Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Sociology 3301: Sociology of Religion Lecture 3: Methodologies for Studying Religion Sociologically Religion is one of those subjects that people often know very little about, yet feel very certain that they do. Thus, it is important to approach this topic with both a healthy does of humility and a corresponding openness to new ideas. By listening to one another and paying close attention to where the evidence leads us, we can broaden our understandings. In this course, I will be presenting a sociological perspective on religion and explaining some of sociology’s findings and theories. Sociology does not offer the whole truth any more than any other discipline, but it does offer helpful insights that are useful regardless of whether one is a believer or a skeptic. Not seeking to undermine anyone’s religious faith, we are seeking useful understanding – to expand our comprehension of religion. It is but one of many vantage points – or perspectives - from which to study this topic (e.g. the psychologist studies religious experience as a mental or emotional experience of the individual; the philosopher compares, contrasts, and analyzes the beliefs of various faiths; the theologian logically and critically formulates doctrines about God and his/her relationship to humanity and the cosmos). None of these are complete in and of themselves. Still, the sociologist offers a unique perspective that differs from these others and that can contribute to a holistic analysis of this, ultimately, multidimensional phenomena we call religion. The Sociological Perspective: In sociology, we focus on religious groups and institutions (their formation, maintenance, and demise), on the behavior of individuals within those groups (e.g. social processes the affect conversion, ritual behavior, or decisions to defect to another group), and on conflicts between religious groups (such as Catholic vs. Protestant, Muslim vs. Christian, mainline denomination vs. cult). For the sociologist, beliefs are only one small part of religion. In modern society, religion is both a set of ideas (values, beliefs) and an institution (structured social relationships). Sociologists look at both to understand how they affect human behavior – not in terms of truth or falsity, but precisely in terms of how they influence people and their understandings of the world. Religious institutions can also affect behavior quite independently of beliefs, indeed, sometimes contrary to the beliefs of that religion. For example, religious institutions, at different levels, may both contribute to racism and combat it at the same time. Beliefs may state that it is wrong, while the predominant ethnic composition of a denomination leaves little room for diversity. As well, beliefs themselves can sometimes have contradictory effects. Thus, while they may teach that antipathy towards others is immoral, some 1st century Christians felt that women were incapable of being saved, reinforcing a cultural view of women as inferior beings, and between 1400-1700 between 1 500,000 and 1 million women were burned as witches (though some of this had to do with changing sex roles in society, expressed in “religious” behavior). Indeed, religious behavior can be either a cause or effect of other social processes. In short, there are many ways in which religious groups, values, and secular social processes can be interrelated. Beliefs are not always at the heart of religious behavior. Indeed, people sometimes become committed to new religious groups with little knowledge of their beliefs, but through groups pressures and social processes. Thus, to focus solely on belief misses other important matters. While many people focus on differences in belief between groups, there are many interesting and important variations in style of worship, authority structures, and psychological appeal that can have an important impact on how one views life, others, nature, and one’s own experiences. Sociology, then, focuses on the social dimensions of religion – including how religion affects society and the ways that society influences religion. We look more to the common patterns rather than the unique characteristics of each religion. Beyond specific theoretical perspectives, sociology aspires to take a scientific approach to studying religion. This is characterized by: (1) objectivity; and (2) methodological empiricism. In the first instance, sociologists try to be objective. This does not mean that we are above error or in some way can tell the whole truth. What it means is that we try to prevent personal beliefs, values, or other biases from affecting the study. We try to remain open to the data before us, to follow it where it leads us, and to avoid pre-judging any particular group – whether we agree with them or not. Our goal is to try to understand them in their own terms. This type of objectivity is not always easy (e.g. both skeptic and believer sociologists may have their own biases, be more or less favorable to religion, and personal interest in the subject matter regardless of belief may contain its own influences). To maintain professional objectivity when studying religion, one must do more than simply declare one’s religious background (if any), but engage in ongoing soul-searching into previously unrecognized feelings and biases in a constant effort to understand each group on its own terms (e.g. like feminists who study gender inequality or Marxists who study class inequality). Such personal connections are not problematic in themselves. They may influence the topics sociologists choose to study and the questions asked, but they should not affect the data collection, analysis, or the conclusions that are drawn. While this is something of an ideal in the face of human imperfection, it remains a standard to live by, a pursuit that is worthwhile. The second fundamental principle of the social scientific perspective is methodological empiricism. We can only make claims about the social world based on systematically gathered data. The social scientist considers only evidence that are