Research Methodology and Methods

Reflexivity in research

Reflexivity is the practice in which a researcher systematically examines how her own social position, values, assumptions and interactions shape the questions she asks, the data she gathers and the conclusions she reaches. It rests on the recognition that no researcher observes society from nowhere. The detached 'bird's eye' stance is therefore treated in reflexive methodology as something to be checked, not as an ideal the researcher can simply claim. Self-reflexivity names the careful consideration of how a researcher's past experiences, viewpoints and roles affect both her interactions in the research setting and her interpretation of what she finds. Reflexivity is closely linked to positionality, which refers to the social location of the researcher, such as gender, class, caste, nationality, education and socialization, from which she approaches the field. When a researcher reflects on her personality, temperament and life experiences during fieldwork, she is practising both positionality and reflexivity.

Reflexivity should be distinguished from objectivity. Objectivity aims to minimise bias, whereas reflexivity accepts that bias cannot be removed entirely and must instead be made visible in the write-up. Good practice includes keeping a field diary, recording how the researcher's presence may have altered informants' answers, and explaining how her biases and values bear on interpretation and analysis. Helping the community under study is an ethical or applied concern. It may accompany reflexive research, but it does not define reflexivity.

Example: a researcher who is an outsider to a village notes in her memo that informants may have softened their criticism of local leaders because she was introduced through a dominant-caste patron, and she revises her reading of the interviews accordingly.

Scientific method and research design

The scientific method is a systematic procedure of posing a question, formulating a testable explanation, collecting evidence and evaluating that explanation against the evidence. A research design is the plan that links the research questions to the evidence, specifying what will be studied, how data will be collected, and how the data will be analysed. In Indian sociology, two contrasting ways of approaching rural society illustrate how the choice of design shapes the understanding obtained. The book view derives understanding of rural India from the study of texts, particularly the classical Sanskrit sources used by Indologists. The field view derives understanding from direct empirical investigation of village life by sociologists and social anthropologists.

Kinship and marriage research also depends on precise terms, which must be used exactly in any design that classifies households or marriages:

• Hypergamy — marriage in which the bride's family is of lower status, so the groom's family is superior. Example: a woman from a lower sub-caste marrying into a higher one.
• Hypogamy — marriage in which the bride's family is of higher status than the groom's. Example: a woman from a dominant family marrying a man of lower standing.
• Exogamy — marriage outside a defined group such as a caste, sub-caste or clan.
• Polygyny — one man has two or more wives. Example: a household in which one man has two wives.
• Polyandry — one woman has two or more husbands.
• Polygynandry — several men and several women are married to one another at the same time.

A design is sound only when each term is operationalised consistently, so that different researchers classify the same case in the same way.

Experimental design and hypothesis

An experimental design tests a causal relationship by deliberately manipulating one variable and observing its effect on another. The variable that the researcher manipulates is the independent variable. The variable that is measured and expected to change in response to the manipulation is the dependent variable. In a classic design, participants are randomly assigned to an experimental group, which receives the manipulation, and a control group, which does not, so that differences in outcomes can be attributed to the manipulation rather than to pre-existing differences between the groups.

A hypothesis is the tentative, testable proposition that a researcher places at the centre of such a design. It proposes an explanation for a set of facts and gives the study something definite to verify or reject. For example, a researcher may hypothesise that a short information session on civic rights (independent variable) increases voter registration among young people (dependent variable). The experiment then compares registration rates in the two groups.

The independent variable should not be confused with the dependent variable, because reversing them changes the meaning of the causal claim. Experimental designs are also distinct from ethnography and interpretive studies, where variables are not manipulated and the meanings held by informants may be the main object of analysis.

Objectivity and replicability in social research

In social science, objectivity is often defined in terms of transpersonal replicability. A finding is objective in this sense when it does not depend on the particular researcher who obtained it and can be reproduced by other researchers under similar conditions and using similar methods. The criterion is therefore procedural and public: the procedures must be described in enough detail that another investigator can repeat them and check the outcome.

Objectivity should not be equated with the complete removal of personal feelings, though a researcher should aim to minimise personal bias. Max Weber's principle of value-freedom distinguishes empirical statements, which can be tested, from value-judgments, which concern what ought to be done. Ethical decisions about how costs and benefits are distributed among groups are value-judgments, and they are not settled by the empirical study itself. Basic honesty in applying professional norms is a matter of research ethics, which supports objectivity, but honesty alone does not make a result replicable.

Objectivity is also debated across theoretical traditions. Andre Gunder Frank criticised the dual economy model that modernisation theorists applied to Latin America, and dependency theory explains underdevelopment through unequal exchange and exploitation rather than through a backward traditional sector. Each such claim can be checked against historical and economic data, which is why the replicability criterion applies to explanatory theories as well as to individual measurements.

