7 Approaches to Data: Qualitative, Quantitative and Triangulation

Biswajit Ghosh and Tanima Choudhuri

1. Objective

   In this module you will learn about the quantitative and qualitative approaches to data and the utility of mixing those methods for collection and analysis of data. As triangulation of methods is a significant strategy in social science research these days, you will be introduced to different advantages and limitations of mixing methods in social science research.

2. Introduction: Approaches to Data

It is a normal practice in sociology to distinguish between quantitative and qualitative research. Such a distinction relates mostly to the use of numerical (quantitative) and non-numerical (qualitative) data (Babbie 2010). When, for instance, we say that a person is old, we make a qualitative assertion. But, if we say that s/he is 60 years old, this is a numerical and precise assertion. It is possible to quantify many aspects of social living. For instance, we can quantify age, height, weight, literacy, income or distance. Such quantification makes our observations about objects/phenomena explicit. But there are many non-numerical concepts like love, hatred, relationship, sentiment, emotion, feeling that are difficult to quantify. Even though it is often possible to numerically present a qualitative aspect of social living, say measuring maturity of a person by his age and work experience, the depth of meaning attached to the concept is often lost in the process. Obviously, as compared to quantitative data, qualitative data can be richer in meaning even though they may generate ambiguity.

The distinction between quantitative and qualitative research serves as a useful means to classify different methods of social research and arrange the range of issues concerning the practice of research. One of the reasons for doing so is the fact that these broad types represent distinctive approaches in carrying out social research. Each approach is associated with a certain cluster of methods of data collection: quantitative approach is associated with social survey technique like structured interviewing and self-administered questionnaires, experiments, structured observation, content analysis, the analysis of official statistics and the like. On the other hand, qualitative research is typically associated with participant observation, ethnography, semi- and unstructured interviewing, life history, focus group, participatory rural appraisal, narratives, and qualitative analysis of texts (Bryman 2004: 506). Some writers prefer to identify quantitative and qualitative research as distinctive epistemological positions and they often use the term ‘paradigm’ to refer to each of them. Such an evaluation has serious implication for exploring any possibility of reconciliation or mixture between the two.

It cannot be denied that the character of both quantitative and qualitative research is deeply influenced by certain epistemological and theoretical positions. As Bryman (2004) argues, the quantitative research has clearly been influenced by the natural science model of research and its positivist form in particular. As against this, qualitative research has been influenced by an epistemological position that rejects the appropriateness of a natural science approach to the study of humans. In theoretical approaches like phenomenology, ethnomethodology and symbolic interactionism, such a position is highly celebrated. Obviously, these epistemological strands have influenced the concerns of the two research approaches. Thus, the concerns for numerical data, causality, measurement, generalizability, objectivism and deductive approach etc., can be traced back to the natural science roots of quantitative research. On the other hand, due to its epistemological roots, qualitative research is mostly concerned with views of individuals being studied, meanings of human action, in-depth understanding of the reasons, opinions,and motivations, introspection into the quality of life/relations, case history, the detailed description of the context, the sensitivity to the process, inductive approach and constructionism (Bryman 1988, 2008).

While it is often useful to contrast the two research strategy, it should be recognised that the status of the distinction is ‘ambiguous’ (Bryman 2008: 21), ‘not distinct’ (Dabbs 1982), and hence one needs to be careful in stressing too much about this. Conducting research in real life is a complex and challenging endeavour and there are many instances where the boundary has been transcended. Quantitative and qualitative approaches to research are not always rooted to their original epistemological positions. It is worth noting here that research methods are much more free-floating than we suppose them. The reason for such flexibility lies in the fact that each has its own strength and weaknesses. The researchers working in the field therefore find it better to integrate them, treat them as complementary. Since exact prediction and validation by replication is difficult and often impossible in social science, use of multiple methods offers the prospect of enhanced confidence. Moreover, as Marvasti (2004) argues, methodological approaches should not be viewed as diametrically opposites as there is much overlap between the two, both in theory and in practice. There is, therefore, a distinct tradition in social science that advocates the use of multiple methods. We do not conduct research only to collect data. The purpose of research is to discover answers to questions through the application of systematic procedures which may be both quantitative and qualitative. Even feminists researchers, who once resisted the use of numerical data, have of late softened their attitude towards it. The growing popularity of mixed methods research would also mean that the age-old argument of ‘paradigm wars’ (Hammersley 1992; Oakley 1999) has lost its vitality to certain extent. There is therefore every need to understand and explore possibilities of triangulation in social research and we would do this in this module.

