This page summarizes A Primer in Theory Construction, written by Paul Davidson Reynolds. Reynolds states that the purposes of his book are:
For more extensive information about theory construction than is provided here, including descriptions of testing theories and strategies for developing a scientific body of knowledge, see:
- to describe the various types of concepts and statements that comprise a scientific body of knowledge.
- to describe what form they should take to be accepted within the community of scholars.
Definitions of Theory
- Gibbs, Jack P. 1972. Sociological Theory Construction. Hinsdale, IL: Dryden Press.
- Hage, Jerald. 1994. Formal Theory in Sociology: Opportunity or Pitfall? Albany, NY: State University of New York.
- Reynolds, Paul D. 1971. A Primer in Theory Construction. New York: Bobbs-Merrill.
Definitions of scientific theory fall within three general forms:
Reynolds defines theory as "abstract statements that are considered part of scientific knowledge in either the set-of-laws, the axiomatic, or the causal process forms." Thus, Reynolds definition focuses upon the notion of a set of abstract statements that describe "things we feel certain about." This definition differs somewhat from the one provided by Dr. Sapp ("A set of abstract, empirically falsifiable statements about reality"). In Dr. Sapp's definition, a theory might not yet have sufficient empirical support to "feel certain about it," but it nevertheless meets the criteria of science as an epistemology in that it posits abstract statements that can be falsified through observation.
- The "set-of-laws" form defines theory as a set of well-supported empirical generalizations, or "laws." Here, theory is thought of as "things we feel very certain about."
- The "axiomatic" form defines theory as a set of interrelated propositions and definitions derived from axioms (i.e., things we feel certain about).
- The "causal" form defines theory as a set of descriptions of causal processes. Here, theory "tells us how things work."
Abstract concepts are independent of a specific time and place. Because scientific statements must predict future events, they cannot be specific to past events. Scientists prefer theories that are as general as possible to time and place.
Abstract concepts are independent of specific circumstances or conditions. This independence permits efficiency in understanding and predicting future events. Thus, the statement, "the greater the human capital investment, the greater the life chances," contains two abstract concepts: human capital investment and life chances. This statement can be used to derive and test a large number of related hypotheses, such as:
The process of science is one of moving continuously from one level of abstraction to another. Scientists "borrow" abstract statements from theories to derive hypotheses suitable to their specific study. They test these hypotheses through observation. They "return" the results of their studies to the theory by reporting to the community of scholars the efficacy of the theory in explaining their observations. Supported hypotheses provide further support for and confidence in the theory. Rejected hypotheses prompt consideration of revising the theory or noting that it is less broadly applicable than originally believed. A scientific body of knowledge is accumulated by this ongoing process of borrowing, testing, revising, and building new theories.
- The greater the formal education, the greater the income.
- The greater the job experience, the greater the likelihood of promotion.
- The greater the communication skills, the greater the job performance.
- ... and so on.
Empirical relevance refers to meeting two conditions of observation:
- Scientific theories must be falsifiable. The distinguishing feature of science, in contrast with other epistemologies, is that its statements can, in principle, be rejected through observation.
- Scientific theories must be supported by observations. When theories receive strong empirical support, then we gain confidence in them, which allows us to build safe bridges, send satellites into orbit, design effective crime prevention programs, etc.
It is impossible to be completely objective, value-free, and unbiased in any human endeavor, including scientific inquiry. Thus, as described on the Primer in Science web page, the community of scholars relies upon the "safety of numbers" in reducing the likelihood that science reflects human frailties. Intersubjectivity refers to shared understandings among the community of scholars within a particular discipline. This intersubjectivity must be achieved with respect to the definitions of concepts and the interpretation of the results of empirical observations.
