I am following an online college/graduate-level class about research methodology and writing. There are things in the class that I have not known before. (If you have any recommendation about literature on research methodology and writing, feel free to let me know. I have heard of some in the past, but not really read any. It is always nice to know or be reminded of which sources are good.)

The class covers the following topics:

  • Select and refine a research topic. 

  • Locate and analyze the relevant literature. 

  • Formulate specific research questions that address practical or conceptual problems in light of the relevant theoretical and empirical literature. 

  • Formulate a set of hypotheses to address your research questions.

  • Craft a research plan, identifying an appropriate design, methods, and analyses to test your 

  • ...

During the step of hypothesis formulation, it distinguishes between

  • The Theoretical Plane, where we think in terms of Propositions comparing Constructs. A construct is how we express our concepts, and a proposition is a tentative and conjectural relationship between constructs that is stated in a declarative form.

  • The Empirical Plane, where we think in terms of Hypotheses comparing Variables.

    Abstract constructs cannot be directly tested. Variables can be discovered that can be used to indirectly test these constructs and their relationships with other constructs.

    Constructs cannot be directly measured and must be operationalized before you can enter the Empirical Plane and actually start doing any work. For measuring constructs, develop metrics, or measurable indicators. A combination of indicators is a variable. Comparing indicators may assess accuracy.

I am not doing research at frontier, but self studying a new topic. So I am trying to write some survey about comparing concurrency models in computer science. More specifically, compare Actor Model and Communicating Sequential Processes (CSP). From some literature, I have seen that the models are compared by the syntax and semantics of the formal languages that describe them, and how they are used to solve some common concurrency problems (mutual exclusion, dining philosopher), and some conclusions are drawn:

  • CSP is more flexible than actor model: In actor model, the medium of communication is tightly coupled to the unit of execution: each actor has precisely one mailbox; In CSP, channels are first class and can be independently created, written to, read from, and passed between tasks.

  • Nothing stops CSP from supporting distribution and fault tolerance, but historically CSP has not had the same level of focus and support of the two as actor model does.

  • Both actor model and CSP do not directly support parallelism. Parallelism has to be created based on concurrency building blocks.

The above aspects are, as far as I know, about whether a model possesses a property, or some quality. They feel like "constructs", and make me look like staying in "The Theoretical Plane". In order to "enter the Empirical Plane and actually start doing any work", I am trying to look for and come up with some metrics, or things directly measurable, but find it hard to measure the models in the aspects, if "measurable" means being able to quantify.

I wonder if it is common in all or most scientific research areas that "For measuring constructs, develop metrics, or measurable indicators. A combination of indicators is a variable. Comparing indicators may assess accuracy."?

Does that (looking for metrics or measurable indicators/variables) also apply to some computer science theories, which study about formal models? (Similarly to mathematics, which is also a formal science?)

I should also mention that the background of most of the attendees of the class is analytics or statistics, where I happen to have some experiences. But the class is not open just to folks in that area. Even in analytics or statistics, do all or most researches have metrics or measurable indicators/variables? (I believe we can study some formal mathematical models in statistics, but I am not sure about the question)


  • 1
    'Constructs cannot be directly measured and must be operationalized before you can enter the Empirical Plane and actually start doing any work' It ain't necessarily so. In the physical sciences, grand-theoretical propositions are often written in terms of quantities that are directly measurable; and in the social sciences, there's an entire discipline of "qualitative research" that seeks to achieve direct empirical investigation of target concepts without needing in intervening operational definition. Commented Dec 16, 2020 at 13:20

1 Answer 1


Much of research deals with the messy real world and often depends on statistics to achieve approximate answers or answers that are true of populations but not individuals (not necessarily human).

So the answer to your topline question is likely yes. "Most" research areas look for metrics. But not all.

But pure mathematics and some other things aren't like that. Theoretical physics, for example, might not be like that at all. Mathematical theorems aren't "empirical".

Some things in applied mathematics are close to the given description. And statistics itself, as a mathematical construct, isn't empirical, but its applications are. Social Sciences are almost entirely empirical (there may be exceptions that I'm not aware of).

But, since you tag mathematics here, let me note that "hypothesis formulation" holds, but in its own way. To do (pure) mathematics effectively you need insight into a subfield, not just knowledge of what is known. That insight leads you to ask questions about what might be true but isn't yet known. This leads you to formulate a potential theorem (a hypothesis) and then to seek mathematical evidence (a proof or a contradiction) to settle the question.

Back when, someone asked how the derivative is related to the integral. The result of the exploration was the fundamental theorem of calculus.

Computer science can be a lot like mathematics. Theoretical Computer Science especially. But some things are empirical. Asking whether one algorithm is superior to another can be either a theoretical question or an empirical one. How much better is one compiler optimization scheme than another is often in the latter category.

  • This answer contains a great explanation of how mathematicians do research! I may have to borrow it the next time people try to ask me about what math research is
    – GageMartin
    Commented Dec 18, 2020 at 14:58

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .