What Is Methodology? How to Understand and Present It in a Thesis

What is methodology

Methodology is the system of perspectives, principles, and approaches that guides the entire research process. It answers several questions: How do you view the problem, what data are considered appropriate, what evidence is sufficient for a conclusion, and why did you choose that research design.

In a quantitative thesis, methodology is more than the name of a technique such as running regression, EFA, or PLS-SEM. It operates at a broader level. Methodology is what allows you to explain why your topic uses a questionnaire, measures variables with a Likert scale, collects numerical data, and tests hypotheses with SPSS or SmartPLS.

You can view the relationship in the following order:

methodology → research approach → research design → data collection method → data analysis technique.

For example, suppose you are studying the factors that influence the intention to use an e-wallet. If you choose a quantitative approach, you build a model with independent and dependent variables, operationalize concepts as items, distribute a questionnaire, code the data, and test the relationships statistically. Methodology explains the logic behind this entire chain of choices.

If this term is new to you, you can also read what is scientific research to distinguish research objectives, research questions, and methodology.

The role of methodology in quantitative research

Methodology gives your thesis a logical line from the research problem to the conclusion. When the committee asks, “Why did you choose a questionnaire survey?” or “Why did you use SmartPLS instead of another tool?”, your answer should not stop at “because previous studies did the same thing.” You need to connect that choice to the nature of the research question and the type of data you need to collect.

For quantitative research, methodology usually guides five key decisions:

1. Identify the type of research question. Are you measuring a level, testing a relationship, comparing groups, or predicting an outcome. Questions about effects between variables often fit a hypothesized model and numerical data.
2. Choose the approach. Quantitative research uses data that can be coded and analyzed statistically. Qualitative research has a different purpose and usually focuses on interpreting textual data. You can also read what is qualitative research to avoid confusing the two terms.
3. Identify the unit and research population. Questionnaire respondents must fit the scope of your topic. The research subject needs to be stated clearly, such as users of digital banking applications in Ho Chi Minh City, rather than simply “customers.”
4. Choose how to measure the concepts. An abstract concept such as satisfaction or perceived usefulness must be measured through several items. Your questionnaire needs variable codes, a scale, and wording appropriate to the context.
5. Choose the analysis technique. Cronbach's Alpha, EFA, regression, mediation testing, and PLS-SEM are not steps that can be combined arbitrarily. Each technique serves a particular question and a particular type of data structure.

Methodology also helps you set boundaries for the topic. A study based on a one-time survey conducted during a specific period is usually suitable for describing or testing relationships within the research sample. Be careful when writing an absolute causal relationship if the data design cannot support that conclusion.

What counts as an adequate methodology

There is no single number that makes a methodology “acceptable.” Readers assess consistency among the research problem, model, scale, sample, data collection method, and analysis technique. Still, quantitative research includes several technical criteria that commonly appear in the methodology and results sections.

The table below gives reference points. Read each one together with its conditions of application, rather than using it as an automatic reason to remove or retain an item.

These reference points belong to different parts of the process. Cronbach's Alpha assesses the internal consistency of a scale. KMO, Bartlett's Test, Eigenvalue, and Factor loading are usually read in EFA. AVE, CR, HTMT, and R² belong to the assessment of the measurement model or structural model in PLS-SEM.

You also need to distinguish methodology from data quality criteria. A model can meet the thresholds and still have problems if the wrong population was sampled, items were coded incorrectly, or too many responses are missing. Conversely, one result that falls below a threshold does not mean that the entire methodology is wrong.

How to read methodology in the output

Methodology usually does not appear as one table in SPSS or SmartPLS. You read it through a sequence of outputs to check whether the initial research choices are supported by the data.

In SPSS, you may see the tables Reliability Statistics, Item-Total Statistics, KMO and Bartlett's Test, Total Variance Explained, Rotated Component Matrix, and Coefficients. In SmartPLS 4, the commonly reviewed tables include Outer Loadings, Construct Reliability and Validity, Discriminant Validity, Path Coefficients, R Square, and Bootstrapping results.

The following is illustrative output, used only to show how to read the output and not to represent the results of a real study.

For example, simply seeing Path Coefficients equal to 0.321 is not enough to conclude that the hypothesis is supported. You must check the bootstrapping results, confidence interval, and p-value under the research settings. A positive coefficient may not be statistically significant.

Likewise, Outer Loadings shows the relationship between an item and a latent variable in the measurement model. It does not show whether an independent variable affects a dependent variable. These are two different questions within the methodology.

When reading the output, save each run. Record which item was removed, why it was removed, how the indices changed, and which variables remained in the final model. This record helps you explain the result when your supervisor asks why the current results table differs from the first run.

What to do when the methodology does not meet the criteria

First identify the level at which the problem occurs. It may involve the sampling design, item quality, test assumptions, or model selection. Do not begin by deleting the item with the lowest index merely to make the results table look better.

Step one, check the raw data. Open the .sav or .csv file and check whether the variables have the correct numeric type, whether the Likert scale was entered in reverse, and how missing values were coded. A variable that should run from 1 to 5 but contains 55 usually reflects a data-entry error, not a statistical characteristic of the scale.

Step two, check reverse-coded items. If the questionnaire contains an item worded in the opposite direction, recode it before running Cronbach's Alpha or EFA. In SPSS, you can use Transform > Recode into Different Variables. Keep the original variable and create a new one so that you can compare them.

