What Is Theory in Quantitative Research?

What Is a Theoretical Model?

You may be looking at your thesis file and seeing a requirement to “build the theoretical basis” or “present the theoretical framework,” without knowing where to start. In quantitative research, a theory is a system of concepts and proposed relationships used to explain a phenomenon. It helps you answer three questions: why variable X might affect variable Y, in which direction the effect may operate, and how that relationship will be tested with data.

For example, the Technology Acceptance Model, usually called TAM, explains technology acceptance through concepts such as Perceived Usefulness and Perceived Ease of Use. TAM is associated with the work of Davis (1989). The Theory of Planned Behavior, or TPB, explains behavior through attitude, subjective norm, and perceived behavioral control (Ajzen, 1991). UTAUT extends the explanation of technology acceptance behavior by incorporating more factors (Venkatesh et al., 2003).

When you read a theory, separate the information into three layers. The first layer is the concepts, such as “perceived usefulness” or “intention to use.” The second layer is the definition of each concept. The third layer is the set of relationships the theory proposes between those concepts. You then convert these three layers into a research model, hypotheses, and measurement scales.

You can also read what a model is to distinguish a system of concepts from the model used in a specific study. The two terms are closely related, but you should not use them interchangeably throughout the thesis.

Components of a Model

A theoretical framework in a thesis is usually more than a paragraph summarizing a theory. You need to identify the variables, their roles, and the relationships you expect to find. The table below shows how to read a basic quantitative model.

The following is illustrative output.

You identify independent and dependent variables from the research question, not only from the name of the theory. A concept can be an independent variable in one study and a mediator in another. When you draw the model, state the role of each variable in your own study clearly.

A mediator answers the question, “Through what mechanism does the effect occur?” For example, service quality may increase satisfaction, and satisfaction may then increase the intention to return. A moderator answers the question, “When, or for which group, is the effect stronger?” This is why you should first distinguish what a hypothesis is before writing the model section.

Each variable in your concept table should have an operational definition. This definition states which observable manifestations you will use to measure the variable. For example, trust may be measured through the extent to which respondents believe that a platform protects their information, keeps its commitments, and handles transactions reliably. If the definition is too broad, your choice of items and interpretation of results will become inconsistent.

The Original Model and Common Extensions

The original model is the initial version proposed by an author to explain a phenomenon. Present the original model before discussing adjustments for a particular national context or industry. This lets the reader see which variables belong to the original foundation and which ones you have adopted from later research.

The TAM proposed by Davis (1989) focuses on Perceived Usefulness, Perceived Ease of Use, attitude, and intention to use. When applying TAM to electronic payments, you may add Perceived Security or Trust if your topic has a theoretical basis and suitable scales. Any added variable needs to connect to a specific research gap, such as a different context, user group, or type of technology.

The TPB proposed by Ajzen (1991) includes attitude toward the behavior, subjective norm, perceived behavioral control, and behavioral intention. A study of green product purchase intention may use TPB as its foundation and then add Environmental Concern. You need to explain this addition with verified literature rather than adding the variable simply because it appears in many other papers.

UTAUT by Venkatesh et al. (2003) explains technology use intention and behavior through factors such as Performance Expectancy, Effort Expectancy, Social Influence, and Facilitating Conditions. UTAUT2 extends the context to consumers and adds factors related to hedonic motivation, price value, and habit (Venkatesh et al., 2012).

The point to remember is that extending a model does not mean adding as many variables as possible. Each new variable increases the number of hypotheses and items, the sample-size requirement, and the amount you need to explain in Chapter 4. If you are choosing between several foundations, the Research Model page can help you see the difference between a foundational theory and a proposed model more clearly.

For models involving an industry or competitive environment, Porter's Five Forces is an analytical framework often used to examine pressure from competitors, suppliers, customers, substitute products, and potential entrants. You can read more about Porter's Five Forces, but check its fit before turning the framework's factors into survey scales.

Common Scales for Each Concept

Theory tells you which concept needs to be measured. A scale tells you how to measure that concept with specific items. When selecting a scale, prioritize its original source, the context in which it was used, the number of items, and its fit with your respondents.