Theories are stories, stories about how reality works. They differ from other stories in the ways described above: they are abstract, causal, and falsifiable. Nevertheless, they are stories about reality. And they come from somewhere. Much has been written in the philosophy of science about induction and intuition, the twin processes by which new theories are crafted, where induction refers to designing theories by combining and raising to an abstract level empirical generalizations and intuition refers to the "great thought" about how something works. Whether meticulously crafted or designed by inspiration, or some combination of both, it is the idea that describes how reality works and helps us improve the human condition by this knowledge. Davidson notes that the ultimate test of any idea is its utility in meeting the goals of science. "There is no substitute for a good idea."
For the purposes of Sociology 401, we will direct our attention to sociological thought rather than to philosophical inquiries into how ideas are formed. Also, we will cover very few sociological theories except where we discuss them as examples of sociological thought. Thus, we will focus our attention on the "big stories," or "worldviews" that guide the development of sociological theories. These grand schemes are called paradigms. They influence theoretical development throughout a scientific discipline. Evolution, for example, is the predominant paradigm of the life sciences; all theories in the life sciences rely in part upon the principles of evolution to depict biological events.
As you would expect, paradigms are more stable than theories. A theory might be revised or even rejected without necessarily revising or rejecting the paradigm that influenced its development. However, with sufficient anomalous findings, revisions to theories, and rejected hypotheses, scientists might begin to question the paradigm itself. This questioning of the paradigm can create "crisis events," wherein much of the body of knowledge within a scientific discipline is questioned. For example, after many years of collecting sufficient evidence to the contrary, astronomers eventually rejected the Ptolemian paradigm of the solar system (i.e., the geocentric or Earth-centered perspective) in favor of the Copernican paradigm (i.e., the heliocentric or Sun-centered perspective). The dynamics of paradigm development, crises, and revolutions in scientific thought are described in detail by the philosopher Thomas Kuhn in The Structure of Scientific Revolutions.
Theories: An Overview
As a generalized schema, consider this depiction of a scientific theory. This presentation incorporates elements from various authors on the philosophy of science. Later, we will rely upon Reynolds' text to describe each feature of a theory in more detail.
The results of the statistical test of the research hypothesis (presuming it is measured quantitatively) might lead the researcher to reject the null form of the hypothesis (i.e., "There is no relationship between formal education and income."). If so, then the results of observation lend support for the hypothesis, the proposition, the theory, and the paradigm. If the null hypothesis is not rejected, then the community of scholars will explore reasons why it was not supported, including the notion that the theory (and perhaps the paradigm) might not be a correct depiction of reality.
- Paradigm: A worldview. "What is reality?"
Example: Functionalism posits that people negotiate the rules of society to meet their survival needs.
- Theory: A set of empirically falsifiable, abstract statements about reality.
Example: Social systems theory posits that to achieve their goals people meet socially-defined role expectations associated with their statuses.
- Proposition: One abstract statement within a theory.
Example: "The greater the human capital investment, the greater the life chances."
- Hypothesis: A specific case of the proposition.
Example: "The greater the formal education, the greater the income."
- Operational Definition: The description of how each concept will be measured.
Example: "The greater the years of formal schooling, the greater the total household income before taxes."
Concepts, the building blocks of theories, are symbols designed to convey a specific meaning to the community of scholars. They must be defined, operationalized, and reviewed by the community of scholars for meaning and accuracy. The concept self-esteem, for example, is defined as, "an individual's sense of his or her value or worth," and most often is measured using Rosenberg's Self Esteem Scale, which is widely accepted by the community of scholars.
- Concepts are defined with either primitive or derived terms. Primitive terms cannot be defined with other symbols or language (e.g., colors, sounds, attitudes, some relationships between individuals), but can only be further described through the use of examples. A derived term is a set of primitive words and symbols that further describes a concept.
- An abstract concept refers to two or more events (e.g., temperature, human capital investment). A concrete concept refers to a specific event (e.g., temperature of the sun, years of formal education).
- Concepts can be measured either quantitatively or qualitatively. There is no epistemological reason to suspect that either type of measurement is more or less scientific, objective, or valid.
