherents.

Some writers, like Dodd and Lundberg , have endorsed such a view, but, in spite of the prominence that such authors are often accorded in philosophi- cal treatments of the social sciences e. Keat and Urry, , their influence has been fairly marginal. There is, however, a diffuse com- mitment to the operationalist position which has broad support among quantitative researchers.

This commitment takes the form of an avowed obligation to specify the meaning of particular concepts pre- cisely and to develop sound measuring procedures which will stand for them. According to many textbook accounts, as we reflect on the nature of the social world we come to recognize certain patterns of coherence.

We recognize, in particular, that there are classes of objects which seem to exhibit a commonality. To facilitate this exercise we give a name to this collectivity and we now have a concept. The problem, then, is to demonstrate whether the concept actually exists and to classify people, organizations, or whatever, in relation to it.

This last phase is often referred to as the operationalization of the concept, that is, we want to measure it. It would seem, then, that people vary markedly in relation to how they feel about their job. We come to think of these feelings as forming a collectivity and give it a name—job satisfaction.

Here, then, is a concept. But as soon as we start to ask questions about job satisfaction—why do some people exhibit greater job satisfaction than others? Concepts like alien- ation, power, bureaucratization, and so on, are all very difficult to pin down. He saw the flow as involving four stages in a sequence see Figure 2.

At the outset, as a consequence of our reflections in connection with a particular theoreti- cal domain, we develop an imagery about a particular facet of that domain.

Thirdly, it is necessary to develop indicators for each of these dimensions. This third step is the crux of the operational process: the development of a group of ques- tions which can stand for each of the delineated dimensions. Thus for each dimension is developed questionnaire items which collectively act as signposts for that dimension. Finally, Lazarsfeld proposes the formation of indices, whereby the indicators are aggregated, either to form one overall index of job satisfaction or whatever, or to form an index of each of the constituent dimensions.

Why use more than one indicator? Lazarsfeld reasoned that only a battery of questionnaire items would allow each dimension to be captured in its totality. For example, spe- cialization was taken to comprise functional specialization i. Empirical indicators were then developed for each dimension and sub- dimension.

Answers to the individual ques- tions were then aggregated to form scales relating to each dimension or sub-dimension. Thus, in order to measure functional specialization, respondents were asked whether their firm had at least one person who spent all of his or her time devoted to each of sixteen specialist areas.

Each respondent was then asked the same question in respect of other functional areas like accounts, market research, sales and service, and so on. Each of these questions can be viewed as an indicator; they are then aggregated to form an overall index or, more technically, a scale of functional spe- cialization. Similarly, in order to operationalize formalization, respon- dents were asked about such things as whether their firms used infor- mation booklets, job descriptions, written policies, and so on.

The Aston programme reflects the concerns of quantitative research in both its preoccupation with the development of measurement devices for its central concepts and also in the sense that its chief conceptual focus— organization structure—was only loosely related to a wider body of theory, namely the writings on the functioning of bureaucracies by authors like Weber The basic steps in the approach of the Aston researchers to the operationalization of organization structure are presented in Figure 2.

The approach to the measurement of concepts used in the Aston Studies, and also recommended by Lazarsfeld, is rigorous and system- atic. However, the practices associated with it are by no means as widespread as might be assumed from the frequent reference to it in the literature on social research procedure.

The departure from the Lazarsfeld approach can be discerned in two areas. First, a great deal of social research is conducted in such a way that steps in the flow pro- posed by Lazarsfeld are bypassed.

A battery of indicators of a particu- lar concept is often developed with little if any consideration of the underlying dimensions to that concept. The failure to examine the pos- sibility of there being constituent dimensions means that the battery of indicators is suggestive of only one strand of meaning that the concept reflects. Even more frequent is the use of just one or two indicators of a concept or its constituent dimensions. The point is that many researchers do not adhere to a lengthy procedure of operationalizing all of their key concepts in the manner proposed by writers like Lazarsfeld.

The widespread use of factor analysis in the social sciences exemplifies this point. Factor analysis seeks to delineate the underlying dimensions to a battery of questionnaire items. A classic use of this approach can be seen in the influential research on leadership developed by the Ohio State Leader- ship Studies. In one particularly notable study Halpin and Winer, , questions i. Such bunching can be taken to be indicative of underlying dimensions of leader behaviour.

The factor analysis revealed four dimensions, that is, groups of questionnaire items which tended to cling together. Of the four dimensions, two were particularly prominent: consideration and initiating structure. Considera- tion is denoted by responses to questionnaire items like: does personal favours for crew members; is friendly and approachable. Initiating structure is represented by questionnaire items like: assigns crew mem- bers to particular tasks; makes his attitude clear to the crew.

However, the designations, consideration and initiating structure, were arrived at after the factor analysis revealed the underlying dimensions.