observable through the five senses (i.e. we would not consider supernatural claims “really real,” though as a belief it may be real in its influence on beliefs and behaviors). Sociologists deal only in facts that can be measured, observed, and tested. They are not satisfied with general impressions, but seek concrete, verifiable data to verify or disprove any generalization. Of course, we may never achieve the precision of measurement or 2 explanatory power of the natural sciences – given the subjectivity and agency of our subject matter. Nevertheless, we should do our best to provide evidence-based analyses. Sociological Methods of Studying Religion: Sociologists attempt to use systematic and rigorous methods of data collection and analysis when studying religion. Unlike journalists who accesses whatever sources will provide a story, partisans who talk only with those they agree with, or novelists who create their own stories, sociologists restrict their claims about social life to those for which they have good empirical evidence. Thus we turn to the various methods by which social scientists gather their data. Survey Research and Statistical Analysis: One, probably dominant mode of sociological research throughout the 20th century has generally involved statistical analysis of data from closed-ended surveys of individuals. Such quantitative research involves a survey instrument (what questions are asked) and a survey sample (who is asked the questions). The questions are generally fixed (i.e. have pre-specified answer categories, such as in an ordinal ranking of how often one prays). Responses to such questions can easily be converted to numbers amenable to statistical analysis (e.g. the % of people who pray weekly or more). Alternatively, if we have answers to a second question, such as frequency of attending religious services, we can cross-tabulate and find out the percentage of people who both pray weekly or more and attend religious services at least weekly. The most common way to analyzing survey data in sociology is called multivariate inferential statistics. Here, the researcher tries to infer the causal effect of one or more independent variables on some dependent variable. The former are sometimes called causal variables or explanatory variables as they are believed to cause or explain something else that occurs as a result – which we call a dependent variable. Such studies also introduce “control variables,” factors that the investigator is not directly interested in but which could influence the dependent variable and cause a mistaken assumption about the causal relationship in question. Thus, these have to be held constant to properly assess the real influence of the main variables being studied. Creating the questions that will go on a closed-ended survey must be done very carefully as, once administered to respondents, it cannot be changed. Questions should get at the underlying concept in a way that makes sense to both the researcher and those being surveyed (a process called operationalization). When we operationalize a concept, we translate abstractions like “religiosity” into specific questions to be answered (e.g. how often do you pray?) Because most survey research seeks to generalize from those surveyed to a broader population, who is given the survey is very important (i.e. sampling procedure). The ideal for making such generalizations is a simple random sample of the population, one where every member of the population has an equal chance of being selected. When 3 done, researchers know within a certain margin of error that the conclusions drawn from the sample represent the entire population (e.g. a random sample of 3000 Americans can be generalized to the American population within a certain margin of error). Because such studies are costly, however, most surveys involve a smaller population base (e.g. college students, residents of a particular city, or members of a specific denomination). Beyond individuals, it is also possible to survey organizations (e.g. asking a representative sample to name their denomination, if any, to generate a random sample of congregations in the country). Yet, statistical analysis of survey data is not without its problems. It may generate a lot of descriptive information about which characteristics can be found in a population, but does not necessarily tell us which factor causes which. (e.g. does lower education cause fundamentalism, or vice versa?) Moreover, cross sectional surveys, done at a single point in time, cannot answer this question, so it may be necessary to conduct longitudinal surveys, done at separate points in time, to help determine which variable comes first. (these are costly). A second problem is that the data produced by large scale surveys involves correlations of hundreds of answers to specific questions, which means that the interpretations of the meaning of the responses for the respondents themselves is sometimes lost. Sometimes it is hard for someone to fit their experience into the limited answer categories provided. Even a simple question like “how often do you pray?” does not demonstrate the variations in an individual’s prayer life over time, even from day to day or week to week. Another problem is that sociologists sometimes assume that negative responses to certain questions mean that the person is “less religious” than others. Yet closed-ended questions often don’t allow people to express alternative modes of religiosity, nor that what is a core tenet for one sect of a religious community may be insignificant or ridiculous to another. The presuppositions of the researcher may serve to filter things out that affect the objectivity of the questions and accuracy in interpreting the answers. A final problem with quantitative surveys is that sometimes what people say is quite different from what they do. For example, over a period of time surveys may show 40% reporting attending religious services weekly, but when sociologists actually go around and count attendance on any given Sunday, they find a much smaller percentage of people actually in church. It