Fundamentals of Social Research
Adam J. McKee, Ph.D.
This content is released as a draft version for comment by the scholarly community. Please do not distribute.
One of the hallmarks of the scientific method is precision. This call for precision reaches into every act of the scientific endeavor, including the meaning of words. To be of value, scientific theories and research findings must be communicated to the community of scholars. The standard is that these professional communications must be unambiguous. That is, everyone must understand exactly what authors mean when they use particular words. The basic building blocks of social scientific theories are concepts and constructs.
A concept can be defined as a verbal abstraction of a number of particular cases. That is, a concept can be viewed as a label social scientists place on observed phenomena. In the social sciences, many of these concepts will deal with things that vary. For example, a researcher may note that some people are short, some are average, and some are tall. From these observations, the concept of height develops. This may all seem very obvious, but it is important to understand that social scientific concepts are always observable.
Social scientists use the term construct to describe an idea similar to concept, but constructs are more abstract. Constructs are said to be abstract because there is no direct link between the observations and the construct. Constructs, then, are not built from direct observations. They are built, rather, from the logical assembly of a number of concepts that are observable. To complicate matters, some constructs are more abstract than others. We can think of very abstract constructs being interpreted in terms of a hierarchy. In other words, a very abstract construct can be defined by less abstract constructs, and these second level constructs can be defined in terms of observable variables. Viewed this way, it is apparent that the degree of abstraction of a construct can be thought of as its distance from observable variables.
A well-developed scientific concept will have three major parts. The first is a label. The label of a concept is the word the researcher chooses to represent the concept. It is important to note that social scientists often choose words because they have a very special meaning in the particular research context. Just as in law, researchers have “terms of art” that reflect a different depth of meaning that the everyday use of the word.
A second major feature of a well-developed scientific concept is the theoretical definition. Theoretical definitions express the meaning behind a particular concept label. The term conceptual definition is also used to describe this type of definition. In science, meaning must be made explicit and unambiguous. Everyday definitions are plagued by vagueness, and word choice is often sloppy. Take the concept of poverty for example. If 30 research methods students were asked to take out a piece of paper and define poverty in terms of annual household income, the likelihood that any two students would come up with the same number is very low.
Everyone knows what poverty is, but the concept may take on slight variations in the minds of every person. Social scientists find this sort of ambiguity unacceptable. They are prone to define their terms with a great degree of specificity. Because of this standard, a hallmark of a quality research study is an explicit definition of all concepts of importance to the study. Note that concepts can be described in a single word (e.g., crime), or they can be labeled using a phrase (e.g., socioeconomic status).
An operational definition defines a concept in terms of how it will be measured in a study. This is an important step in the social science research process. The rules of science require that the concept be subject to objective observation by anyone wishing to repeat the findings of a study. This means that researchers must specify exactly how a particular concept will be measured in practice. If different types of measurements are required to adequately measure a concept, then the researcher must specify exactly how the different measurements will be combined to form the number that will represent the concept. Viewed from this perspective, operationalization is the process of specifying how a theoretical definition can be translated into a measurement plan that will result in data. The researcher should specify in logical or mathematical terms how the single value that represents the construct will be constructed.
A critical aspect of specifying an operational definition is determining the unit of measurement. In the United States, the speed limit is specified in a unit of measurement called miles per hour. For the American tourist trying to drive in Europe, switching to Kilometers Per Hour can be confusing until you grow accustomed to it. The same confusion can arise of the authors and consumers of research do not pay close attention to the unit of measurement used in the study. Imagine the different results that would arise if one research specified the variable “sentence length” in years, and another researcher specified “sentence length” in months. As tired as the cliché may be, consumers of social scientific research must take care not to “compare apples to oranges.” Similar warnings are given about the level of measurement that the researcher chooses in a different section of this book.
One of the most basic tasks of social research is to model reality using numerical representations of social phenomena. Social scientists usually refer to these numbers simply as data. When a concept can be measured directly, the major issues are the reliability and validity of the data. In other words, the researcher is interested in whether measurements precisely measure what they are supposed to measure. Some concepts of importance to social researchers cannot be measured directly. Variables in a research study that cannot be measured directly are often referred to as latent constructs.
