Collecting and Analyzing Data

In order to test a hypothesis and determine if it is supported or refuted, researchers need to collect information. To do so, they must employ one of the research designs described later in the chap­ter. The research design guides them in collect­ing and analyzing data.

Selecting the Sample: As a means of evaluating the relationship between mental illness and homelessness, David Snow and his colleagues developed a sample from information collected by social service agencies in Austin, Texas (Snow et al., 1986). A representative sample is a selection from a larger population that is statistically found to be typical of that population. There are many kinds of samples, of which the random sample is frequently used by social scientists. For a random sample, every member of an entire population being studied has the same chance of being selected.

By using specialized sampling techniques, soci­ologists do not need to question everyone in a population. Thus, if researchers wanted to exam­ine the opinions of persons listed in a city direc­tory (a book that, unlike the telephone directory, lists all households), they might call every tenth or fiftieth or hundredth name listed. This would constitute a random sample. In the study of homelessness, the researchers drew a random sample of 800 names from the 13,881 homeless men and women who had registered at least once at the Salvation Army during a 14-month period. (The Salvation Army was chosen primarily be­cause it operated Austin's only public shelter for the homeless during this period and was the only facility in the city to provide free breakfast and dinner.) This sample of 800 persons was then compared with the records of six other state and local agencies, such as hospitals, mental health institutions, and law enforcement agencies. Ulti­mately, a usable sample of 767 persons, primarily men, was selected for the study of homelessness. All names were expunged from the researchers' files in order to protect the anonymity of the per­sons being sampled.

Creating Scales and Indices: It is relatively sim­ple to measure certain characteristics statistically, such as level of education, income, and size of a community. However, it is far more difficult to

CHAPTER TWO ♦ METHODS OF SOCIOLOGICAL RESEARCH

By using specialized sampling techniques, researchers do not need to question everyone in a population.

"All right. Now, how many would prefer Bayer?"

measure attitudes and beliefs such as patriotism, respect, and tolerance. Sociologists create scales in order to assess aspects of social behavior that require judgments or subjective evaluations. The scale and index are indicators of attitudes, behav­ior, and characteristics of people or organiza­tions.

A scale or index typically uses a series of ques­tions to measure attitudes, knowledge of facts, events, objects, or behavior. For example, sociolo­gists might want to learn not only whether re­spondents favor a constitutional amendment al­lowing a prayer in public schools but also how knowledgeable they are about different policy al­ternatives such as a "silent time" for prayer or a daily ecumenical statement read by a teacher. In this type of situation, sociologists can develop a scale to measure citizens' awareness of the school prayer debate. Throughout this textbook, we will consider how social scientists have developed scales to measure even such elusive concepts as love (see Chapter 12).

While the study of homelessness in Austin, Texas, did not employ a scale, other such studies have used scales. For example, a large-scale study of homeless people throughout Ohio developed a scale of how many symptoms of mental illness had been detected in each of 979 homeless per-

sons sampled. This scale revealed that 31 percent of the homeless had one symptom of mental ill­ness, while 13 percent displayed two or more symptoms, and less than 5 percent were regarded as "candidates for highly structured protective settings." Such data led Snow and his colleagues to question the popular view of the homeless as mentally ill (Roth et al., 1985:113; see also Ohio Department of Mental Health, 1984; Snow and Anderson, 1987).

Ensuring Validity and Reliability Scientific method requires that research results be both valid and reliable. Validity refers to the degree to which a measure or scale truly reflects the phe­nomenon under study. A valid measure of work­ers' productivity would accurately indicate how much they had produced over a specified period of time. Similarly, in the study of homelessness, researchers used definitions accepted by the American Psychiatric Association and genuinely believed to describe mental illness.

Reliability refers to the extent to which a mea­sure provides consistent results. A reliable mea­sure of workers' productivity would lead to the same results even when utilized by different re­searchers. The Austin, Texas, study provides de­tailed information concerning the research meth-

PART ONE ♦ THE SOCIOLOGICAL PERSPECTIVE

Experiment

ods used, thereby allowing other social scientists to test the conclusions in other locales.

In developing the Texas study, researchers carefully evaluated the reliability of their sample as originally developed from Salvation Army rec­ords. Drawing upon field work conducted among the homeless in Austin, Snow's colleagues found that two subgroups of homeless persons were underrepresented: homeless women (many of whom were accompanied by children, leading the agency to classify them differently) and a small proportion of "hard-core" homeless men who avoid the Salvation Army. Both these subgroups were found to constitute a very small minority of the homeless; as a result, the researchers felt confident that their sampling techniques were reliable.

Дата добавления: 2015-08-27