Theoretical research gap:definition, steps of identifying theoretical gap and examples

1.1 Background

What is theoretical knowledge gap? Different authors define this term in dissimilar ways as they trying to come up with a common meaning or understanding. But most of the times they do not get it right. We have a different and a more appropriate way of defining and elaborating this concept.

First, what is a gap in research?

Definition

A gap is the link which is still missing on the side of the researcher between one variable and another or amongst several variables due to inefficient knowledge or due to lack of knowledge at all. Remember that lack of enough knowledge does not mean that the two variables do not relate to one another. The fact is that there is an underlying relationship only that the researcher/scholar has not discovered or has not unearthed that fact. So failure to unearth or discover that there exists a plausible/logical relationship is the research or knowledge gap.

Two, what is then “Theoretical Gap?”

Theoretical gap is also referred to as Conceptual knowledge gap hence the two terms are used interchangeably by most researchers for the two terms are one and the same thing. To answer this question, we will start small by looking at some basics in research and grow up to the full understanding.

Definition

Theoretical gap (Conceptual) is the missing gap of knowledge which is related to the role a variable plays in an already existing body of knowledge and it is a sub-set of research gap. This means that the researcher or you as a postgraduate student you may propose the role that you strongly feel that a variable or factor may play in a relationship. The role can be of predicting nature, so the variable in the model is classified as a predictor or independent variable. It can also be treated as a dependent variable, or intervening or moderating hence treated as either dependent/outcome, intervening or moderating variable in that order.

 

In other words, as explained in our sub topic on formulation of effective research hypothesis, the role the specific variable plays in explaining a model, is referred to as a variable position and it dictates the type of character/behavior that variable explains in the model. So a variable that explain a certain concept assumes a conceptual variable position. Now this means theoretical or conceptual knowledge gaps are further classified on based on the role the variable statistically and significantly explains in a model. Those roles are commonly categorized as;

 

1.2 Theoretical or Conceptual Variable Position

1.    Independent variable
2.    Dependent variable
3.    Intervening variable
4.    Moderating variable
 

Each type of variable has a distinct conceptual/theoretical variable position as explained in the following examples;

1.2.1 Independent Variable Position

Carry explanatory power.

Example 1: Independent Variable

An independent variable plays the explanatory role of how or the extent to which the variable/construct influence another variable referred to as the dependent or criterion variable. So the researcher manipulates this variable to observe the changes on the dependent variable. The independent variable is used to predict the changes that will occur on the dependent variable.

Therefore, if the results are statistically significant (remember the case of p-value being less than critical value e.g. 0.05), holding other factors constant, then it is empirically proven that there was a conceptual gap which was in existence only that there was no adequate data or information to prove or disapprove whether that particular variable played the role of predicting the outcome of the dependent variable.

 

1.2.2 Dependent Variable Position

Carry dependency of outcome power

Example 2: Dependent Variable

On the other hand, the dependent variable plays the role of being influenced by another variable referred to as the independent or predictor variable. So the researcher manipulates the predictor variable to observe the changes on the dependent variable. The dependent variable is used to observe the changes that will occur due to independent variable and whether it will be significant or not.

Therefore, if the results are statistically significant (remember the case of p-value being less than critical value e.g. 0.05), holding other factors constant, then it is empirically proven that there was a conceptual gap which was in existence only that there was no adequate data or information to prove or disapprove whether that particular variable played the role of dependent variable.

Illustration 1: In this first illustration, we will combine both independent and dependent variable to bring out the theoretical/ or conceptual knowledge gap well for there is no concept or theory which is represented by one variable/construct. So, if the researcher hypothesized that variable “P” play the role of independent variable while variable “W”, the role of dependent variable, Figure 1.1 below demonstrates this concept.

 

 

Note that if P relates with W in the natural/physical phenomenon in a statistically significant manner, then it implies that there existed conceptual knowledge gap between P and W only that may be there was no information or data available to support that relationship. The point here is that it was not known or proven before; hence, there was a theoretical or conceptual knowledge gap that was in existence. This is what is referred to as a THESIS.  