Classic sociological field studies

Several classic field studies are associated with particular researchers and particular settings, and examination questions often test the pairing between the researcher and the study. Laud Humphreys published Tearoom Trade in 1970, a study of homosexual encounters in public restrooms based on covert observation. The study is remembered as much for its ethical controversy as for its findings. William Foote-Whyte's Street Corner Society, published in 1943, is a participant observation study of an Italian American community in Boston, in which he spent years learning the social structure of the local street-corner groups. Elton Mayo led the Hawthorne studies at the Western Electric plant in the 1920s and 1930s, which examined worker productivity and gave rise to the notion of research reactivity discussed elsewhere in this chapter.

A.R. Radcliffe-Brown conducted his early fieldwork among the Andamanese of the Andaman Islands between 1906 and 1908, and this work informed his structural-functional approach to social organisation. The Trobriand Islands are associated with Bronisław Malinowski, and the highlands of Burma with Edmund Leach's ethnography of the Kachin. Students should therefore remember the field site together with the researcher rather than the region alone.

Key informant in fieldwork

A key informant is a person who possesses specialised or in-depth knowledge of a community or of the subject under study, and who is willing to share that knowledge with the researcher. Key informants are valued because they can explain norms, relationships and meanings that an outsider would otherwise miss. A key informant may hold a privileged social position, such as a village elder, a teacher or a local leader, and this position allows him or her to interpret the society from within.

A key informant often facilitates the fieldwork by referring the researcher to other people who have relevant knowledge, which extends the reach of the study through a chain of trusted introductions. This practice is closely related to snowball sampling, although a key informant is chosen for what he or she knows, whereas snowball sampling is a method of building a sample.

The key informant should be distinguished from a gatekeeper. A gatekeeper controls access to a site or to participants and may restrict the researcher's movement or contact with people. Such a person obstructs the study rather than helping it, and so does not fit the concept of a key informant. Because informants can shape what the researcher sees, their accounts should be cross-checked against other sources and reflexively assessed.

Hypothesis characteristics

A hypothesis is a tentative proposition that anticipates a relationship or a solution before the evidence is collected. It often begins as a shrewd hunch about a possible explanation of a problem, which the researcher then formalises so that it can be examined. Typically it proposes a relationship between two or more variables, and this relationship may be causal, in which one variable is expected to produce change in another. A hypothesis is also a provisional conclusion that is expected to be verified or rejected by empirical evidence.

The defining requirement of a hypothesis is testability. A proposition must be capable of being refuted by evidence, and a theoretically derived statement that rules out refutation cannot serve as a hypothesis, however well it fits received doctrine. Karl Popper's emphasis on falsifiability makes this requirement central to the scientific method. Testability depends on precise operational definitions of the variables, so a hypothesis stated in vague terms cannot be tested.

Example: 'Households that receive a cash transfer will send more children to school' is a hypothesis, since it relates two variables and can be refuted by data showing no difference. 'Social harmony is the natural state of society' cannot be refuted and therefore is not a hypothesis in this sense.

Positivism and scientific supremacy

Positivism is the philosophical position holding that valid knowledge is obtained through the methods of the natural sciences, and that knowledge derived from observation and verification is superior to speculation, metaphysics or theology. The claim that science is supreme over the non-sciences is attributed to positivism. Auguste Comte arranged the sciences in a hierarchy, with mathematics at the base and sociology at the top, and he argued for the unity of method, meaning that the social sciences should adopt the procedures of the natural sciences.

Positivist social research typically seeks law-like generalisations, relies on systematic observation and quantification, and aims at value-free explanation. Its critics, including interpretive sociologists, argue that human action is meaningful and must be understood from the actor's point of view. Max Weber's concept of Verstehen, or interpretive understanding, emphasised this alternative, and Neo-Kantian philosophers distinguished the cultural sciences from the natural sciences on similar grounds.

Positivism should not be confused with postcolonial critiques of Western knowledge, which question whose knowledge counts, nor with Verstehen, which defends the distinctiveness of social understanding. Positivism is the position that most directly asserts the scientific supremacy of its method.

Snowball sampling

Snowball sampling is a non-probability sampling technique in which the researcher begins with a small number of initial participants who meet the study's inclusion criteria, and then asks each of them to nominate further participants who also meet those criteria. The sample grows through these chains of referral, as each participant recommends a colleague, friend or relative. The method is especially useful when the population is hidden or hard to reach, or when no sampling frame exists, as with some stigmatised groups or small specialised occupations.

Because selection depends on referral, the sample reflects the social networks of the initial participants and is not drawn with known probabilities. The findings therefore cannot be statistically generalised to the whole population, and the researcher must acknowledge this limitation. Snowball sampling should be distinguished from random sampling and stratified random sampling, which select units by chance from a known list, and from purposive sampling, in which the researcher uses judgement about who is most informative rather than accepting nominations.