Janice M. Morse
DEVELOPING QUANTITATIVE THEORY
Theories cannot be verified absolutely and forever; however, they can be falsified—i.e., they can be proven to be wrong—given a certain degree of certainty (or probability).
—Karl Popper (1959)
Quantitative theories are tools developed by the investigator to be deductively tested using particular statistical methods: theories are constructed logically from what is already known about the topic, considering the feasibility of measuring the concepts or variables in the theory. Once tested, the theory may be supported (i.e., statistically significant), or considered “weak” (i.e., having poor results but with some indication toward a positive relationship), or not statistically supported. Importantly, as Popper (1959) stressed, theories are not proven, but may be supported. They do not show us absolute truth, but rather probable truth, with some (not necessarily complete) degree of certainty. However, they can show us when we are wrong. In this way, theories can be modified and retested and positivistic science progresses.
Quantitative research therefore rests upon two components: the development of a theoretical model and the measurement and statistical procedures for testing the model. It is the purpose of this chapter to give you an introduction to the construction of theories and models used in quantitative inquiry. Although the concomitant research design and statistical interpretation are outside of our scope, these matters are discussed in many texts. However, an exception is with factorial designs, which allows the researcher to develop the model directly. The model itself serves as a framework, which is then “tested” in the setting. From these data, significant variables are selected. For example, Bubela et al. (1990) surveyed patients’ perceptions of their learning needs on discharge. Seven subscales were identified, including medications, activities of daily living, feelings related to their condition, community, and follow-up. These findings were used by Smith (1999) for a follow-up survey of patients to evaluate nursing discharge teaching.
THE ROLE OF QUANTITATIVE THEORY
Theory in quantitative inquiry has an extremely important role. Theory is the reason for conducting the inquiry in the first place; it is conjecture, with hypotheses to be tested, and therefore it must be worthwhile and linked to what is known. Testing a hypothesis is important. If supported, the findings will enable knowledge to move forward; if not supported, at least the investigator will know what does not work. Therefore, theory must be innovative and contain new ideas, but at the same time must be based on work that has been previously supported. It must be cumulative, logical, and convincing. And, in nursing, we often add another criterion: It must be applicable (i.e., useful to practice), or else lead to a more distant goal that will be applied and improve practice.
The development of the theory and subsequent testing is a deductive process, and sometimes a risky endeavor. The paradigmatic assumptions on which the theory is based may not be correct, yet because of the strength of the prevailing paradigm, the theory may be supported when tested. These “errors” in science are more easily identified in historical documents (e.g., in “The Diseases of Masturbation”; see To Do 5.1) and in class discussion you may think of other examples. But science generally, sooner or later, corrects itself.
“If I Can’t Measure It, It Doesn’t Exist”
Consider the McNamara fallacy:
The McNamara fallacy refers to Robert McNamara, the United States Secretary of Defense, involves making a decision purely on quantitative data, and ignoring others. The first step is to measure whatever can be easily measured. This is OK as far as it goes. The second step is to disregard that which can’t be easily measured or to give it an arbitrary quantitative value. This is artificial and misleading. The third step is to presume that what can’t be measured easily really isn’t important. This is blindness. The fourth step is to say that what can’t be easily measured really doesn’t exist. This is suicide. (Daniel Yankelovich, 1972)1
Although in this book we are not going to discuss measurement strategies, the role of measurement cannot be put entirely aside. Quantitative theory must, by definition, deal with what is measurable. Thagard (1999), a Canadian philosopher, writes extensively about theories that may be proposed but cannot be empirically supported (at least initially), and are therefore not considered credible. One recent example is the “bacteria theory of peptic ulcers” proposed by Warren and Marshall (1983) that was “considered preposterous” by his colleagues (Thagard, 1999, p. 41). The research trajectory of Warren and Marshall came in three stages: (a) from serendipitous observation, which they conducted from studies that identified and revealed the presence of bacteria, Helicobacter pylori, in the stomach; (b) that which showed that H. pylori caused peptic ulcers by locating the bacteria in the ulcers, and (c) that peptic ulcers can be treated with antibiotics. As their research continued, the medical community moved from rejection, to acceptance, to explanatory coherence between hypotheses and evidence, as other researchers replicated their studies. But the road was a little rough whereas the investigators gained credence for their work.2 The use of antibiotics for peptic ulcers is now standard treatment.