may be that survey respondents are simply giving socially preferred responses, answering what is expected of “good people.” Such a problem is compounded given that some groups may be more eager to please than others, potentially making it incorrectly appear that there is a correlation between two variables when there really is not. More generally, survey information does not involve a direct study of religious experience itself, but focuses on reports of religious experience or its consequences. Yet, if people often operate on assumptions and respond to symbols of which they are only 4 partially conscious, we can conclude that the value of survey information is affected by how “hard” (verifiable, objective, and unchangeable) such data are. Statements of one’s religious preference or political party affiliation are relatively hard data. On the other hand, to get at the importance of religion to respondents, it may be necessary to ask several questions (e.g. how significant is religion to you? How important is it in everyday life?) Even then, it may be hard to pin this down, and any correlations must be interpreted with great care lest bias creep in. Interviewing: Interviewing has some similarities with survey research, but these more often employ semi-structured or open-ended questions that more readily speak to issues of meaning and process (e.g. an interviewer can more readily probe into the meaning of an answer if respondent is vague or unsure). As with surveys, interviews begin with questions, but they do not prescribe a set of responses. Good interview questions get respondents to begin talking about the topic of interest and then specify certain follow-up questions (called probes) that an interviewer should use to encourage the person to elaborate. An “interview schedule” with very few general questions and few probes is called open ended; one with more specific questions and more numerous probes is called semi-structured. For an example that falls somewhere in the middle, Wuthnow (1998) asks generally about someone’s involvement in religious organizations since childhood, asking various clarifying questions along the way (one of 32 questions in his 13 page interview schedule). Usually interviews are recorded and later transcribed for analysis. These can be quite lengthy and make data analysis challenging, to say the least. In the analysis, researchers seek to identify patterns in the data, often in the types of language people use to describe their experiences and understandings – language not unique to each person but drawn from larger social patterns, cultural frames or toolkits. The point is not what is going on in a person’s head, but what is going on in lots of people’s heads. For example, Wuthnow (1998) identified in his interviews a new “spirituality of seeking” in contemporary society: an emerging way of thinking about being spiritual that is flexible and less dogmatic. Such research is often called qualitative because the focus is typically nonnumeric. The richness of such data is often a strength, but the downside is the considerable investment of time involved in interviewing, transcription and analysis. This also means that samples are necessarily small, often not random, and thus hard to generalize to the broader population. Another problem is that there is a doubleinterpretation standing between the social reality being investigated and the findings of the study. First, the individual is interpreting their experience to the researcher, then the researcher has to interpret these in terms of establishing a broader pattern. Thus, interviews may speak more to process than surveys, but this is still not the same as directly observing social processes. 5 Participant Observation: Participant observation – or ethnography – involves immersing oneself in the situation under study. This can be for various lengths of time and involve engagements ranging from periodic visits to actually living among the group in question. By participating in the group researchers can observe the religious beliefs and behaviors of people in a concrete social context. This can be done overtly (with subjects aware of the study) or covertly (with subjects believing the observer is a new member). This latter type has been increasingly viewed as unethical and today has been frowned upon by university ethics boards. There are 3 related advantages of this method. First, the data produced are very “thick” – unlike surveys which tell us a little about a lot of people, they tell you a great deal about a few. Second, the data are social rather than individual: the method does not generalize about groups, organizations, or social systems from what individuals say, but from observation of the groups and organizations in the first place. Third, participant observation does not rely on people’s self-reporting, but on direct observation. This allows sociologists to see social processes in action, to separate what people say and what they do (e.g. rather than simply ask people what they think or believe about religious racial segregation, some like Marti (2009) spent considerable time in multi-racial congregations observing how people do religious racial integration in stages – something that probably wouldn’t have come to light through interviews alone. The downside of this method relates to the researcher’s observations. First, the data are normally limited to one case (e,g, a congregation), limiting the ability of the researcher to make generalizations (i.e. breadth is sacrificed for depth). A large number of case studies is needed before generalized patterns can be identified, and then they are often catalogued by persons who did not have firsthand experience in each group. Second, reported data are bound to be somewhat biased by the observer’s own “filtering system” (e.g. interests, system of relevancies). Data can be unconsciously filtered, and even if this were not the case, what the observer chooses to write up cannot possibly include all of the observations made. There must also be conscious filtering for relevance. This, and the fact that these studies are usually