Skip to content

Thesis writing basics (2.1) Theory and evaluation criteria of theory

This article summarizes a peper, “Organizational Theories: Some Criteria for Evaluation” written by Samuel B. Bacharach in 1989. The original text can be found at the following link. In this article, we will look at what are the theories and the criteria for the evaluation of theories. For a list of articles related to the basics of writing a thesis, please see the links.

■ Theory is not classification of data, typology, and metaphors!

Descriptions are different from theories. Descriptions can be the source of a theory, but it cannot be a theory itself. Theories explain and predict reality based on how, when, and why, while descriptions mainly focus on what.

There are three types of description. The first is categorization of the raw data, the second is typology, and the third is metaphor. Often, articles describing analysis are confused with theories, but the classification of quantitative or qualitative data cannot be a theory. The analytic description answers the question of “What is the phenomenon”.

Typology comes from the “ideal type” by Max Weber, German economist and sociologist. Ideal type is one that is constructed through a thought process, not a model by experience. For example, Etzioni’s compliance structure is not a theory but a typology. Etzioni classified the organization into 9 types based on power and involvement, which are abstract than categorization of the raw data, but still focus on the “what”. Typology answers to the question of “what is the most important factor in the phenomenon?”.

Metaphor answers to the question of “How is this phenomenon different from another phenomenon?”. For example, metaphors are like loosely coupled systems (Weick) and garbage can model (Cohen, March, and Olsen). Metaphors are great literary tool because they adds imagination, but they cannot be theories in themselves.

■ What is a theory?

A theory is a statement of the relationship between observed units or approximated units in the empirical world. Since the essence cannot be observed directly, the object must be estimated in terms of observation. A unit of observation is a variable that operates empirically, and the main goal of a theory is to answer questions about how, when, and why rather than answering what.

In other words, theory refers to an entire system of constructs and variables. There is a proposition between the two constructs, and a hypothesis between the variables. For an explanation of constructs, variables, propositions, and hypotheses, see this article.

Figure 1 in the paper

Theory has scope. The scope includes limits on space and time either explicitly or implicitly set by the scholar. The more specific the assumptions about the scope, the more difficult it is to generalize. For example, let’s say you have a theory that is limited only spatially. In other words, this theory can only be applied to certain types of organizations, not to organizations at other times. The degree of generalization of this theory will be higher than that of a theory with both time and space constraints.

■ Evaluation of theories – Falsifiability and Utility

Falsifiability means whether a theory is empirically falsifiable. The ideal goal of science is the pursuit of universal truth, and most philosophers of science believe that “theories cannot be proven, they are only disproved” (Nagel 1961; Popper 1959). In other words, the theory is like a defendant in a court of law before he/she is found guilty.

Utility means the usefulness of a theory. Utility is the bridge between theory and research (Bierstedt 1959). Theories are evaluated as useful if they can explain and predict. Explanations establish constructs, variables, and the associations between them, and predictions test their practical meaning against empirical evidence. In the case of ancient astronomers, it was predictable, but it was not possible to adequately explain the observed phenomena. Thus, it remains an unreliable theory.

Let’s take a closer look at this according to its falsifiability and utility.

  1. Falsifiability of variables
  2. Falsifiability of constructs
  3. Falsifiability of relationship
  4. Utility of variables
  5. Utility of constructs
  6. Utility of relationships
■ Falsifiability of variables, constructs, and relationships

Variables must be defined and consistent in terms of measurement. Therefore, it is necessary to pass a measurement model that satisfies the validity, noncontinuousness, and reliability. Only by including variables that can be measured meaningfully and accurately can the theory be confirmed. In addition, the experiment must be discontinuous, but if it is continuous, it will not be able to be verified accordingly without constraining time and space. The falsifiability of variables is thus associated with ‘measurement’.

The falsifiability of the constructs is related to construct validity. It focuses on attributes rather than the measurements themselves. Construct validity indicates how properly the measurement tool measured the abstract concept. To meet this requirement, responses from measurements of the same constructs must share variance. That is, it must meet convergent validity, which means the concordance of the questions being measured (Schwab 1980). In addition, different concepts should not share the same attributes and should be empirically distinct. This is called discriminant validity. Accordingly, it should be possible to distinguish it from other constructs that may be similar and point out irrelevant things.

Now it is necessary to evaluate the adequacy of the relational elements of the theoretical system. In other words, it is necessary to prove that the connection between the elements is sound, and to examine the logical and empirical adequacy here.

Logical adequacy is the logic that guarantees that hypotheses and propositions may not be confirmed. Therefore, propositions and hypotheses are nontautological, and the characteristics between antecedent and outcome must be specified. In the sense of nontautological, in order for a proposition and hypothesis to be disprovable, the existence of an antecedent may not automatically mean the existence of a consequent. This means that the characteristics between the antecedent and the consequent should be stated explicitly whether the prior condition is a necessary condition, a sufficient condition, or a necessary and sufficient condition.

Empirical adequacy is the second criterion for assessing the falsifiability of relationships inherent in a theory, meaning that the object of analysis must be at least one and must exist at least one point in time.

Moving on to utility, the utility of variables and constructs depend on scope. Variables should cover the area of the constructs by assuming sufficient but minimal changes (parsimonious), and the area of problem phenomena should be treated with the assumption that the constructs is sufficient but only minimal change. In the utility of a relationship, it means explanatory potential and predictive adequacy.

The explanatory potential of the theory can be compared with (1) the degree of embodiment of the assumptions associated with the analysis, (2) the degree of embodiment of the assumptions associated with the relationship between the antecedent and the consequent, and (3) the scope and parsimony of the propositions. Theories in which assumptions are stated are preferred over theories that do not. In addition, a theory with a wide range of analyses is preferred over a theory with a narrow scope.

■ Falsibility of variables, constructive concepts, and relationships

Today’s research aims to enable theories and statements to be empirically tested, and to provide a source of explanations and predictions. To write a good theory, you need to remember the following:

1. It describes the scope of the theory and at the same time specifies the description of the assumptions (values, space, times) that define the scope of the theory.
2. Use common language for constructs and variables at all levels.
3. Clearly distinguish between propositions and hypotheses, and the relationships contained therein.
4. Improve the parsimony of the theory.


  • Organizational theories: some criteria for evaluation (Samuel B. Bacharach 1989)
  • Educational organizations as loosely coupled system (Weick, K. 1976)
  • A garbage can model of organizational choice (Cohen, M., March, J., & Olsen, J. 1972)
  • The structure of science: Problems in the logic of scientific explanation (Nagel, E. 1961)
  • The logic of scientific discovery (Popper, K. 1959)
  • Nominal and real definitions in sociological theory (Bierstedt, R. 1959)
  • Construct validity in organizational behavior (Schwab, D. P. 1980)

Leave a Reply

Your email address will not be published. Required fields are marked *