3. Learning Outcome

This module would acquaint you with the similarities and differences of quantitative and qualitative research. In doing so, it would explicate in detail the issue of triangulation of research including its advantages. As there are challenges of mixing methods, this module would also make you familiar with best possible way of making triangulation possible with concrete examples from the history and practice of social science research.

4. Comparing Methods

While it is a normal practice to differentiate between quantitative and qualitative research, scholars working in the field have noted down many similarities between them as well. Let us here note down these similarities and differences:

4.1. Similarities

According to Neuman (2007: 328), ‘both styles of research involve researchers inferring from the empirical details of social life’. In other words, researchers involved in both types of research regardless of their methodological and theoretical differences believe that there is reality worth of further investigation. Hence, they collect useful field data to find relation among variables/events, pass judgements on them, make reasoning to explain them and finally establish their arguments with appropriate evidences. Both forms of data analysis remain faithful to the data being collected and try to represent the social world, though in different styles. As a result, processes like use of reasonin  simplifying the complexity in the data and abstracting from the data are commonly used to generate conclusion.

Secondly, researchers following both these methods share the conviction that the scientific study of society should have a certain logic and consistency (Marvasti 2004: 8). Bryman (2008: 395) argues that both types of research are concerned with answering research questions. While doing so, they also try to ensure that they select research methods and approaches to the analysis of data that are appropriate to those questions. This means that social research of both varieties requires scientific rigor or systematic adherence to certain rules and procedures. Thus, they systematically review literature, frame question(s), gather data, record them carefully, describe and analyse them, look for general trend and make these accessible to others. It is a standard practice in social science research to document the logistics of data collection and explain these stratagems clearly before discussing the findings. All Ph.D theses in social science, therefore, contain a chapter on research methodology.

Thirdly, both are concerned with relating data analysis to the research literature. The evidences gathered from a particular field are compared with those collected by others. The similarities and differences of findings with the body of existing literature allow the researchers to develop definite patterns in the processes, causes, consequences, properties or mechanisms of any given variable(s)/ event(s)/phenomena. By doing so, both types of research try to nullify/modify/substantiate existing theories/explanations or give rise to new theory/explanation. In other words, both try to explore ‘how people (or whatever the unit of analysis) differ and to explore some of the factors connected to that variation, although, once again, the form that the variation takes differs’ (Bryman 2008: 395),

Finally, in both types of research, attempt is made to generate standardised data by avoiding errors, biases, misrepresentation, false conclusions, illusions and misleading inferences. Scholars carrying out such research always told to remain alert and sensitive to these fallacies. Hence, stress is always put on collection of valid, authentic, unbiased, transparent, and reliable data so as to explain them to make sense of the real world and happenings. However, most researchers nowadays recognise the fact that it is impossible to be an entirely objective dispassionate student of social life. There should also be occasion when researchers can be partisan (Bryman 2008: 395). Yet, this does not mean that ‘anything goes’ in research. Rather, ‘wilful bias’ is strictly avoided. The quality and importance of any research findings are also judged through public debates and discussion.

4.2. Differences

Neuman (2007: 328) has shown that qualitative data analysis differs from quantitative analysis in four ways. Thus, to begin with, quantitative researchers choose from a specialised, standardised set of data analysis techniques. Hypothesis testing and statistical methods are commonly used across different types of quantitative research. By contrast, qualitative data analysis is less standardised. There are many approaches to qualitative data analysis resulting in a wide variety of such research.

Secondly, data analysis in quantitative research does not begin until all the data are collected, classified and arranged into numbers/figures. Quantitative researchers look for patterns of relationship among variables when they are able to manipulate numbers through certain statistical tests at the end of their research. By comparison, qualitative researchers start looking for patterns or relationships among variables/events from the early days of their research. Information gathered at a stage of research guides subsequent research and often a break through at certain stage of a research becomes critical in accomplishing a ‘thick description’. Hence, unlike quantitative data, analysis of qualitative data is not a distinct final stage endeavour and it stretches across all stages.