undertaken by only one researcher, provides the danger of subjective bias. Having a “research guide” (e.g. one that says what to do, when, and where) may help, but these are still dangers. Despite shortcomings, this method has proved very valuable in the sociological study of religion. Content Analysis: Still another method of research is content analysis: analyzing the content of cultural “texts” (broadly understood to include written documents, spoken words, or various other media. In this method, the researcher tries to ferret out underlying religious themes or unarticulated assumptions by analyzing written materials (e.g. the relative 6 importance of religion and science can be examined by looking at the content of TV news and current affairs programming). Content analysis has been useful, but a difficulty is its assumption that the texts being analyzed accurately present the views of the people. Sermons, for example, may not reflect the attitudes or values of the congregation, TV may not appeal to the actual values and interests of the viewers. Researchers using content analysis have to be cautious in generalizing from the texts under consideration to wider views in the population. Historical-Comparative Analysis: Historical-comparative analysis is a method that helps us understand that events in the past shaped the present situation in which we are living and why things turned out differently here than in other places. For example, Max Weber’s famous study of the rise of capitalism in Western Europe sought to understand the connection between religious beliefs and economic practices in parts of the world dominated by different religious traditions. Examining Confucian, Hindu, Buddhist, Christian, Islamic, and Jewish religious ethics, he found that only Protestant Christianity provided the motivation necessary to facilitate the rise of modern industrial capitalism. In this, he explained both the seeds of the world we live in today and why these seeds flowered first in the West instead of elsewhere. When sociologists use historical material, they do so differently than historians. Rather than engage in detailed description of historical situations, their causes and consequences, sociologists are more likely to be interested in whether a particular social situation is usually accompanied by – or followed by – some other “typical” situation or circumstances. In other words, sociologists are looking for patterns, general rules in the relationship between social events and religious characteristics. The goal is to develop a generalization or a theory that explain the relationship – both in the particular situation and more broadly. This has led to things like typologies of religious groups and theories about the evolution and decline of religion in modern society – among other things. The danger with this method is a tendency to impose one’s own pattern on the data and thus distort history. The goal of this method is to recognize and uncover historical patterns that are relevant today, but an unbiased execution of this method can be hard to accomplish. Experimentation: The most powerful tool of the social sciences in terms of ability to control variables and isolate causal influences is experimentation. In a pure experimental model, subjects are randomly assigned to “treatment” and “control” groups, where the only difference is the exposure of the treatment group to the causal force in question. 7 Often used in medical research, and possessing great explanatory power, experimentation is the method least often used by sociologists. Why? Well, for example, we could not experiment on students with several techniques for converting people to a new religion: this would be a gross violation of social norms and research ethics, especially given the deeply personal nature of religion. On the other hand, informing the subjects up front would bias the outcome. Thus, true experimentation on religious behavior has been very limited. Instead, sociologists of religion have tended to employ various types of “quasiexperimental” designs. Such studies employ the logic of treatment and control groups without the actual random assignment of individuals. Thus, Darley and Batson (1973) investigated seminary students by trying to measure effects of biblical stories on subsequent behavior. Students had been asked to speak on a particular topic (some on the Good Samaritan), and an apparently ill, sinister-looking man was placed en-route to the location where they were to speak. It was found that those assigned this story were not, in fact, more likely to help, though other factors came into play. Other “quasi-experimental designs allow a natural event to serve as the manipulator of variables (e.g. do measurements of psychological variables and values before or after a revival meeting, a natural disaster, etc). While not conforming to the normal standards of experiment, such studies can still shed light on how events can affect faith. Nevertheless, most research on religion must proceed without the powerful tool of experimentation due to the ethical concerns noted above. Triangulation: Before closing this survey of research methods in the study of religion, one must speak of triangulation of methods. To some extent, the weaknesses of each research design can be compensated for and overcome. No one design is entirely adequate in itself. The variety of approaches available allows social scientists to check the accuracy of their theories from a variety of data sources. While often questioning which design is most appropriate, and most researchers tend to prefer one approach over others, sociologists’ ability to gather data in several ways provides checks on the weaknesses inherent in any one approach – a process called triangulation of the data. Together, using various approaches when approaching a problem enables us to substantiate or dismiss various generalizations about religious behavior that would simply be impossible using one alone. Looking at an issue from different vantage points and in different ways makes for much stronger conclusions than using one method alone. 8