Illustration 2: Suppose another researcher proposed that Y also predicts W. This is a new concept and if again it is empirically proven that Y statistically and significantly ((remember the case of p-value being less than critical value e.g. 0.05) explain the changes in W, then this represents a theoretical/conceptual knowledge gap that has been successfully filled through further research. This is represented as follows as per Figure 1.2 below

 

 

Note that if Y relates with “W” in the natural/physical phenomenon in a significant manner, then it implies that there also existed conceptual knowledge gap between Y and W just like in the case of P and “W” as proved earlier on, only that may be there was no information or data available to support that relationship. The point here is that it was not known or proven before; hence, there was a conceptual knowledge gap that was in existence. This is also what is referred to as a THESIS.

NB: That illustration one, where W is a function of P i.e. W=f(P) represents body of knowledge that already exists as discovered by the first researcher. If the second researcher empirically proves that W is also a function of Y i.e. W=f(Y), then the second researcher has added new knowledge that Y statistically and significantly explains the changes in W, the dependent variable as it is in the case of P although the magnitude of change may vary.

 1.2.3 Intervening Variable Position

Carry intervening or a mechanism power

Example 3: Intervening Variable

An intervening variable plays the intermediating role of more effectively explaining the reason as to why the independent to dependent variable relationship exists. The intervening variable provides the mechanism behind the independent variable explaining the changes in the response/dependent variable in a statistically significant manner.

Qualifications of a Variable as Intervening Variable;

Not all variables are intervening variables. Many students suppose that any variable a researcher introduces in a direct empirical model is intervening in nature. That is not true. Again, all intervening variables may intervene a relationship at different levels such that some variables are more intervening than others. Ok, in our current matter, a variable qualifies to be an intervening variable in a certain relationship if;

One, it is statistically and significantly predicted by the independent variable.

Two, the intervening variable itself statistically and significantly influences the dependent variable in the absence of the independent variable.

Three, the variable is inside the model. In other words, it is part of the model for it first has a direct influence to the dependent variable when the independent variable is absent and when the three variables are incorporated in the model, it still has an intervening effect.

 So if the researcher successfully identifies such a variable which has the two qualifications aforementioned, then the role it plays of explaining the mechanism behind the current model represents a theoretical or conceptual knowledge gap. Therefore, if the results are statistically significant (remember the case of p-value being less than critical value eg 0.05), holding other factors constant, then it is empirically proven that there was a conceptual gap which was in existence only that there was no adequate data or information to prove or disapprove whether that particular variable played the role of intervening the relationship between the independent and dependent variables.

Illustration 3

Suppose in the body of knowledge researcher three maintained the position status quo of variable Y as the independent variable while W still played the role or position of dependent variable. However, he further introduced an intervening variable by the name “I” which portrayed significant intervening effect on the relationship between X and Y. This can be presented using Figure 1.3 as follow below.

 

 

This is new conceptual knowledge added to the already the one in existence and represents THESIS of researcher three for it is empirically or scientifically proven that “I” significantly intervenes on the relationship between Y and W. This is also a conceptual role or position taken by variable “I” to explain more on the Y-W relationship.

1.2.4 Moderating Variable Position

Carry conditional power

Example 4: Moderating Variable

A Moderating variable plays the intermediating role of providing a condition that has to prevail for the relationship between independent and dependent variables to exist. The moderating variable provides either enhancing, buffering or antagonistic conditions behind the relationship under investigation. 

Qualifications of a Variable as Moderating Variable;

A moderating variable has to have characteristics of a moderator as substantiated in our intervening and moderating variable article in our Accountingnest.com website.

 So if the researcher successfully identifies such a variable which is truly a moderator as aforementioned, then the role it plays of explaining the condition(s) behind the current model represents a conceptual knowledge gap. Therefore, if the results are statistically significant (remember the case of p-value being less than critical value eg 0.05), holding other factors constant, then it is empirically proven that there was a conceptual gap which was in existence only that there lacked data or information to prove or disapprove whether that particular variable played the role of moderating the relationship between the independent and dependent variables.

Illustration 4

Suppose in the body of knowledge researcher four maintained the position status quo of variable Y as the independent variable while W still played the role or position of dependent variable. However, he further introduced a moderating variable by the name M which portrayed significant interaction term effect on the relationship between Yand W. This can be presented using Figure 1.4 as follow below

 

 

This is new conceptual knowledge added to the already the one in existence and represents THESIS of researcher four for it is empirically or scientifically proven that “M” significantly moderates on the relationship between Y and W. This is also a conceptual role or position taken by variable “M” to explain more on the Y-W relationship.