Thirdly, the relationship between data and social theory differs in both types of research. As Neuman (2007: 329) argues, ‘Quantitative researchers manipulate numbers that represent empirical facts to test theoretical hypotheses. By contrast, qualitative researchers create new concepts and theory by blending together empirical evidences and abstract concepts’. In other words, instead of testing a proposed relationship between two or more variables (hypothesis), qualitative researchers try to illustrate an event, object or relationship in order to reveal that a theory, generalization, or interpretation is plausible.

Finally, both types of research also differ in the degree of abstraction or distance from the details of social life. Thus, in quantitative research, raw data are categorised and manipulated to reveal any pattern. When the responses from an interview schedule or questionnaire are codified and the numbers are arranged in tabular/graphical forms to carry our statistical tests/analysis, certain features of social life get revealed. Obviously, the assumption involved in such an exercise is that social life can be represented by using large numbers. Qualitative researchers, on the other hand, put stress rather on analysis of meaning of non-numerical data expressed through words, sentences, emotions, sentiments, behaviour and practices. Obviously, the sample size remains small as such researcher cannot observe or talk to large number of participants. More importantly, the data collected are relatively imprecise, diffuse, context based and carry more than one meaning.

It appears from the foregoing discussion that quantitative research involves the use of methodological techniques that attach importance to numerical categories or statistics, while qualitative research provides detailed description and analysis of the quality or the substance of human experience. The qualitative approach is more aligned with idiographic explanations, while nomothetic explanations are more easily achieved through quantification. Bryman (2008: 393) has identified the following common contrast between qualitative and quantitative research:

Quantitative	Qualitative
Numbers	Words
Points of view of researcher	Points of view of participants
Researcher distant	Researcher close
Theory testing	Theory emergent
Static	Process
Structured	Unstructured
Generalization	Contextual understanding
Hard, reliable data	Rich, deep data
Macro	Micro
Behaviour	Meaning
Artificial setting	Natural setting

These differences made Hughes (1990: 11) to argue that ‘every research tool or procedure is inextricably embedded in commitments to particular versions of the world and to knowing that world’.Smith (1983) has equally argued that each of the two research strategies supports different procedures and has different epistemological implications. He, therefore, advised researchers to avoid the assumption that methods are complementary. Smith and Heshusius (1986: 8) also went against the mixed method strategy as it ignores the assumptions underlying research methods and transforms ‘qualitative inquiry into a procedural variation of quantitative inquiry’.

The quantitative/qualitative distinction, however, has serious limitations given the fact that there is much overlap between the two in both theory and practice and both research strategy carry distinctive strengths and weaknesses. The chief difficulty with the arguments of writers like Hughes, Smith and Heshusius is that research methods do not always carry with them fixed epistemological and ontological implications. Onwuegbuzie and Leech (2005) suggest that the two approaches have more similarities than differences and hence social research is strengthened by the use of both. Notwithstanding the differences between the two methods, the relationships are not absolute. Both approaches rather present considerable “grey areas” (Babbie 2010: 25). Recognising the distinction between these approaches does not mean that one should specifically choose one by excluding the other. A complete understanding of a topic often requires both techniques. Hence, contributions of both these approaches are recognised today and they are put to a wide variety of tasks. Since early 1980s, therefore, the ‘amount of combined research has been increasing’ (Bryman 2008: 603-04).

5. Problems with Contrast

The quantitative/qualitative distinction is often ‘overblown’. As has been suggested earlier, the epistemological and ontological commitments may be associated with certain research methods, but the connections are not deterministic. For instance, researcher using the technique of interview or questionnaire might not be committed to a natural science model. Conversely, those relying on participant observation need not be necessarily interpretivists. Platt (1996: 275), therefore, has argued that “In many cases general theoretical/methodological stances are just stances: slogans, hopes, aspirations, not guidelines with clear implications that are followed in practice”. Interestingly, qualitative research often exhibits features that are associated with natural science model. For instance, many writers on qualitative research display an equal emphasis on the importance of empiricism or direct contact with social reality which is typically associated with quantitative research. The empiricism in qualitative research is most notable in conversation analysis stressing on precise transcription of talk. The idea that theory is to be ‘grounded’ in data contributes to ‘covert positivism’ in qualitative research (Bryman 2008: 589).

