a Research Study
To study Hector’s situation, the researcher uses Step 1 to define a topic or problem, including the variables to be studied. Step 2 requires the researcher to review past studies on related topics to see what has been done and how variables were defined in other studies. This review provides the basis for Step 3.
In Step 3, to formulate hypotheses, the researcher must link concepts, such as poverty or dropping out of school, to specific measurements. For example, the researcher could hypothesize that poverty is a major cause of favela teenagers dropping out of school because they need to earn money for their families. Who is a dropout might be determined by school records indicating whether that child has attended school during the past 6 months. Poverty could be defined as having a low annual income—say less than half of the average income for that size of family in the country—or by assessing ownership of property such as cattle, automobiles, and indoor plumbing. It is important for researchers to be clear, precise, and consistent in how they define and measure their variables.
In order to conduct research to test a theory, researchers formulate a hypothesis, a statement they can test to determine if it is true. This is called deductive research. It starts with a theory that you then test. Inductive research, on the other hand, starts with observations that then lead to hypothesis development and, potentially, theory formation. Researchers make an observation and then begin to collect more data to determine if what they witnessed initially was a social pattern. Once they start to notice social patterns, they can begin to analyze those patterns using appropriate existing theories, or they can create a new theory if existing ones do not provide needed explanations.
Thinking Sociologically
Think of a research question based on a theoretical perspective. For example, you might ask how Hector’s peers affect his decisions, using a micro-level theory. Then write a hypothesis and identify your variables in the hypothesis.
Whether you use inductive or deductive research, you must always carefully define your variables and determine how they interact with and relate to one another. The relationship between variables is central to understanding causality. Causal reasoning and other statistical terms are discussed in the next Sociology in Our Social World.
Sociology in Our Social World
Being Clear About Causality
Sociology as a science tries to be very careful about language—more precise than we usually are in our everyday conversations. What do we really mean when we say that something causes something else? At the heart of the research process is the effort to find causal relationships (i.e., one variable causes another one to change). The following key terms are important in understanding how two variables (concepts that vary in frequency and can be measured) are related.
CORRELATION
Correlation refers to a relationship between variables (such as poverty and low levels of education), with change in one variable associated with change in another. The hypothesis earlier predicts that poverty and teenagers dropping out of school are related and vary together. That is, when the poverty level is high, dropping out of school is also high. If we claim a correlation, however, that is only the first step. We have not yet established that change in one variable causes a change in the other.
CAUSE AND EFFECT VARIABLES
Cause-and-effect relationships occur when there is a relationship between variables so that one variable stimulates a change in another. Once we have determined that there is probably a relationship, or correlation (the fact that the two variables, such as poverty and dropping out of school, both occur in the same situation), we need to take the next step: analyzing which comes first and seeing if one variable causes change in another. The independent variable is the variable in a cause-and-effect relationship that comes first in a time sequence and causes a change in another variable— the dependent variable. If we hypothesize that poverty causes Hector and others to drop out of school, poverty is the independent variable in this hypothesis and dropping out of school is the dependent variable, dependent on the level of poverty. In determining cause and effect, the independent variable must always precede the dependent variable in the time sequence if we want to try to determine whether the independent variable causes a change in the dependent variable.
SPURIOUS RELATIONSHIPS
Spurious relationships occur when there is no causal relationship between the independent and dependent variables, but they vary together, often due to a third variable affecting both of them. For example, if the quantity of ice cream consumed is highest during those weeks of the year when most drownings occur, these two events are correlated. However, eating ice cream did not cause the increase in deaths. Indeed, hot weather may have caused more people both to purchase ice cream and to go swimming, with the larger number of swimmers resulting in more drowning incidents. The connection between ice cream and drownings is a spurious relationship.
CONTROLS
Controls are steps used by researchers to eliminate all variables except those related to the hypothesis—especially those variables that might be spurious. Using controls helps ensure that the relationship is not spurious. Using the ice cream example, we might have studied beaches where lots of ice cream was sold and beaches where none was available in order to compare water death incidents. If there was no difference in death rates, the drownings could not have been caused by the ice cream.
Correlation alone can never prove causality. We need studies conducted over time, with control groups, to establish causal relationships. For example, cigarette use correlates with lung cancer. The causal relationship was only proved, however, by comparing lung cancer rates among similar groups of people whose only difference was whether they smoked.
Designing the Research Plan and Method for Collecting the Data
Researchers must always make clear how they collect their data. Every research study should be replicable—capable of being repeated—by other researchers. So, enough information must be given to ensure that another researcher could repeat the study and compare results.
The appropriate data collection method depends on the level(s) of analysis of the research question (micro, meso, or macro) the researcher is asking. For example, if you want to answer a macro-level research question, such as the effect of poverty on students dropping out of school in Brazil, you should focus on large-scale social