The researcher has to use his or her imagination to determine what the items which make up each dimension actually mean. This procedure entails examining what appears to be common to the items which make up each dimension. Thus, whereas the flow implied by Figure 2. Whereas Fig- ures 2. They are then asked to evaluate their LPC in terms of a series of at least the number varies sixteen pairs of adjectival opposites, each pairing being on an eight-point scale.

Exam- ples are: Pleasant 8 7 6 5 4 3 2 1 Unpleasant Rejecting 1 2 3 4 5 6 7 8 Accepting Fiedler found that leaders who described their LPC in favourable terms e. Thus, the measure came first and then it was decided what the measure might be referring to.

Fur- ther, at different times Fiedler has offered no fewer than four distinct ways of envisioning the leadership styles of leaders with different LPC scores. Because concepts and their measurement are so central to quantita- tive research, there is much concern about the technical requirements of operationalization.

This concern is usually portrayed in textbooks and by writers on methods as a need to consider the validity and relia- bility of measures. The question of validity refers to the issue of how we can be sure that a measure really does reflect the concept to which it is supposed to be referring.

Textbooks invariably routinely adum- brate the procedures that are available for establishing validity. There is increasing concern among many writers that it is also neces- sary to test for the validity of a measure by using a different approach to measuring it e.

Campbell and Fiske, Frequently, examina- tions of validity point to the problems associated with simply assuming a fit between concepts and their measures. The Aston approach is not the only method used for operationalizing organization structure. However, he used a different approach to the operationalization of these dimensions in that he administered a questionnaire to samples of members of a num- ber of organizations. Thus, rather than adopting the Aston approach of using key informants who spoke for the organization, Hall drew his measures from the views of a broader constituency of people within each organization and then summed the scores.

Azumi and McMillan and Pennings have combined both procedures within a single study and found a very poor correspondence between the appar- ently kindred dimensions of the Aston and Hall approaches. It is not difficult to see why there might be a disparity between the recom- mendations of textbooks and much research practice, for validity issues can easily become fairly major projects for researchers who may see such issues as excessive distractions.

The issue of reliability is concerned with the consistency of a mea- sure. Consistency is taken to comprise two distinct questions. The first is internal consistency, which is really a matter for measures which are in the form of scales or indices, because it is concerned with the inter- nal coherence of a scale—does it comprise one unitary idea or separate components?

There is a veritable artillery of procedures and techniques which can be deployed to investigate this issue. It is probably the case that social scientists tend to be more concerned about the reliabil- ity than the validity of their measures. Textbooks tend to give the two issues equal attention, but researchers seem more inclined to report that reliability tests have been carried out. This creates the illusion that reli- ability is more important, and allows measures to be evaluated mainly in terms of this criterion.

The real reason is probably that validity test- ing is highly time consuming and can easily turn into a major project in its own right. The chief purpose in this section has been to point to the importance of concepts within the framework of quantitative research and to high- light how the preoccupation with their operationalization has led to a number of concerns such as validity and reliability.

In passing, departures from the textbook approach to the measurement of concepts have been mentioned. Causality Quantitative research is often highly preoccupied with establishing the causal relationships between concepts. This concern can be viewed as a transposing of what are deemed to be the ways of the natural sciences to the study of society. As the author of one textbook on research meth- ods has observed: One of the chief goals of the scientist, social or other, is to explain why things are the way they are.

Typically, we do that by specify- ing the causes for the way things are: some things are caused by other things. Babbie, , p. There is much dis- cussion in the literature about the proper practices to be employed in order to be able to make robust claims about cause. Such discussion tends to revolve around the two main approaches to the generation of causality—those associated with experimental and cross-sectional social survey research designs.

Central to the exercise of establishing internal validity is the ability to rule out alterna- tive explanations of a posited causal relationship. As indicated above, the presence of a control group, coupled with the use of random assignment to the experimental and control groups, means that experi- mental designs are particularly strong in this respect.

Consequently, experimental designs are invariably depicted in textbooks on research methods as particularly effective in the context of establishing defini- tive causal connections. Very often, such experimental research is depicted as a model of quantitative research precisely because of the ability of its practitioners to make strong claims about the internal validity of their findings Hughes, , p. Foremost among approaches to quantitative research which seems to be poorly equipped in this respect is the cross-sectional survey design.

In a survey, data are typically collected by postal questionnaire, interview schedule, or whatever from a sample of individuals at a sin- gle juncture. The data allow the researcher to establish whether there are associations among the various variables that are reflected in the questionnaire. The concern to establish causal connections between variables can be discerned in the widespread preoccupation among many survey researchers with the development of methods for imput- ing cause-and-effect relationships e.

Blalock, ; Davis, , in spite of the fact that survey investigations are generally thought to be primarily geared to the establishment of simple associations and corre- lations among variables. In order to be able to establish causal relationships among variables in a cross-sectional study three conditions have to be met. First, it has to be established that there is a relationship among the variables con- cerned, that is, that they are not independent of each other. Well- known statistical techniques e.