Again, qualitative research can be employed to investigate research questions that are associated with natural science model of research process. Thus, qualitative researchers frequently discuss hypothesis and theory testing while conducting research as in Grounded Theory (read Module RMS 26 for a detail discussion on this theory). Becker (1958) tried to show how hypothesis can be discovered and can be tested while using the method of participant observation. The classic study of Festinger, Riecken and Schachter (1956) on millenarian religious cult used participant observation to test a theory. There are many other instances of use of natural science model in conducting qualitative research (Bryman [2008: 589-591] has cited many such examples). For instance, ethnographers have often used interview to cross check their data.

The opposite experience of quantitative researchers entering into the qualitative domain of studying meanings is not rare. Thus, the study of attitudes is widespread in social surveys based on interviews and  questionnaires. Even though survey questions devised by the researcher might have limitations to understand the meaning of human actions, yet it is possible to bring out a range of attitudinal positions on an issue by devising questions based on prior knowledge. It should be noted that survey researchers frequently ask their respondents the reasons for their actions. This proves their concern to uncover issues of meaning. Interestingly, there are instances of survey based studies to explore qualitative issues like class identities (Sturridge 2007).

There are many other problems with the quantitative/qualitative distinction. As Bryman (2008: 594) argues, any hard-and-fast set of distinctions and differences carry the risk of exaggeration. Thus, for instance, the degree to which the ‘behaviour versus meaning’ contrast coincides with quantitative and qualitative research should be properly understood. Both types of researchers are interested in knowing what people do, what they think, how they act in real life even though they go about the investigation of these areas in different ways. It should also be acknowledged that qualitative researchers often ‘over-estimate the degree of rapport and shared meaning with their respondents’ (Payne and Grew 2005: 907). Hence, questions are being raised about the ‘superior capacity’ of qualitative research to uncover meaning.

Again, as against the common perception, quantitative research is far less driven by a hypothesis testing strategy. In reality, except the experimental investigations, survey based studies are often more exploratory in nature. Hence, the nature of interconnection among variables is frequently not specified in advance. For instance, a field based survey research on the nature of trade unionism among the informal sector workers in Kolkata concluded by generating hypotheses (Ghosh 1989). The findings of such studies allow researchers the scope to generate hypothesis, concepts and theories. On the other hand, qualitative research can be used to test theories. For instance, life history method entails a theory-testing approach to the collection and analysis of qualitative data (Miller 2000). Theorists of grounded theory also believed in revealing the nature and functioning of the social world (Charmaz 2006).

Finally, differences between quantitative and qualitative research with respect to lofty points like ‘numbers versus words’ and ‘artificial versus natural’ appear too rigid. It is not unusual for qualitative researchers to undertake a limited amount of quantification in their data. For instance, a qualitative study on trafficking in women and children in the tea gardens of Jalpaiguri, West Bengal, used numerical estimates of such victims provided by the stakeholders to measure the volume of trafficking (Ghosh 2014). Such quantification of findings from qualitative study becomes very helpful to uncover the generality of the phenomena being described. Similarly, the tag ‘artificial’ can be applied to qualitative studies using unstructured interview, focus group or Participatory Rural Appraisal (PRA) as techniques of data collection. Even in case of participant observation, the presence of researcher in the field situation may be a source of interference that makes the situation less natural than it is presumed.

It is, therefore, fare to argue that the contrast is not very useful and both types of research contribute positively towards better understanding and interpretation of the social world. The problem lies not with any chosen method, but the way it is utilised to collect and analyse events/facts/opinions and the like. Instead of concentrating on the quantitative/qualitative distinction, we need to focus more on the selection of particular method(s) for studying a particular problem and its proper execution.

Self Check Exercise 1

1. What are the epistemological roots of quantitative research?

The quantitative research has been influenced by the natural science model of research and its positivist form in particular. Concern for numerical data, causality, measurement, generalizability, objectivism and deductive approach etc., can be traced back to the natural science roots of quantitative research.

2. What are the epistemological roots of qualitative research?

Qualitative research has been influenced by an epistemological position that rejects the appropriateness of a natural science approach to the study of humans. In theoretical approaches like phenomenology, ethnomethodology and symbolic interactionism, such a position is highly celebrated.

3. What are the arguments against quantitative/qualitative distinction?

The quantitative/qualitative distinction is often ‘overblown’. Though the epistemological and ontological commitments may be associated with certain research methods, the connections are not deterministic. The distinction has serious limitations given the fact that there is much overlap between the two in both theory and practice and both research strategy carry distinctive strengths and weaknesses. Recognising the distinction between these approaches does not mean that one should specifically choose one by excluding the other. A complete understanding of a topic often requires both techniques.