- Concepts can be measured at the nominal level, indicating no inherent ranking (e.g., male, female; Christian, Hindu, Muslim, Jewish), the ordinal level, indicating ranking without a continuous ordering (e.g., large, medium, small), the interval level, indicating ranking with a continuous ordering, with no known zero-state (e.g, attitudes about same-sex marriage expressed on a 1-7 response scale), or the ratio level, indicating continuous ordered ranking with a known zero point (e.g., age in years).
Statements are expressions about reality. They can be classified within two groups, those that claim the existence of an event and those that describe a relationship between two concepts.
Forms of Theory
- Existence statements state that a concept exists (e.g., the object is a primary group) or describe a relationship that exists (e.g., each individual in the group contacts all other individuals at least once each week).
- Relational statements posit a causal or associational relationship between two or more concepts (e.g., the greater the formal education, the greater the income). Note that, formally, a relational statement can include only two concepts that vary because hypothesis testing, whether quantitative or qualitative, can only be conducted on two variables. For example, the expression, "the greater the formal education the greater the income and self-esteem," cannot be tested because it might be true that the greater the formal education the greater the income, but it might not be true that the greater the formal education the greater the self-esteem. The statement must be split into two expressions, one for income and one for self-esteem.
An apparent exception to this rule occurs when one tests whether a theory fits the data because it seems like one is testing multiple hypotheses at once. But this is not the case; rather, one is testing just one relationship: the one between the whole theory and the data. It is possible, however, to include one or more constant concepts within a formal relational statement to restrict the domain of the statement (e.g., Among males, the greater the formal education, the greater the income).
Note also that articles published in professional sociological journals often include three or more variables within an "hypothesis" to convey multiple relationships in a concise manner.
- Associational statements state a relationship without implying cause. For example, we might state that, "locus-of-control and self-esteem (two concepts with similar meanings) are related," meaning they will vary together but not necessarily cause one another.
- Causal statements imply that x causes y (e.g., the greater the formal education, the greater the income).
- Theoretical propositions state relationships in an abstract form (e.g., the greater the human capital investment, the greater the life chances).
- Hypotheses state relationships in a concrete form (e.g, the greater the formal education, the greater the income).
Theories can be expressed as a set of laws, in axiomatic form, or as a set of causal statements.
Note Regarding the Format of Theory
- The set-of-laws format expresses relationships as a set of highly supported laws (i.e., typically in causal form). Consider, for example, the Theory of Reasoned Action, proposed by Martin Fishbein and Izak Ajzen. Within this theory we might state as one law, "the greater the attitude about the behavior, the greater the intention to engage in the behavior." All the other paths implied by the diagram would be listed as laws within the set of laws that define the theory of reasoned action.
- The axiomatic format expresses relationships as a set of axioms. For example, within the theory of reasoned action, we might state as one axiom, "If attitude toward the behavior, then intention toward the behavior." All the other paths implied by the diagram would be listed as axioms within this format.
- The diagram shown for the theory of reasoned action represents the causal statement form. Each diagrammed path represents a theoretical proposition. For example, we might infer from the diagram of the Theory of Reasoned Action that, "the greater the attitude about the behavior, the greater the intention to engage in the behavior."
The typical format used in sociology to express a theory is the set of causal statements, often shown in a concise manner by the use of a diagram. In the 1980's, as part of an effort to make sociology "more scientific," sociologists began to present their theories in axiomatic format (see volumes of The American Sociological Review for examples of this effort). Sociologists learned quickly that the formatting of a theory provided few advantages toward accumulating a scientific body of knowledge; what mattered was the quality of the theory, not its formatting. Note, however, that some sociologists will argue that "theory" should be expressed either as a set of laws or in axiomatic format (see: Formal Theory in Sociology: Opportunity or Pitfall?, edited by Jerald Hage).