Step three, review the item content and scale source. If an item has a low Corrected Item-Total Correlation, reread its wording, context, and the concept it represents. Removing an item only because Alpha increases may remove important content from the concept.

Step four, remove items for a defensible reason. You may consider removing an item with a low factor loading, high cross-loading, or unsuitable item-total correlation, but record every round. The decision should rely on both theory and statistics, rather than on one number alone.

Step five, review the model. If several items have problems at the same time, the cause may lie in how the concepts are distinguished, the sample size, or how respondents understood the questionnaire. Running the same data repeatedly cannot repair a weak measurement design.

When the data have already been collected and an important assumption is violated, discuss an alternative technique or the reporting of limitations with your supervisor. SPSS, SmartPLS, AMOS, JASP, and R use different settings. Choose a tool because it fits the model and can be explained during the defence, rather than only because one output looks better.

Distinguishing methodology from research methods

Methodology is the framework of thinking and principles that guide the research. Research methods are the specific ways you collect, process, and analyze data. In short, methodology answers “why did you choose this direction,” while research methods answer “how will you carry it out.”

For example, “questionnaire survey” is a data collection method. “Linear regression” is an analysis method. “Quantitative” is an approach. Methodology explains why these choices fit your research question.

You should also distinguish methodology from the research field and research subject. Research field tells you the area in which you are working, while the subject identifies the specific phenomenon, group, or relationship to be analyzed. These concepts are related, but they do not replace one another.

Common mistakes

Naming a technique instead of a methodology. Writing “the methodology of this study is Cronbach's Alpha and EFA” confuses the conceptual levels. These are scale-testing techniques, not the entire methodology.

Listing methods without giving reasons. A paragraph that only says “the study uses SPSS to analyze the data” does not explain the choice. You need to state who provided the data, how the variables were measured, and which output addresses which hypothesis.

Using thresholds without sources. The same index may be interpreted differently depending on the context. When writing reference points for Cronbach's Alpha, Factor loading, AVE, or HTMT, give the source in the sentence or criteria table.

Mixing CB-SEM and PLS-SEM. CFI, TLI, RMSEA, and SRMR are commonly used in CB-SEM, with the source (Hu and Bentler, 1999). Do not put these indices into a separate SmartPLS criteria section unless you have correctly explained the model and tool.

Deleting an item based on one output. An item with a low loading needs to be considered together with the scale content, cross-loading, reliability, and theoretical purpose. Repeatedly deleting items until a threshold is met can make the model difficult to explain during the defence.

Writing the methodology after running the analysis and then editing it to match. You can update the methodology when the data require an adjustment, but you need to keep a decision trail. The final thesis must be consistent across the methodology chapter, questionnaire, data file, and results chapter.

Frequently asked questions

Is methodology the same as research methods?

The two concepts are related but not identical. Methodology is the foundation that guides the study and explains the choices, while research methods are how you collect, process, and analyze the data.

In a thesis, the methods section may present the methodology, research design, sample, questionnaire, and analysis techniques together. State the role of each part clearly instead of using every term as a synonym.

What does quantitative research methodology include?

Depending on the topic, this section commonly includes the quantitative approach, model and hypotheses, operationalization of concepts, questionnaire design, population and sample size, data collection process, data cleaning, and analysis plan.

If you use SPSS, you can present Cronbach's Alpha, EFA, correlation, and regression in that order when appropriate. If you use SmartPLS, separate the assessment of the measurement model from the structural model, then report bootstrapping and the relevant indices.

Does methodology need citations?

Yes, when you state a concept, criterion, or procedure based on academic literature. For example, the Cronbach's Alpha criterion can cite (Nunnally, 1978), while the two-step measurement-then-structural approach can cite (Anderson and Gerbing, 1988).

You should also describe your own research procedure specifically, but do not attach a source to a number that the source did not propose. Check your reference list before adding a citation to the thesis.

Does methodology require a choice between qualitative and quantitative research?

This is a choice of research approach, while methodology is the broader framework that explains the choice. For a topic that aims to measure and test relationships using questionnaire data, a quantitative approach is often suitable.

Choose based on the research question, data type, and model you need to test. Do not choose quantitative research only because software makes tables easy to produce, and do not call a statistical technique a methodology.

How should I write the methodology if I only use SPSS?

You can present the quantitative approach, a cross-sectional research design if appropriate, questionnaire-based data collection, sampling method, scale, data-cleaning process, and the techniques run in SPSS.

The explanation needs to connect each technique to its purpose. For example, Cronbach's Alpha assesses scale consistency, EFA examines the factor structure, and regression tests the effects of independent variables on the dependent variable in the proposed model.

If the results do not meet the criteria, do I need to change the methodology?

Not necessarily. First check data-entry errors, reverse-coded items, missing data, sample size, and command settings. Then review the item content and the conditions of the test.

If the problem comes from the questionnaire design or the wrong population, software cannot fix the underlying issue. Record the limitation, discuss it with your supervisor, and consider a defensible option instead of repeatedly running the same file to search for attractive results.

How to write this in your thesis

Open your data file now. Write a short outline from the research question to the approach, sample, scale, and analysis technique, then compare each choice with the methodology chapter. If you need to run and explain the analysis on your own .sav or .csv file, you can use DoThesis M4 analysis.