Research Methodology: How to Write a Quantitative Thesis
What is research methodology in a thesis
Research methodology explains how you answered your research questions and tested your hypotheses. In a quantitative thesis, this section must connect four points: the research questions, the model and scales, the data collection process, and the analysis technique you used in SPSS or SmartPLS.
You can consult the graduation thesis topics to check the scope of your topic, but when you write the methodology chapter, focus on what you actually did with your data file. The committee often asks about very specific points: Why did you choose this sample, where did the items come from, why did you use EFA or PLS-SEM, and which criteria did you use to decide whether a hypothesis was supported.
In English, the question “what is research methodology” is often answered by listing quantitative, qualitative, mixed-methods, survey, experimental, or case study research. That classification can be useful in the literature review, but it is not enough for a methodology chapter. You need to describe the specific design of your study, the unit of analysis, sample size, questionnaire, data-cleaning procedure, and testing steps.
Standard structure of a research methodology chapter
A methodology chapter is often 10 to 20 pages long, depending on your university's requirements and the complexity of the model. Length matters less than whether each section answers a question that the reader can check.
| Subsection | What to write | Reference length |
|---|---|---|
| Research design | Quantitative design, objective, unit of analysis, implementation period | 1 to 2 pages |
| Model and hypotheses | Variables, relationships, basis for developing the hypotheses | 2 to 3 pages |
| Scales and questionnaire | Scale sources, variable codes, Likert scale, language adjustments | 2 to 4 pages |
| Sample and data collection | Participants, sampling method, sample size, collection channel | 2 to 3 pages |
| Analysis procedure | Data cleaning, Cronbach's Alpha, EFA, regression, or PLS-SEM | 2 to 4 pages |
| Evaluation criteria | Thresholds and rules used to read the results | 1 to 2 pages |
This section should match your results chapter. If Chapter 4 contains a KMO and Bartlett's Test table, Chapter 3 should say that you planned to assess KMO and Bartlett's test. If you ran the model in SmartPLS, describe the measurement model, structural model, bootstrapping, R², f², Q², and path coefficient instead of placing CFI and TLI from CB-SEM in the same procedure.
You can also consult the scientific research process to arrange the order of your work. The scientific research methodology textbook PDF can support the conceptual section, but the numbers in your thesis still need to match the cited sources and the way you actually processed the data.
Steps for writing the research methodology
Step 1: Finalize the research questions and quantitative design
Open your proposal and write a paragraph answering this question: What data could help you test the problem you are studying. If your question asks about the effect of factors X on Y, a questionnaire-based survey and quantitative analysis are often suitable. If your objective is to compare means between two groups, state the grouping variable and the outcome variable clearly.
Do not write vaguely that the study uses “scientific research methods.” State clearly that your study uses a cross-sectional survey, primary data, a structured questionnaire, and statistical analysis. If you use secondary data, describe the data source, observation period, measurement unit, and treatment of missing values.
Step 2: Describe the model, variables, and hypotheses
Draw the model before you write. Each arrow in the model should correspond to a hypothesis, and each latent variable should have a group of indicators. For example, a study on the intention to use digital banking could include perceived usefulness, perceived ease of use, trust, and usage intention.
A hypothesis should be written as a sentence with a clear direction of the relationship: “Perceived usefulness has a positive effect on the intention to use digital banking.” Then explain the theoretical basis and previous research instead of simply copying the variable names. If you have a mediator or moderator, state that variable's role clearly in this chapter.
Step 3: Identify the scale sources and design the questionnaire
For each variable, create a list of codes such as PU1, PU2, PU3 or TR1, TR2, TR3. Record the scale source, the initial number of items, the translation process, and any adjustments made for the Vietnamese context. Use item or indicator when discussing a scale item, and use questionnaire for the complete data collection instrument.
If you use a Likert scale, state the number of points and the meaning of both endpoints. The Likert scale originated from Likert's technique for measuring attitudes, which you can cite as (Likert, 1932). The questionnaire should include screening questions, demographic questions, and groups of items in the same order as the variables coded in your .sav or .csv file.
Do not describe your own translation as a validated scale. You can write: “The items were translated and adapted to the study context from the original scale, then assessed preliminarily using survey data.” If you remove an item, keep a record of the questionnaire version and the reason for removal so that you can explain the decision if needed.