The source names in the table are starting points, not confirmed translations for every industry. Read the original items, translate them for your context, have the wording checked, and pilot the questionnaire before official data collection. A word-for-word translation that sounds unnatural to respondents can still produce poor-quality data.

After collecting data, the process usually includes reliability testing, EFA, or measurement-model assessment in SmartPLS, depending on your research design. Cronbach's Alpha from 0.7 upward is generally considered acceptable (Nunnally, 1978). Corrected Item-Total Correlation from 0.3 upward is commonly used when reviewing items (Nunnally and Bernstein, 1994). For a new or exploratory scale, Alpha from 0.6 may be considered depending on the context (Hair et al., 2010).

If you use PLS-SEM, you usually examine Outer Loadings, Composite Reliability, AVE, and HTMT. An outer loading from 0.7 upward is a commonly referenced threshold (Chin, 1998). CR from 0.7 and AVE from 0.5 upward are used to assess reliability and convergent validity (Fornell and Larcker, 1981). HTMT should be below 0.85, or below 0.90 for theoretically close concepts (Henseler et al., 2015).

Applying Theory to Your Study

You can turn a theory into a model in five steps. First, write the research question in the language of your topic. For example: “Which factors affect university students' intention to use e-wallets in Ho Chi Minh City?” Then choose a foundational theory that can explain the intention to use technology.

Next, make a list of the concepts from the theory. For this example, you might choose Perceived Usefulness, Perceived Ease of Use, Trust, and Intention to Use. Do not add a variable to the model merely because you found it in a reference paper. Write down, beside each variable, why it fits your context and how you plan to measure it.

The third step is to draw arrows between the variables. The arrows should reflect an argument, rather than simply creating an attractive diagram. For example, users who consider an e-wallet useful may develop a stronger intention to use it. This relationship can be converted into a quantitative hypothesis.

The fourth step is to write each hypothesis as a complete sentence. For an illustrative model with four hypotheses, you could present them as follows:

• H1: Perceived Usefulness has a positive effect on university students' Intention to Use e-wallets.
• H2: Perceived Ease of Use has a positive effect on Perceived Usefulness.
• H3: Perceived Ease of Use has a positive effect on university students' Intention to Use e-wallets.
• H4: Trust has a positive effect on university students' Intention to Use e-wallets.

This is an illustrative model and set of hypotheses, not the result of a specific study. For your actual thesis, replace the population, context, and concepts with the content in your own model. You also need to identify the theoretical or foundational research supporting each relationship instead of simply stating that one variable “has an effect” without explaining the mechanism.

The final step is to create a table connecting theory to data. For example, Perceived Usefulness is a latent variable, PU1 through PU4 are items, a 5-point Likert scale records the responses, and Intention to Use is the outcome variable. Each row in the questionnaire needs a stable variable code so that, when you enter the .sav or .csv file, you still know which concept each variable belongs to.

How to write this in your thesis

You can use the following template: “Based on [the name of the theory] by [author, year], this study proposes that [independent variable] has a [positive/negative] effect on [dependent variable] in the context of [population and location]. Therefore, hypothesis H[ number ] is proposed: [hypothesis statement].” Replace every bracketed section with real information and check it against your model diagram.

How to Test the Model in SPSS or SmartPLS

SPSS is suitable when your model uses linear regression, mean comparisons, EFA, or analyses based on composite variables. After checking the scale, you can calculate the mean score for each concept and run regression through Analyze > Regression > Linear. The Coefficients table reports the coefficient, t, and Sig., while Model Summary reports R Square.

If you use SPSS to check the scale, the usual path is Analyze > Scale > Reliability Analysis. Select the items belonging to the same concept, click Statistics, and select Scale if item deleted to see Alpha if each item is removed. For EFA, use Analyze > Dimension Reduction > Factor, then check KMO and Bartlett's Test, Total Variance Explained, and Rotated Component Matrix.

SmartPLS is suitable when your model contains latent variables and several path relationships, especially when you want to assess the measurement model and structural model within the same process. In SmartPLS 4, create a project, import the .csv file, drag the items into constructs, connect the constructs according to the hypotheses, and run PLS-SEM Algorithm. Then run Bootstrapping to examine the significance of the path coefficient.

