First, you need to identify the constructs that you are going to measure. These constructs need to be tightly worded (e.g., customer loyalty, trust, social capital, knowledge sharing, etc.). You do not need to include an operational definition of your construct, which would be too verbose (i.e., wordy) when used in a diagram (i.e., in your theoretical model). If you are unsure about the difference between conceptual and operational definitions of constructs, a good starting point is the article, Concepts, constructs and variables, in the Fundamentals part of Lærd Dissertation.
Next, you need to think about the role that your different constructs play. Now whilst constructs can consist of a number of variables, they will often be treated in the same (or a similar) way as variables when creating a theoretical model; that is, constructs can be thought of as playing the role of dependent, independent or moderator variables. Therefore, you need to identify what role your different constructs play (e.g., customer loyalty might act as the dependent variable, with service quality acting as an independent variable, whilst trust acts as a moderator variable). If you are unsure about these different types of variables, a good starting point is the article, Types of variables, which is part of the broader article, Concepts, constructs and variables, in the Fundamentals part of Lærd Dissertation.
When you identify what constructs you are going to measure, and whether these act as independent, moderator, or dependent variables, it is useful to include them in a table (see Table A below). We do this because the role that constructs play (i.e., an independent, moderator, or dependent variable role) influences where they are placed in the theoretical model that you create (i.e., from left to right in the diagram).
Table A: Constructs and their roles
Construct name | Independent variables | Moderator variables | Dependent variables |
Service quality | X | ||
Customer satisfaction | X | ||
Trust | X | ||
Customer loyalty | X |
Second, you need to ask yourself: What do my hypotheses tell me about how these constructs are related? This is where you are not only interested in the relationships between constructs, but also the nature of the relationships (i.e., whether they are direct or indirect/moderators; positive or negative). If you need a recap on hypotheses, see the article: Research (and null) hypotheses. Some example hypotheses for the constructs highlighted in Table A include:
H1 | There is a positive relationship between service quality and customer loyalty |
H2 | There is a positive relationship between customer satisfaction and customer loyalty |
H3a | There is a positive relationship between trust and service quality |
H3b | There is a positive relationship between trust and customer satisfaction |
H4 | Trust mediates the relationship between customer satisfaction and customer loyalty |
You can also use a table to illustrate the hypotheses, the constructs they represent, and the directionality of the hypotheses (see Table B below):
Table B: Hypotheses, their directionality, and the constructs they represent
Hypothesis | Construct #1 | Construct #2 | Directionality |
H1 | Service quality | Customer loyalty | (+) |
H2 | Customer satisfaction | Customer loyalty | (+) |
H3a | Trust | Service quality | (+) |
H3b | Trust | Customer satisfaction | (+) |
Third, you need to think about whether you want to organise your constructs in a particular way. After all, the constructs that you are going to measure can be (a) linked or grouped, or (b) discrete. Sometimes you will be studying a number of constructs that have a logical or theoretical relationship with each other (i.e., construct grouping). You may also want to show how the different constructs that you are studying are underpinned by certain theories, especially if you are trying to show how different theories overlap (i.e., theoretical grouping). To explain further:
Construct grouping
Some constructs are more complex than others. Construct grouping is a way of showing that two or more constructs are part of a broader, overarching construct. Imagine that we were interested in the relationship between construct A, organisational commitment, and construct B, employee turnover. When we look at the construct, organisational commitment (i.e., how committed employees are to the organisation that they work for), we typically do not study this as a single construct. Instead, we look at three different types of organisational commitment, which means that we look at three different constructs of organisational commitment: normative commitment, continuance commitment and affective commitment (Meyer & Allen, 1991). Whilst we may believe that each of these different types of organisational commitment has a different relationship with (i.e., effect on) employee turnover, we may choose to group them in our theoretical model in order to show that they are part of the same, overarching construct: organisational commitment. In other words, we call it construct grouping because two or more of the constructs we are interested in are part of an overarching construct.
If the approach that you are taking towards Route C: Extension requires you to add a new construct to the constructs in the main journal article, ask yourself whether the new construct is distinct from the existing constructs in the main journal article, or whether it is part of an overarching construct. For example, if the main journal article examined the relationship between normative commitment and employee turnover in a bank, and your dissertation examined the relationship between normative and affective commitment in a factory, the construct you have added in your dissertation, affective commitment, is not distinct from the construct in the main journal article, normative commitment, but is part of the overarching construct, organizational commitment. Therefore, in this case, you would group the two constructs (i.e., normative commitment and affective commitment) in your theoretical model (something that we show you how to do later in STEP FOUR: Put your diagram together).
If you are unsure whether the construct you are adding is distinct or part of an overarching construct, refer back to the review articles you used in STEP ONE: Understand the broader literature within which your main journal article and chosen route fit. After all, these review articles typically explain the various dimensions of the constructs you are interested in. To recap, a construct can be multi-dimensional, such as the example of organizational commitment we gave above, which is an overarching construct consisting of three constructs (i.e., normative, affective and continuance commitment), or one-dimensional (e.g., constructs such as weight or height). Such review articles will help you identify the role that the construct you are adding plays, whether (a) the overarching construct, (b) part of a multi-dimensional construct or (c) a distinct, one-dimensional construct.
Theoretical grouping
In quantitative research, we typically build on or test theory. These theories help to explain how different constructs are related to each other, which is reflected in the hypotheses we set and the directionality we give to our hypotheses (i.e., the assumptions we make about how one construct, such as continuance commitment, affects another, perhaps employee turnover). For example, imagine a hypothesis based on these two constructs: There is a negative relationship between continuance commitment and employee turnover (i.e., as the perceived costs of leaving the organisation increase - continuance commitment - the number of employees leaving the organisation decreases - employee turnover). Such a negative relationship between these two variables can be explained using organizational commitment theory and turnover theory. Ideally, you will know what theories explain how the different constructs in your dissertation are related, especially because you are taking on a Route #1: Replication-based dissertation, where you can benefit from the literature review of the main journal article often explaining what theories are. However, this is not always the case (i.e., not all quantitative research, especially dissertations at the undergraduate or master's level are theoretically strong, since this is often associated with 1st class dissertations). If you can identify these theories, you need to decide whether to theoretically group the constructs in your diagram (i.e., your theoretical model) or not. Theoretical grouping is simply a way of illustrating (a) how different constructs are underpinned by certain theories, and (b) how the theories and constructs you are interested in may overlap. In STEP FOUR, which follows, we show you how you can theoretical group your constructs within your theoretical model.