Step 4: Choose the participants, sampling method, and sample size
Describe the study population before discussing the number of questionnaires. For example, the population may consist of people who have used ride-hailing applications in Ho Chi Minh City during the last six months. This criterion must appear consistently in the screening question and the sample description.
Sample size needs to match the analysis technique. For regression, the rule n ≥ 50 + 8m, where m is the number of independent variables, is stated in (Tabachnick and Fidell, 2013). For a questionnaire with many items, the rule of 5 to 10 observations per item can be consulted in (Hair et al., 2010). This is a planning basis, not a reason to retain invalid responses.
If the population size is known and finite, you can present the sample-size formula from (Yamane, 1967). State the parameters you used, the expected margin of error, and the number of questionnaires distributed. Report the final number of valid responses separately from the number of questionnaires collected.
Step 5: Collect and clean the data
Record the survey period, survey channel, participation conditions, and response-exclusion rules. The checks should include questionnaires with many unanswered sections, unusually short response times, the same response repeated across all items, and responses that did not pass the screening question.
After cleaning, keep one raw-data file and one analysis-data file. Use short variable names, do not use Vietnamese diacritics in column names, and code responses according to one consistent rule. If you have reverse-coded items, reverse the codes before running Cronbach's Alpha. A reverse-coding error can change the direction of item correlations and lead you to remove the wrong item.
Step 6: Select and describe the analysis technique
A common quantitative procedure includes descriptive statistics, scale reliability testing, EFA when required by the research design, and then regression or PLS-SEM. You can consult the separate guide on quantitative research methodology to distinguish the design section from the analysis procedure.
If you use SPSS, describe the expected output tables, including Reliability Statistics, Item-Total Statistics, KMO and Bartlett's Test, Total Variance Explained, Rotated Component Matrix, and Coefficients. If you use SmartPLS 4, describe Outer Loadings, Construct Reliability and Validity, Discriminant Validity, Path Coefficients, R Square, f Square, Q Square, and the bootstrapping results.
Step 7: Record the evaluation criteria before looking at the results
Recording the criteria first helps you avoid choosing thresholds based on the result you want. You can include the table below in Chapter 3, but adjust it to match your model and your supervisor's instructions.
| Evaluation item | Reference criterion | Source |
|---|---|---|
| Cronbach's Alpha | At least 0.7; an exploratory study may consider 0.6 or above | (Nunnally, 1978); (Hair et al., 2010) |
| Corrected Item-Total Correlation | At least 0.3 | (Nunnally and Bernstein, 1994) |
| KMO and Bartlett's test | KMO at least 0.5, with Bartlett's p-value below 0.05 | (Kaiser, 1974) |
| Factor loading | At least 0.5 | (Hair et al., 2010) |
| Total variance explained | At least 50% | (Hair et al., 2010) |
| Composite Reliability and AVE | CR at least 0.7, AVE at least 0.5 | (Fornell and Larcker, 1981) |
| HTMT | Below 0.85, or below 0.90 for closely related concepts | (Henseler et al., 2015) |
| VIF in PLS-SEM | Below 5 | (Hair et al., 2019) |
| R² in PLS-SEM | 0.75 substantial, 0.50 moderate, 0.25 weak | (Hair et al., 2011) |
| Bootstrapping in PLS-SEM | 5,000 bootstrap subsamples used in the reference procedure | (Hair et al., 2022) |
These thresholds do not replace contextual judgment. For example, an item with a low loading should not be deleted only because of one number if deleting it removes theoretical content from the construct. Check the scale source, content meaning, and changes in the indicators after every run.
Usable sample sentences
You can adapt the following sentences directly. Replace the bracketed sections with real information from your study and do not retain the illustrative numbers.
| Location | Sample sentence |
|---|---|
| Design | “The study used a quantitative approach through a questionnaire survey to assess [the relationship between the variables] in the context of [participants and location].” |
| Participants | “The participants were [respondent characteristics] who had [screening condition] during [period].” |
| Sample size | “The minimum sample size was determined based on [rule or formula], with [number of independent variables or items] and [parameter].” |
| Scale | “The items were adapted from [scale source], translated and adjusted to fit [context], and then coded from [first code] to [last code].” |
| Analysis | “The data were processed through descriptive statistics, reliability assessment, [EFA or measurement model], and testing of [regression or structural model].” |
| Results | “The results show that [independent variable] has a [positive or negative] effect on [dependent variable], with [coefficient] = [value] and p-value = [value].” |
How to write this in your thesis
The paragraph on how to write the methodology should appear after the procedure description. It should not only list software names. For example: “The study used SPSS [version] to assess reliability and conduct EFA, followed by SmartPLS 4 to assess the measurement model and structural model. The evaluation criteria were determined in advance based on [source], and the results were reported using valid data from [number] observations.”