When reporting PLS-SEM, assess the measurement model first and interpret the structural model afterward, following the two-step approach (Anderson and Gerbing, 1988). You can check Outer Loadings, CR, AVE, HTMT, VIF, R², f², and Q². VIF below 5 is a commonly reported threshold in PLS-SEM (Hair et al., 2019). Q² greater than 0 indicates predictive relevance for the assessed construct (Stone, 1974).

For bootstrapping, state the number of subsamples and the confidence interval clearly. A total of 5,000 bootstrap subsamples is a recommended choice in current PLS-SEM guidance (Hair et al., 2022). When the confidence interval for the indirect effect does not contain 0, the indirect effect is considered significant under the bootstrapping approach (Preacher and Hayes, 2008).

You should not place CFI, TLI, or RMSEA in a SmartPLS assessment table simply because you saw them in SEM literature. These fit indices belong to the CB-SEM context, with commonly cited thresholds from Hu and Bentler (1999). If you run AMOS, that is a different procedure and needs to be reported according to CB-SEM logic.

Common Errors When Using This Model

The first error is confusing a theory with a topic. “A study of factors affecting purchase intention” is a research direction, not a theory. A theory should give you a system of concepts and an argument for explaining those factors.

The second error is using several theories without identifying which one is the foundation. You can combine TAM and TPB, but state clearly what concept or relationship each theory contributes. If you only list model names in Chapter 2, the committee may ask why your model is not duplicative or internally inconsistent.

The third error is adding variables after collecting the data. Once the model, questionnaire, and hypotheses have been approved, adding a variable simply because a p-value is not significant changes the original design. If an adjustment is needed, record the reason, discuss it with your supervisor, and distinguish exploratory analysis from testing the registered hypotheses.

The fourth error is using a scale that does not have the same meaning as the concept. “Service quality,” “satisfaction,” and “perceived value” may be related, but they are not the same variable. Read the original definitions, check every item, and avoid combining concepts merely because they are correlated.

The fifth error is writing hypotheses after seeing the results. The data may show a negative or nonsignificant relationship, but you still need to report the results against the model you proposed. Do not redraw an arrow or change H1 simply to make the findings look better.

You should also distinguish “assumption” from “hypothesis.” Assumption or hypothesis explains this common wording problem in a thesis. In quantitative research, a hypothesis is a statement about a relationship that can be tested with data.

Frequently asked questions

What is theory in scientific research?

A theory is a system of concepts, definitions, and relationships used to explain a phenomenon. In a quantitative thesis, theory helps you choose variables, build a model, write hypotheses, and find suitable scales.

What is a research theoretical framework?

A theoretical framework is a structured presentation of the foundational theory and concepts related to your topic. It usually leads to a research model in which variables and arrows represent the relationships you will test.

How many theories should a study use?

There is no single number for every study. One suitable theory explained clearly is usually more valuable than several theories that are only listed. If you combine theories, identify the distinct role of each one and explain why their concepts can appear together in one model.

How are theory and a research model different?

A theory is a broader explanatory foundation, usually developed for a group of phenomena. A research model is a version specified for your study, containing the selected variables, relationships, and hypotheses to be tested.

Can I add variables to a theoretical model myself?

You can extend a model when you have a suitable argument and supporting literature. An added variable needs a clear definition, an appropriate scale, an explained relationship, and a sufficient sample size for testing. Do not add a variable merely to increase the number of hypotheses or because the initial results are not significant.

Should I use SPSS or SmartPLS to test the model?

SPSS is suitable for EFA, regression, and tests using items or composite scores. SmartPLS is suitable for models with latent variables, a measurement model, and a structural model. Choose the tool based on your model, scale type, and department's methodological requirements, not simply because one software package produces more output tables.

Open your own literature and data files first. Create a table containing the foundational theory, concepts, items, scale sources, and hypotheses H1 through H4, then compare every row with the model on paper. If you need to run the analysis on your own .sav or .csv file, DoThesis M4 analysis supports this step.