Social scientists often model reality using latent constructs such as socioeconomic status, anger, and intelligence. Unlike measures such as weight and volume, there is no device with which to measure anger. Because they are not directly observable, latent constructs are essentially theoretical in nature. Since they are not directly observable, they are not directly measurable. Consider intelligence. When observers determines that a person is intelligent, they are not directly observing intelligence. They are observing behavioral indicators believed to indicate the construct of intelligence. Because those behaviors are observable, they stand in as proxies for the real variable of interest. For example, answering questions on an IQ test tells the researcher little if they do not indicate the latent construct of intelligence.
This is a relatively simple concept, but it is critical to the social scientific endeavor. Science is empirical, so scientific research findings must be based on observations by definition. To deal with important unobservable variables, social scientists must measure them with observable variables that indicate the presence and magnitude of the variable of interest. Without these indicator variables, there could be no scientific inquiry into abstract concepts.
A potential problem with using indicator variables as measures of underlying constructs is that there is no way of knowing whether or not the indicators accurately capture the underlying construct. Many believe that IQ tests do a good job of measuring the underlying construct intelligence, but some argue that there are “multiple intelligences” and that IQ tests only capture a small range of these salient characteristics. A potential method of dealing with this problem is to use a preexisting measure that has been subjected to the scrutiny of the scientific community. Published instruments will have undergone the peer review process and endured the criticisms of a potentially wide readership of scientific journals.
Intelligent consumers of social scientific research pay very close attention to how the operational definition of a concept relates to the theoretical definition of that term. This is because science requires researchers to show that they are measuring what they say they are measuring (and doing so with precision). If the measurements used by the researcher do not represent the construct, then the research lacks construct validity. If a researcher operationalizes intelligence in terms of scores on an IQ test, most people will acknowledge that the operationalization links back to the theoretical definition. If the researcher were to measure intelligence using hat size, consumers of the research would not take it seriously. Thus, the degree to which a measurement is considered valid is directly related to how good of a job it does in measuring what it is supposed to measure. It is safe to say that hat size as a measure of intelligence has no validity whatsoever.
Note that the idea of construct validity is not an all or nothing proposition. When a complex social phenomenon is translated into an operational definition, the odds are very good that something will be lost. In other words, a perfect measurement of abstract concepts is difficult to the point of being practically impossible. The idea is to do the best job that is practically possible in coming up with numerical representations of the social phenomenon of interest in a study. This points back to the importance of starting the process with good theoretical definitions. The theoretical definition is the starting point of measurement, and if it is not done well, the results will be suspect. As computer scientists are fond of saying, “garbage in, garbage out.”
Models in Science
When concepts and constructs are combined to specify how something in the “real world” works, the result is often referred to as a model. The term model is used because social scientific inquiry can rarely specify the myriad variables that determine human behavior. Think about what a model is in the everyday sense of the term. For example, architects often build models of bridges during the design phase. This model serves few of the purposes of a real bridge; you cannot drive vehicles across it. For explanatory purposes, the model is much more useful. It easily communicates the salient features of the design to others. From this example, it can be seen that a primary purpose of a model is to represent the real thing. That is exactly what social scientists use them for: To represent a social phenomenon.
Social scientific models can be descriptive (i.e., written in words), or they can be mathematical. Mathematical models are useful because they can demonstrate how the variables being considered in a study are related. A common type of mathematical model that is encountered in the social scientific literature is the regression model. This type of modeling uses a specific set of statistical techniques to specify the relationship between variables. When social scientists refer to “model specification” they are talking about the process of explaining how something in the social world works by specifying relationships between certain variables.
Students of the social scientists are introduced to how these models are specified and interpreted in a statistical methods course. Researchers specifying such models must take great pains in choosing which variables to include. Including variables that do not have much impact on the phenomenon being studied can muddy the waters and make the model substantially incorrect. Likewise, not including important variables creates a similar problem. The consensus among the scientific community is that these problems are best avoided by deriving models from theory. That is, models should be specified to be related in a way that is specified by a particular social scientific theory.
Modification History File Created: 07/24/2018 Last Modified: 08/09/2018
This work is licensed under an Open Educational Resource-Quality Master Source (OER-QMS) License.