Short research methodology example
Suppose you are studying “Factors affecting students' intention to continue using e-wallets in Hanoi.” The model includes four independent variables: perceived usefulness, perceived ease of use, trust, and social influence. The dependent variable is continuance intention. Each construct is measured by three to four items on a 5-point Likert scale.
You could write the shortened section as follows: “The study used a quantitative design and collected primary data through an online questionnaire from students who had used an e-wallet during the last six months. After removing questionnaires with missing data and repeated response patterns, the valid data were analyzed using descriptive statistics, Cronbach's Alpha, EFA, and multiple linear regression. Sample size was considered based on the number of independent variables and items in the model, with reference to (Tabachnick and Fidell, 2013) and (Hair et al., 2010).”
The shortened results table below is illustrative output, not a finding from a real study.
| Variable | Cronbach's Alpha | Lowest Corrected Item-Total Correlation | Illustrative conclusion |
|---|---|---|---|
| Perceived usefulness | 0.842 | 0.566 | Retain the scale |
| Perceived ease of use | 0.791 | 0.481 | Retain the scale |
| Trust | 0.866 | 0.603 | Retain the scale |
| Continuance intention | 0.824 | 0.544 | Retain the scale |
When you write Chapter 4, replace every number in the table with the actual output. Do not insert this illustrative table into the results section as if it were your data. If an item was removed, report the item code, its value before removal, the reason for removal, and the result after rerunning the analysis.
Common problems that make the committee ask follow-up questions
The first problem is writing “the study used a survey method” without saying who was surveyed, during which period, or what the valid-response criteria were. The committee will immediately ask about the study population and why the sample can represent the problem you are analyzing.
The second problem is a mismatch between the model, questionnaire, and data file. The proposal may contain the variable Trust, the questionnaire may use “safety,” and the SPSS file may call it TC. Create a codebook connecting the construct name, item code, questionnaire statement, and data column.
The third problem is citing a threshold without stating its source, or using the same threshold for every type of analysis. CFI, TLI, and RMSEA are fit indices for CB-SEM according to (Hu and Bentler, 1999), so you should not include them in a SmartPLS procedure if you did not run CB-SEM.
The fourth problem is describing the analysis in the order in which you clicked through the software without explaining the statistical purpose. Each step needs three parts: what it checks, which output is read, and how the result affects the hypothesis or scale.
The final problem is changing the data to reach a threshold. You may remove invalid responses according to the stated criteria, remove an item with theoretical and statistical justification, or choose a more suitable technique. You should not delete data rows arbitrarily because they cause the p-value to miss the threshold.
Frequently asked questions
What is research methodology in a thesis?
It is the section describing the design, data, sample, instrument, and analysis techniques used to answer the research questions. In a quantitative thesis, it should allow another reader to understand how to reproduce the process from the questionnaire to the SPSS or SmartPLS output.
What are the common research methods?
You may encounter quantitative, qualitative, mixed-methods, survey, experimental, case study, and secondary-data analysis methods. For a quantitative paper, choose the label that accurately reflects the data and technique you actually used instead of listing several methods that you did not implement.
How should you choose a research method in economics?
If your study measures relationships between factors using a questionnaire, a quantitative method with regression or PLS-SEM can often be implemented in SPSS or SmartPLS. If you use firm-level or time-series data, describe the data source, observation period, unit of analysis, and estimation model clearly.
Do you need to state the SPSS or SmartPLS version?
Yes. State the version you used, particularly because menus and exported tables can change between versions. The version does not automatically make the results better, but it helps readers understand and check your procedure.
How should you write the results if they do not meet the threshold?
First check the coding, reverse-coded items, missing data, outliers, and applicable assumptions. Then consider removing an item with a defensible basis, rerun the analysis one round at a time, and save every decision. If the result still does not meet the threshold, report the limitation of the scale or model honestly instead of changing the data.
Open your proposal and data files now, create a mapping table connecting each hypothesis, item, questionnaire statement, and analysis technique, then complete the methodology section using the structure above. If you need help running SPSS or SmartPLS on your own .sav or .csv file, you can use DoThesis M4 data analysis.