6. Research Methods as Free Floating

Research methods are much more free-floating in terms of epistemology and ontology than is often supposed. This can be particularly demonstrated by reference to historical and other studies of social research (Bryman 2008: 593). Thus, the findings of a research by Snizek (1976) prove that Ritzer’s argument that ‘sociology is characterised by three paradigms’ is not correct. Snizek examined 1434 articles published in sociology journals in between 1950 and 1970. He did not find any explicit pattern that could link such research with any given paradigm. Equally, Platt’s (1986) research proves that the presumed association between Functionalism and survey method is exaggerated to a large extent. She has shown that the two originated independently and the leading functionalists had no propensity to use survey. Her research on trends of research in American Sociology in between 1920 and 1960 further reveals that methodological choices are steered less by fundamental theoretical assumptions and more by quite other considerations (Platt 1996: 275).

Considering the contemporary trend of research in sociology, Bryman (2008: 597-99) has correctly suggested that in order to undermine the barrier between quantitative and qualitative research, we should look for developments in which each is used as an approach to analyse the other. He has also suggested a series of practical ways to integrate quantitative and qualitative styles, characterizing Smith and Heshusius as ‘doctrinaire and restrictive’ (2004: 506-508). Let us now look into some of these possibilities in brief:

6.1. A Qualitative Research Approach to Quantitative Research

There has been a growing interest among quantitative researchers to use some of the methods associated with qualitative research. This trend can also be seen as an extension of the growth of interest of ethnographers like Van Maanen and Atkinson in quantitative research (Cited in Bryman 2008: 597). Both instances reveal a common concern for persuading the reader of the credibility of the findings. Many of those relying on survey techniques to collect large pool of data very often also use life histories of participants or their personal experiences (observation) to explain findings. One definite outcome of this concern is to employ empiricist repertoire when writing the findings. Use of ‘ethnostatistics’ (Gephart 1988), which refers to the study of the construction, interpretation and display of statistics in quantitative research, is another way. Hodson (1996) has tried to use content analysis – a quantitative research approach – to analyse qualitative findings. Such a strategy is also very useful in writing ethnography and there have been attempt to write ‘meta ethnography’ (see Bryman 2008: 579).

6.2. A Quantitative Research Approach to Qualitative Research

The presence of numbers in qualitative research is documented in some research. A certain amount of quantification becomes necessary to explore the reality. One common approach is to look for frequency of themes in transcripts or field notes. Such counting helps researchers to account for prominence given to certain incidents, words, phrases and the like that denote a theme. Moreover, qualitative researchers frequently engage themselves in ‘quasi-quantification’ (Bryman 2008: 598) through the use of the terms like ‘many’, ‘more’, ‘rare’, ‘often’, ‘some’ etc. One may, however, argue that as expressions of quantification, such ‘quasi-quantifications’ are imprecise. But, it is equally true that lack of precise estimate about prevalence of any phenomenon is a powerful criticism against qualitative research. In response to such criticism, qualitative researchers sometimes undertake a limited amount of quantification of their data. Research findings of D. Silverman, Y. Gabriel and Bryman, Stephens and A Campo reveal this trend (see Bryman 2008, for details).

An important development in recent time is the arrival of computer software for analysis of qualitative data (read Module No RMS 30, for details). Programmes like The Computer-assisted qualitative data analysis software (CAQDAS), Non-numerical Unstructured Data Indexing, Searching and Theorizing (NUD*IST), NVivo, Atlas/ti or QDA Miner now allow researchers to code text in the computer and retrieve the coded data. Even though serious questions are being raised about the consequences of use of computer software for qualitative data analysis, several writers have preferred to use the available packages on a variety of grounds. As such, computer application in qualitative research tries to fulfil the twin objectives of scientific research: one, to enhance objectivity in the formulation of research tools; two, to increase the analytical girth of the qualitative researcher without much botheration for developing a standardized platform. CAQDAS, for instance, can make the coding and retrieval process faster and more effective, and hence, to some it is a “new opportunity” as well. Bryman (2008: 567), for instance, took the assistance of NVivo as a tool in the process of qualitative data analysis in his study of visitors to Disney theme parks.