 

Steps of identifying theoretical/conceptual gap

It is very common with researchers especially in academia to arrive at statistically insignificant research findings. So the students are forced to concoct or forge the results to read what is pleasing to the end users. To avoid this embarrassments, the following step by step approach should be used to get statistically significant ((remember the case of p-value being less than critical value eg 0.05) research findings

Step One: Identification of the Researcher’s Area of Interest/Field

This is the right footing. There is no way you can carry out research on an area that you have no interest in. Many of my Postgraduate students keep on asking me for a topic to write a proposal on. Others are assisted by friends or consultants. Its ok, but ones interest is not another. What is good for the duck may not be good for the goose.

At this stage, the researcher need to keenly identify the motivation behind the study he or she wants to undertake. This will then be obviously pegged on your area of interest. Do not expect your supervisor or your audience of your research findings to be excited or motivated if you are not excited or motivated by your own proposal or dissertation in the first place.

I always liken this mentality to a scenario whereby;

“There are mothers who make food for their young ones and they cannot even taste it to know the flavor thereof for they complain that the food look funny, tasteless, vomit initiating and so on. You hear them say mmm…mmm, I cannot taste, its for my baby. So if such mothers cannot eat or taste what they have made for their babies, why give the same food to the baby to eat?”

So, if the topic of your study that you have chosen is not interesting or motivating you in the first place, why do you think it will excite your supervisor or your audience? They will obviously throw it through the window as you watch. Therefore, the starting point is identification of the area of interest you want to focus on. This can be determined of course by either your area of specialization or as guided by your sponsor if you are carrying the study on behalf of somebody else or an interested organization/institution.

Step Two: Review Past Studies on the Area of Interest

In step two, you need to intensively interrogate past literature to get backup of the specific areas already researched which form the already existing body of knowledge from your predecessors in that field. This will also help you to know the commonly studied variables that you do not re-invent the wheel by introducing strange and new variables that does not exist at all in the natural phenomenon.

Through these reviews, automatically you will be able to infer variables with statistically significant relationships and those which does not.

Step Three: Identify Variables that Portray a Linkage in their Natural Phenomenon

In this step, you carefully identify the variables that portray an association in the natural phenomenon.

It is at this point where most researchers or postgraduate students undertaking their study make mistakes and pick variables anyhow. At the end of the exercise, they analyze the data collected only to be embarrassed when they find that the relationship may not portray statistically significant results.

Therefore, in this step, the researcher should consider variables which have theoretical foundation. For such variables will almost give 100% assurance of results that are statistically significant.

Let me give you two examples;

Example 1

Case 1; If you consider the relationship between dog biting a human being-the results are most likely going to be statistically significant. This is because in the natural phenomenon, dogs do bite human beings. Therefore, the results will be statistically significant.

Note: Dog bite is the predictor variable while man bitten by a dog is the dependent variable.  

Case 2; If you consider the relationship between a human being biting dog-the results are most likely not going to be statistically significant. This is because in the natural phenomenon, human beings do not bite dogs. Therefore the results will be insignificant. Even by the virtue that you may not get sufficient data of a man biting a dog, is good enough to justify why no statistical significance link may be established.

Note: man biting a dog is the predictor variable while Dog bitten by a man is the dependent variable.  

Example 2

Case 1: If you consider the relationship between price of a commodity and quantity demanded of the commodity-the results are most likely going to be statistically significant if the price is the determining factor. This is because in the natural phenomenon, the price of the commodity determine the quantity demanded of the commodity holding other factors constant.

Note: Price of the commodity is the predictor variable while quantity of the commodity is the dependent variable.  

Case 2; If you consider the relationship between quantity demanded of a commodity and price of the commodity-the results are most likely not going to be statistically significant if the quantity demanded is the determining factor. This is because in the natural phenomenon, the quantity demanded of the commodity does not determine the price of the commodity holding other factors constant. 

Note: Quantity of the commodity is the predictor variable while Price of the commodity is the dependent variable.  