Secondly, the relationship must be non-spurious. This means that it is necessary to establish that an apparent relationship between two vari- ables, x and y, is not being produced by the presence of a third variable which is antecedent and related to x and y.

Thirdly, and perhaps most controversially, the data analyst must establish a temporal order to the assembly of variables in question. Since research designs like the cross-sectional survey entail the collec- tion of data at a single point in time, this temporal order has to be imputed.

To some extent this process implies an intuitive component in the analysis of such data. While it may be objected that this approach treats life cycle stages as unproblematic—something which many social scientists have been seeking to question Bryman et al. Researchers using an experimental design do not face this problem, since the experimental treatment is a stimulus, the response to which is deemed to be the effect.

Consequently, there is usually little doubt about questions of causal order. One of the best-known techniques used by survey researchers to unravel the relative importance of a cluster of variables as prospective causes of a dependent variable is path analysis, which is an extension of multiple regression analysis that allows the analyst to tease out the contribution of each causal factor while controlling for the others.

A hypothetical instance of such a procedure in its skeletal form is provided in figure 2. However, a very real problem is that path analysis is agnostic in regard to equally plausible models. The Aston Studies provide a case in point. Researchers in the field of organization studies have been concerned to demonstrate which factors are most instrumental in determining the structure of organizations.

While the Aston researchers tended to emphasize organizational size Pugh and Hickson, , others had found technology to be a crucial determi- nant Woodward, When Hilton re-analysed the original data collected by the Aston researchers, he found that path analysis could sustain a number of different plausible a priori models of the causal interconnections among the three variables, viz.

In recognition of such difficulties, longitudinal designs are often proposed. For example, a researcher may observe a relationship between the extent to which leaders are participative and the job satis- faction of their subordinates; but which is temporally precedent? Does participativeness enhance job satisfaction, or do leaders allow greater participativeness to more satisfied subordinates?

Questions such as this require a longitudinal approach, such as a panel design in which two or more waves of observations of the relevant variables are executed at different points in time. The main point is that there are many instances of patterns of relationships between groups of variables which derive from cross-sectional designs where temporal precedence is very diffi- cult to determine and thereby so too is causality.

Questions of causality, then, greatly preoccupy the exponents of quantitative research. Since cross-sectional research designs pose far greater problems in respect to the establishment of causality than exper- imental ones, survey researchers have sought to develop approaches to data analysis which allow them to infer causal processes.

The preoccu- pation with causality can be readily seen as a consequence of the ten- dency among quantitative researchers to seek to absorb the methods and assumptions of the natural scientist which have tended to be inter- preted in positivist terms. Bhaskar, ; Sayer, Such mechanisms may be directly or indirectly observable, or they may have to be inferred, e. This view of causality departs quite markedly from that which per- vades quantitative research in which the succession of cause and effect is so paramount.

Generalization The quantitative researcher is invariably concerned to establish that the results of a particular investigation can be generalized beyond the con- fines of the research location.

Among survey researchers this preoccu- pation manifests itself in a great deal of attention being paid to sam- pling issues and in particular the representativeness of samples. The widespread preference in textbooks and among many practitioners for random sampling is symptomatic of this concern.

Essentially, the con- cern is to establish that findings can be legitimately generalized to a wider population of which the sample is representative. Further, statis- tical inference techniques like chi-square , which are widely used by survey researchers, make sense only in the context of randomly selected samples which permit inferences to a population.

By verifying generality, the quan- titative researcher draws nearer to the law-like findings of the sciences. Perhaps for this reason, qualitative research, which is frequently based on the study of one or two single cases, is often disparaged by researchers in the quantitative tradition, for the cases may be unrepre- sentative and therefore of unknown generality. How does one know whether a slum in Boston Whyte, is representative of all slums in the USA, and, if one is unsure, how can the fruits of such research be generalized beyond the confines of Boston?

In the next chapter, the arguments against this view of the generalizability of case study research, which have been proffered by a number of writers e. Mitchell, ; Yin, , will be presented. While random sampling can establish within limits the generaliz- ability of findings to the population from which the sample was derived, there may still be problems of establishing the generality of findings to other populations.

National sample surveys are quite rare and more often than not researchers draw from particular regions or cities. These more localized populations may be selected on the basis of convenience e. An example of the latter would be the selec- tion of Luton as the test site for the embourgeoisement thesis because the investigators wanted a setting that was as favourable as possible to it Goldthorpe et al.

An investigation of this same thesis as well as other aspects of the changing class structure of the UK was carried out in two districts of Liverpool by Roberts et al. In other words, survey research findings may lack generality too, even when a random sample has been extracted.



0コメント

  • 1000 / 1000