7. Triangulation of Research

Triangulation entails using more than one method or sources of data in the study of social phenomena. Originating from Greek mathematics, the concept of triangulation is applied first in navigation and military strategy to locate an object’s exact position (Smith 1975, Adami and Kiger 2005). In social sciences, the use of ‘triangulation’ can be traced back to Campbell and Fiskel (1959). It was later developed by Webb et al (1966) as an approach whereby more than one method would be employed in the development of measures, resulting in greater confidence in findings. Later, the term has been employed broadly by Denzin (1970: 310) to refer to an approach that uses four categories: i) data triangulation, ii) investigator triangulation, iii) methodology triangulation, and iv) theory triangulation. We can use the term ‘triangulation’ to refer to the research process where results of an investigation employing a method associated with one research strategy are cross-checked against the results of using a method associated with the other research strategy. It should be noted that even though the terms ‘triangulation’ and ‘mixed methods’ are often used alternately, mixing methods is only a part of the wider process of triangulation. The term ‘mixed methods research’ is used to refer to research that integrates the qualitative and quantitative strategies. Such a strategy allows researchers to make use of the strengths of different techniques while minimising their weaknesses (Bryman 2008: 603). Obviously, mixed method normally does not include investigator or theory triangulation.

According to Denzin, data triangulation is done by using multiple sources of data, but methodology triangulation is accomplished by combining two or more techniques of data collection like. For instance, in the study on women and child trafficking from the tea gardens of Jalpaiguri, West Bengal, four different techniques were used: i) In-depth Key Informants Interviews (IKII), ii) Participatory Rural Appraisal (PRA), iii) Focused Group Discussion (FGD), and iv) Case Study Method (Ghosh 2014). Such triangulation of methods was done to collect critical and sensitive data from children and other participants of research. Investigator triangulation is done by involving more than one investigators in the task in an attempt to overcome biases or personal limitations of a particular researcher. As compared to other types of triangulation, theory triangulation is a bit different exercise as here one has to look for competing theories that can support a data set.

Denzin has further divided triangulation into two types: a) between or across methods, and b) within method. The first type, which is most popular and Denzin’s favourite, uses multiple methods for cross validation of data. For instance, use of field observation to strengthen statistical data. Denzin asked us to think of two researchers studying a psychiatric hospital, one relying on survey, the other using participant observation. This would lead to differences in questions being asked and the observation made. The findings will also be influenced by researchers’ different personalities, biographies and biases. Each uncovers different aspects of social reality, but neither can reveal the complete picture. Obviously, Denzin here is suggesting us to apply the strategy of between methods to collect reliable, valid and complete information. In the second type, multiple techniques within a given method are used. For instance, a participant observer may take multiple comparison groups as units of study. In survey research, this can take the form of multiple scales, stages or indices focused on the same construct. In short, ‘between methods’ triangulation tests the degree of external validity, whereas ‘within method’ triangulation involves internal consistency or reliability.

For many researchers, however, triangulation is restricted to the use of multiple data-gathering techniques to investigate the same phenomenon (Berg 2001: 5). Fielding and Fielding (1986: 9) suggest that the important feature of triangulation is not the simple combination of different kinds of data, but the attempt to relate them so as to counteract the threats to validity identified in each. In other words, validation of research findings can be achieved through triangulation if one collects corroborating findings from the same respondents and on the same topic by using different methods.

In field research, there is a special need for multiple types of evidences gathered from different sources, often using different data collection methods like interview, observation, focus group, documents or life history. Often, the ideal method for studying a problem cannot be employed for strategic, ethical, or economic reasons. In such cases, using multiple methods is helpful to approximate the ideal method, to avoid biases, and to create realistic alternatives (Baker 1994: 285). Triangulation enhances the credibility and validity of research findings in social science. We may also call triangulation a ‘multi-strategy research’ which is more common these days (Bryman 2004: 454).

Triangulation is not however an easy task and there are challenges involved in it. Yet, an overwhelming majority of the social science research today is ‘mixed method research’ where both the qualitative and quantitative approaches complement each other during the entire research process. Qualitative and quantitative data derived from different sources are ‘illuminating’ (Bryman 2008: 603) and when properly combined, one approach should enhance the other. There are many examples of triangulation of research. Thus, Hughes and others (1997) have combined the methods of focus group (qualitative research method) and questionnaire (a quantitative research method) to collect information about consumption of drinks by young people. In this research, the two sets of results were broadly consistent and mutually reinforcing. Interestingly, since 1950, the arguments related to triangulation of research have proceeded through four stages: a formative period, a paradigm debate period, a procedural development period and finally an advocacy period (Creswell and Plano Clark 2007). In the last stage of advocacy that began in the present century triangulation of research is recognised as a distinctive approach, even as a movement.