Explanatory coherence is the removal of alternative hypotheses. This is done by processes of logical argument and by statistically ruling out alternative explanations. Therefore to do this, excellent theory is required to be parsimonious, explicit, logical, and often developed by means of many intermediate goals. Explanatory hypotheses are controlled by research designs.
CREATING THEORETICAL FRAMEWORKS
In your research, you will probably create a theoretical framework. Theoretical (conceptual) frameworks are logical explanations and justifications to guide quantitative inquiry methodologically. Basically, the theory makes a case for conducting the research, the scope and variables to be included, and the selection of the sample and type of analysis to be conducted. Without such theory, the research becomes a descriptive “fishing trip” in which researchers explore large data sets for significant correlations; such correlations, in large data sets, may turn out to be spurious. In such cases, with exploratory data analysis, the researchers create possible explanations for the results (often spurious correlations) and subsequently conduct further confirmatory research.
The process of developing the conceptual framework for a quantitative study begins with (a) evaluating the literature, (b) identifying a quantitative question, (c) identifying the concepts, (d) operationalization, (e) determining measurement, and (d) developing appropriate research design. As the entire process is dependent on your question and the research designs are the purview of many texts, here I will address steps 1 to 5, and provide examples of several quantitative designs.
EVALUATING THE LITERATURE
Reading the literature to identify a topic is the most crucial part of the process, for it gets you started—and for many, “getting started” is the greatest hurdle. First, you must decide on a topic, which is a general area of interest, for without that you will not know where to start reading. Think through what you are really interested in, or think of clinical problems you have encountered in practice when you were hunting for something else online.
Think about what you often think about.
Identifying a topic or an area may be enough for now. Pull the literature and start reading. Think while you are reading. Do not skim; read broadly within your area. As you read, sort the literature into groups, according to the “schools of thought,” types of questions asked and answered, or the author’s theoretical perspective.
It is very helpful, at this point, to conduct a review or a meta-synthesis or meta-analysis, for a class paper. This may develop into an article that may even “do” as one of your articles for your dissertation, or even be publishable. But the most important thing is that it enables you to discuss the literature with authority. Sounding as though you know what you are talking about is a trait that can never be underestimated or undervalued.
Identifying the Research Question
This step is often a nightmare for students because it signifies a commitment. Therefore, the more you try to think of an original question (one that will make a contribution and hopefully fits in your advisor’s research program) and one that is the right size for a dissertation (which may be completed in a year, and does not require a huge sample or fancy instrumentation), and has the right price (affordable, cheap, and does not require outside grant funding), the more your minds becomes completely blank. So, step back and identify your aim or aims.
Identifying Your Aim(s)
You do this by thinking backward. Start thinking about the end of the project and try to identify what you want to know when you finish the project, which is your goal. That is the aim of your study (or those are the aims, if you may have more than one).
Now comes the question—or questions.
A question may just “pop” into your mind in the middle of the night. Grab it, research it, refine it …
Or, you may have to deliberately sit and write it down. Give yourself some leeway by asking “If I ask this _________, then I will have to __________.”
“If I ask that________, then I will have to __________.”
“If I ask thus and so__________, then I will have to __________.”
Looking at the problem from different directions opens your mind and makes things appear less rigid, less narrow.
Now select the best question for your purposes, and refine it. Refining your question may be tricky for quantitative research. First you may have to narrow the question by making it smaller, breaking it into several subquestions.
IDENTIFYING THE CONCEPTS
Often your questions and your reading will have provided you with many concepts in your area. List these. You will have a nice collection of scientific concepts, all with operational definitions with means of measurement. This is more likely to be so if you have a physiological question. But with a social or psychological question, often you will have descriptors rather than concept labels, or lay concepts. Look at these labels analytically: what does the description represent? Are there concepts available in the literature that would account for these descriptions?