NB: That as long as in the natural and physical phenomenon the two or more variables logically (plausibility) relate to each other then there is assurance of statistically significant research findings.

Step Four: Quantify the Specific Variables

This is the litmus test step for the variables identified must be measurable. There must be a certain way of operationalizing the variable. Otherwise, if you cannot measure the variable, then it implies that they do not exist. The methodology of measuring the variables should be authentic and universally acceptable. That means you cannot come up with your own. You cannot re-invent the wheel.

Step Five: Assess the Level of Data/Information Availability for the Variables Chosen

The most challenging step to the researcher is step five. Why? This is because, the variables may be good looking and promising an obvious association or linkage. However, data availability may be a nightmare to the researcher. First things first. Before going very far with your identification of the variables of your concern, you should be sure of how to access the data or information required to back up your study. This calls for one to investigate the most appropriate method to use to get statistically good enough data or information describing the study variable(s).

If no statistically sufficient data, this may disapprove your proposition even when it is valid.

Step Six: Collect and Analyze the Data

This step entails actual data analysis using the diverse tools of analysis such as simple regression, correlation analysis. Multiple regression and other more. The analysis tool used is based on the research problem to be solved by the researcher.

NB: It is only after doing data analysis you can tell if two or more variables you proposed to be related can be proved to have had a theoretical gap.

 

Role of theoretical or conceptual knowledge gap in proposal writing

Whether the proposal being developed is for academic or for funding purposes, development of theoretical/ or conceptual knowledge gap is very important as explained below;

  1. Helps one to come up with empirical models which have theoretical foundations. Hence it will result to addition of new knowledge in the already existing body of knowledge.
  2. Management decision making purposes-Increased possibility of statistically significant research findings guide in well informed decision making by the management.
  3. Avoids the problems of using variables in a study which are not measurable and hence result to research findings which are unreliable.
  4. In academia, it makes it easy for the student to identify the research topic without much straining. As a result, this will help one to avoid repeating already researched topics or area.
  5. Help in optimally allocating resources in research endeavors by the researcher for no resources will be allocated for instance to collect data for a study variable that is not of value addition to the researcher.

Causes of conceptual gaps, which are not statistically significant

What causes studies by researchers and even academic research findings have no statistical significance? Something that is always unpleasant to many students and even professional researchers. For most of us we want to see significant relationship between or amongst study variables!
Some of the main causes are;
1.    Identifying study variables which have no plausibility. That is which have no logic in their link or in other words, they do not relate in the natural or physical phenomenon.
For example, if you regress the number of times a man bites a dog or regress the influence of quantity demanded on the price of the commodity. In this two cases, no logic hence the regression or estimation results will not be statistically significant.
2.    Lack of adequate or available data. In other words, Small or dismal sample size as compared to the whole population may be problematic. This makes the data fail the test of being a true representative of the population.
For instance, if the population is very large and the sample size is small, the regression outcome will be something similar to this empirical equation

 

 

3.    Lack of theory. If the concept or conceptual framework proposed by the researcher has no theory underpinning the relationship, then it implies that no practicality of the relationship portrayed and hence the model is bound to backfire.
For example, if the variables chosen have no practical relationship. Such as when you put a cart ahead of the oxen. Of course you know it should be the reverse.
4.    Wrong sampling methodology-if the method of collecting data is not good enough, then the analysis of data will be adversely affected and the conceptual gap will not give fruit.
For example when the researcher chooses the wrong sampling methodology which will not address the structure and characteristics of the population. Like using simple random sampling to collect data from a population which is dividend in to geographical strata.
5.    Wrong data analysis technique. If the researcher, just like in the case of sampling methodology chooses an inappropriate analysis tool, the conceptual gap will not be able to communicate any reliable results.
For instance, if one performs simple regression instead of multiple regression for in the case of simple regression, only one predictor variable is considered and this may not have a high prediction power as it could be if we considered several predictor variables. Of course we know the changes we see on an individual dependent variable is not solely determined by one predictor variable.

About the Author - Dr Geoffrey Mbuva(PhD-Finance) is a lecturer of Finance and Accountancy at Kenyatta University, Kenya. He is an enthusiast of teaching and making accounting & research tutorials for his readers.