8. Advantages of Triangulation

In defence of triangulation of research, Berg (2001: 4) wrote:

Each method reveals slightly different facets of the same reality. Every method is a different line of sight directed towards the same point, observing social and symbolic picture of reality. By combining several lines of sight, researchers obtain a better, more substantive picture of reality, a richer, more complete array of symbols and theoretical concepts; and a means of verifying many of these elements. The use of multiple lines of sight is frequently called triangulation.

Social realities are inherently complex to be grasped in its entirety with one method of investigation. Moreover, all techniques of data collection have their advantages along with disadvantages. Interviews, for instance, are very easy to organise; but there is no control over the context and the persons. Others do intervene and break the rhythm. On the other hand, the quality of data obtained from observation is better than survey techniques; but large scale study cannot be conducted by using this method. Case studies certainly add meaning and flavour to the analysis of data; but studies based only on case histories are difficult to conduct and they have their limitations in explaining the differences seen in real life.

Research methodologists also tell us to remain vigilant about certain biases that may occur in any type and stage of research. Thus, measurement bias occurs due to weaknesses of data collecting methods;sampling biases occur due to poor and faulty selection of samples; and procedural biases occur when participants hide information or provide false data. Triangulation helps to overcome these biases by generating ‘thick description’. It, therefore, serves two purposes: confirmatory and completeness. As a confirmatory approach, triangulation can overcome the challenges of single-method, single-observer and single-theory biasness and thus can be applied to confirm the research results and conclusions. For completeness purpose, researchers use triangulation to increase their knowledge base and understanding of the phenomenon under investigation. Triangulation for completeness purposes is used mainly in researching the less explored or unexplored research problems (Yeasmin and Rahman 2012: 158).

We can list the following advantages of Triangulation:

a) It allows researchers to be more confident of their result by minimising inadequacies of single source data or method.

b) It can stimulate the creation of innovative and better way of collecting of data.

c) It may help to uncover the complex character of a phenomenon.

d) It reduces the influence of different types of biases in research by the use of complementary methods.

e) It provides rich and comprehensive information by ruling out rival explanation.

f) Such triangulation helps in removing misunderstanding about processes and categories.

g) Identification of participants for qualitative research, say FGD, may become easy if a general survey is undertaken earlier. Similarly, observation of real life situation followed by a survey may enlarge the data base of researcher. Re-visiting participants is extremely helpful to notice changes in their opinion or views.

h) Both quantitative and qualitative researches depend on each other for certain set of data to explain research findings. In many circumstances, the researcher cannot rely on only one technique to explain a process.

1. Challenges of Triangulation

The triangulation strategy is not bereft of challenges and it is highly desirable that researchers try to resolve these before beginning their study. Thus, to begin with, even though it is very common these days to mix methods, such research strategy must be competently designed and conducted. Planned triangulation is better than unplanned one. Second, the research should be focused theoretically and conceptually for satisfactory result. If the research is not clearly focused, all the methods in the world will not produce a satisfactory outcome. Third, it is least helpful to legitimate a dominant and personally preferred method. In other words, any successful triangulation requires equal or significant representation of both quantitative and qualitative methods and not just window dressing. If one method is stronger and more appropriate, the fact needs to be carefully justified and made explicit (Yeasmin and Rahman 2012: 160). Fourth, replication is a bit difficult exercise in triangulation. Replicating a mixed method package, including idiosyncratic techniques, is a nearly impossible task. Qualitative methods, in particular, are difficult to replicate (Jick 1979). Fifth, triangulation may not be suitable for all research purposes as it might require better resources and time to complete the task.