If concepts are available, at this point revisit your question(s) and substitute phrases in your questions with the concepts. Once you understand operationalization,3 you will see that this step will be one giant step in the right direction for your project, for your project will probably be closely linked to an instrument that will enable measurement. Here is the great secret: The concept may be your variable. Alternatively, the attributes within the concept may serve as indicators for the concept, that is, serve as variables. Either way, what you have learned so far about the anatomy of concepts will be very helpful at this point.
If a concept is not available, or you cannot think of a concept that will address or fit into your question, then things will be a little more difficult, but not impossible. It may be that you have a question that needs to be addressed qualitatively, and the concepts developed. Or you may have to develop the concept. Quantitative researchers do not usually develop concepts qualitatively as we have described in Chapter 6. Rather, they often move directly to developing the concept as a scientific concept. They develop a label that fits the phenomenon they have listed in their question, develop a definition, and operational definitions as a means to measure the concept, and move directly to quantitative research.
For example, if you have a description that reads:
“Patient disagreeing with treatment.” Think—is that an example of noncompliance? Or is it an antecedent to noncompliance? (McDonald, 2013)
Using your knowledge of the anatomy of concepts (Chapter 7), analyze the concept of noncompliance as in the literature. Does it fit? Continue with this process of exploring scientific concepts until you find an appropriate concept to fit your statement.
The result of this process will probably be one of the following:
• You find a concept that is defined, operationalized, and has a selection of measurement tools that fits the phenomenon you are seeing. You accept that concept and move forward, creating your conceptual framework.
• You find a concept that is defined, operationalized, and has a selection of measurement tools, one of which fits your needs but does not have a measurement instrument associated with it for your context. This is not very good news, for creating an instrument is too much work and requires a much higher level of expertise than is required for a dissertation. Discuss it with your advisor: Either you will have to find another concept or accept the less than “ideal” match between your concept and the measure.
• You find a concept that is defined but is not operationalized. You may go ahead and operationalize it to fit your needs, if there is a tool that will then measure the concept.
• You do not find a concept label that fits your phenomenon. This is not a good sign. If you cannot find a concept, can you create one? No, that is not a poor idea for your first project. If you decide to continue this line of investigation, you may have no alternative than to change your question slightly and retarget the phenomenon. Discuss the problem with your advisor.
Operationalization is the process of defining something in terms of measurement. We consider our variables (concepts or attributes) in a way that they can be measured. Measurement may be by means of a survey questionnaire, a psychological instrument, a physiological measure, an observed behavior, and so forth.
How do you operationalize? Operationalization differs from a definition: A definition provides the meaning of the concept; when operationalizing, one develops an empirical meaning for the concept. From this operational definition, the variables may be identified. Ask yourself: What indicators, or observations, will show the presence of the concept? If concepts have operational definitions that fit into theory, then they are useful. The operational definition must link to measures that are accurate and precise.
If you have found a concept that is defined, there are two schools of thought about the next step. The first is to identify the concepts, operationalize variables, and locate an instrument that fits the operationalization (Geering, 2012; Loseke, 2013; Figure 27.1a). The second is to operationalize by identifying the question, and from there identify measurement instruments, and refine the definition of your variables to fit the instrument (LoBiondo-Wood & Haber, 2014; Remler & van Ryzin, 2011; Figure 27.1b). All of these authors agree that research consists of interconnected parts, and there must be a logical flow from one step to the next. Loseke (2013) reminds us that it is the research question that drives the process of writing the literature review, identification of the concepts, and further, what data to collect, and how to interpret these data.
Let us walk through an example: Suppose the concept is “disagreeing with treatment.” Our description tells us that the patient did not argue with the physician, but you saw he had not been prescribed the medication that he expected, wanted, or requested. You doubt that he will bother to fill the prescription, or if he does, you guess he will not take the medicine. Noncompliance is defined as “failure or refusal to comply,” but at the point the patient has been prescribed the medication, then he has not refused to comply, according to the definition. If we do decide to use “noncompliance” as our concept, we could define it as “reluctant acceptance of treatment.” Alternatively, we could reject noncompliance as a concept and label the concept as “pending noncompliance” or “pre-noncompliance.” Because that concept does not exist, the choice (and the appropriate operational definition) is yours.
But now you have to operationalize the concept. This consists of translating the concept into observable or measurable events, so that it can be:
• Consistently recognized by others