The outcome of triangulation strategy might also crop up certain critical challenges. Inconsistency and non-convergence of data is a common threat to such research. The challenge becomes very crucial if the data obtained are found to be contradictory and unexpected. Jick (1979) however believed that the process of compiling research materials based on multi-methods is useful whether there is convergence or not. Where there is convergence, confidence in results grows considerable. But if divergent results emerge, researchers need to look for alternative and more complex explanations. Another approach is to treat one set of results as definite, as Newby (1977: 127) did in connection with his research on farm workers. When his survey and participant observation findings did not match, he instinctively trusted the latter as he had much confidence in such data. Bryman (2008: 611) correctly assert here that such a selection should not be done arbitrarily. The experience of Deacon, Bryman and Fenton (1998) prove that one might have to re-examine the whole set of data to arrive at a definite conclusion. In this particular research, numerical responses from mailed questionnaire revealed that journalists and social scientists broadly enjoyed consensual relationships with regard to the reporting of social science research in the media. But qualitative data collected through semi-structured interview suggested greater collision of approaches and values among them. Re-examination of the data finally revealed that social scientists were relatively pleased with the reporting of their research, but, when they reflected on specific problems in the past, their answer became negative.

It appears that triangulation demands creativity, ingenuity and insightful thinking on the part of researchers. Hence, not all writers on research methods suggest integration of different strands of research as desirable or feasible. By no means all researchers have the same skill and training to combine methods. Hence, multi-strategy research should not be considered a panacea for all.

10. How to make Triangulation Possible?

Following Denzin (1970) and Kennedy (2009), we may identify the following major guidelines for conducting triangulation in research:

i) The nature of the research problems and its relevance to a particular method should be assessed first.

ii) Different techniques that balance each other should be combined. Some of these are: quantitative vs. qualitative, individual vs. group, face-to-face vs. remote, self-reported vs. facilitated, short engagement vs. long engagement.

iii) The right tool should be selected for the right job.

iv) Involvement of at least two researchers with different perspectives is helpful to balance the findings of each other.

v) The financial and human resources should be sufficient to carry our triangulation.

vi) Methods should be combined with ‘checks and balances’ approach so that threats to internal and external validity are reduced.

vii) Conducting research in successive layers is also helpful to handle macro and micro issues one after another;

viii) The theoretical relevance of each method should be considered.

ix) Researchers should continually reflect on their methods, being ready to develop or alter them, if necessary.

x) Re-visiting participants would allow researchers to compare and contrast their views. Longer engagement with participants is helpful to notice changes in their opinions, feeling, behaviour and attitudes.

Self-check Exercise- 2

1. 1. Why are quantitative researchers interested in triangulation?

There has been a growing interest among quantitative researchers to use some of the methods associated with qualitative research. One of the reasons for this is social realities are inherently complex to be grasped in its entirety with one method of investigation. Triangulation enhances the credibility and validity of research findings in social science. It also allows researchers to be more confident of their result by minimising inadequacies and biases that are common in social science research.

2. How is between methods triangulation different from within method triangulation?

In between methods triangulation, researchers use multiple methods for cross validation of data. For instance, use of field observation to strengthen statistical data. In the within method triangulation, multiple techniques within a given method are used. For instance, a participant observer may take multiple comparison groups as units of study. In survey research, this can take the form of multiple scales, stages or indices focused on the same construct. In short, ‘between methods’ triangulation tests the degree of external validity, whereas ‘within method’ triangulation involves internal consistency or reliability.

3. What should researchers do if their mixed method results become contradictory and unexpected?

Inconsistency and non-convergence of data is a common threat to triangulation of research. Yet, the process of compiling research materials based on multi-methods is useful whether there is convergence or not. In case of divergent results, researchers need to look for alternative and more complex explanations. Another approach is to treat one set of results as definite. But such a selection should not be done arbitrarily. The best way is to re-examine the whole set of data to arrive at a definite conclusion.

11. Summary

Social research is a highly contested field, in which a variety of styles and positions co-exist often in uneasy way. Yet, it cannot be argued that there is ‘paradigm war’ between quantitative and qualitative approaches. As social research is a personal and professional responsibility, reintegration of different types of skills, data and theory ultimately strengthen our commitment to objectivity and scholarly values. It should be kept in mind that social realities are inherently complex to be grasped in its entirety with one method of investigation. Moreover, all techniques of data collection have their advantages along with disadvantages. Qualitative and quantitative data derived from different sources are, therefore, illuminating, comprehensive and when properly combined, one approach enhances the other. The triangulation strategy is, however, not bereft of challenges and it is highly desirable that researchers try to resolve these before beginning their study. Effective triangulation relies on coordination and collaboration among the stakeholders.