How to Write a Master's Thesis from Proposal to Final Draft

If you are looking at a Word file and do not know which chapter to start with, divide your master's thesis into sections with separate jobs. A thesis usually includes the theoretical foundation, model and hypotheses, research methods, analysis results, discussion, conclusion, and recommendations. Each section should answer a specific question and remain consistent with your research questions and the data you actually have.

This guide focuses on writing a quantitative master's thesis. You can also review the list of undergraduate thesis topics to check how to narrow a topic, or compare your structure with a doctoral dissertation if your university requires a more advanced format.

What is a master's thesis in a thesis?

A master's thesis is a research project showing that you can identify a problem, develop research questions, select a method, collect data, and explain results with evidence. In quantitative research, the core usually consists of the research model, scales, questionnaire, survey data, and statistical tests.

A thesis is not a rewritten theory chapter. The committee usually follows the logic of the whole chain: a practical problem leads to a research gap, the gap leads to objectives and hypotheses, the hypotheses lead to measured variables and a method, and the results then answer each hypothesis.

Each university has its own rules for page count, font, table numbering, and appendices. Follow the requirements of your department or supervisor first. For a typical quantitative thesis, the main text may be around 60 to 100 pages, excluding references and appendices. This is only a reference range and does not replace your university's official guidelines.

If your topic only describes a current situation, its structure will differ from a topic testing relationships between variables. Before you write, list your research questions, independent variables, dependent variables, any mediator or moderator, and the analyses you expect to run in SPSS or SmartPLS.

Standard master's thesis structure

Use the table below as an initial checklist. The page ranges are only a guide for allocating effort, not a rule that applies to every programme.

Create the headings in Word before you write the content. This makes it easier to see which sections are still empty and lets Word generate the table of contents automatically. Number important results tables continuously by chapter, such as Table 4.1 and Table 4.2, instead of naming them informally.

Use one citation style consistently from the beginning. Review the guides on how to format references and how to cite references before you build your source list. Your references must match in both directions: every source cited in the text must appear in the reference list, and every source in the reference list must be cited somewhere.

Steps for writing a master's thesis

Step 1: Finalise the problem, objectives, and research questions

Write a short paragraph describing the practical problem your study aims to explain. Then turn that problem into one general objective and several specific objectives. Your objectives should be measurable. Verbs such as assess, identify, measure, test, and analyse usually fit quantitative research better than broad verbs such as understand or explore.

Each research question should identify a variable or an outcome to observe. For example, the question “What factors affect students' intention to use digital banking?” could lead to perceived usefulness, perceived ease of use, trust, and usage intention as variables. Once the question is clear, it becomes easier to decide which questionnaire and analysis you need.

Step 2: Write the theoretical foundation and literature review

Start by defining the main concepts using academic sources. Then present a theoretical framework that fits the topic, such as TAM, TPB, or UTAUT for research on technology acceptance. Do not place every theory you find in this chapter. A theory belongs here when it helps explain a variable or a relationship in your model.

Organise previous research by topic or by relationships between variables. Each paragraph should explain what the study measured, which context it used, what it found, and what remains unexplained. Listing papers one after another without comparison can make the review longer without establishing a meaningful research gap.

At the end of the chapter, state the gap clearly. The gap may concern the context, the target group, an untested relationship, or the combination of variables in one model. It should lead directly to your model and hypotheses, rather than ending with a general statement that more research is needed.

Step 3: Build the model and hypotheses

Draw the model using consistent rectangles or ellipses. Each arrow needs a theoretical reason or evidence from previous research. If variable A affects variable B, the hypothesis should state the direction of the effect when the theory supports doing so.

For example, you could write H1: Perceived usefulness has a positive effect on students' intention to use digital banking. After each hypothesis, give a short rationale and cite the relevant source. This section should do more than describe the diagram, because the committee may ask why a particular relationship was included.

If your model includes a mediator or moderator, define that variable's role before stating the hypothesis. Do not use the two terms interchangeably. A mediator explains the mechanism through which an effect is transmitted, while a moderator changes the strength or direction of a relationship.

Step 4: Write the research methods

Describe the process from questionnaire design, scale adaptation, pilot testing, and data collection to data cleaning and analysis. For the items, create a table containing the item code, adapted wording, scale source, and the way it was translated or adjusted. Use consistent wording throughout the questionnaire and state the Likert scale clearly, such as 1 for strongly disagree and 5 for strongly agree.

Present the sampling criteria, survey period, questionnaire distribution channel, and number of valid responses. Explain why a response was removed, such as many unanswered items, the same response selected for every item, or failure to meet a screening question.

If you run SPSS, describe the order of Cronbach's Alpha, EFA, descriptive statistics, correlation, and regression according to your model. If you run SmartPLS, separate assessment of the measurement model from assessment of the structural model using the two-stage procedure. CFI, TLI, and RMSEA belong to CB-SEM, so do not include them when assessing a PLS-SEM model.

Step 5: Present the analysis results

The results chapter should follow the order of your method. First describe the sample. Then report scale reliability and validity, followed by the model results and hypothesis tests. Every table needs an introductory sentence, a clear title, and an explanatory paragraph immediately below it.

When reporting Cronbach's Alpha, do not only state whether the scale passed. State the original number of items, any item removed, the final Alpha coefficient, and the reason for removal. For EFA, report KMO, Bartlett's test, total variance explained, the number of factors, and items with low or cross-loadings.

For regression, present R², the F test, unstandardized or standardized coefficients according to your university's requirements, p-value, and the conclusion for each hypothesis. For SmartPLS, report the path coefficient, t-value or p-value, R², f², Q² if the appropriate procedure was performed, and the measurement model indicators.

The table below is illustrative output, not the result of a real study. The column structure may vary depending on the output table and software you use.

Step 6: Write the discussion, conclusion, and recommendations

The discussion explains why the results matter. For a supported hypothesis, connect the result with the theory and previous research. For an unsupported hypothesis, consider the sample context, measurement approach, respondent characteristics, and differences between your study and earlier studies. Do not alter your data just to make every hypothesis pass.

The conclusion should return to the research objectives rather than introduce a new topic. Recommendations must follow from the results. If perceived usefulness has a stronger effect than the other factors, the recommendation should address how to clarify usage benefits instead of presenting a general list of solutions for every business.

Finally, state the limitations specifically, such as convenience sampling, geographic scope, cross-sectional data, the number of items, or the measurement approach. Future research should address those limitations. You can prepare the abstract after the main content is stable, because an abstract written too early often no longer matches the final results.

Usable model sentences

Use the sentences below as frameworks and adjust them to your real data and topic. Do not place illustrative values in your thesis until you have checked them against the output.

A thesis sentence could read as follows: “The analysis shows that [independent variable] has a [positive/negative] effect on [dependent variable], with [beta/path coefficient] = [value] and p-value = [value]. Therefore, hypothesis [H] is [accepted/supported] at the [significance level] level.” Replace every bracketed section with the actual result from your file.

A useful rule is to write the result sentence first and the interpretation second. The first sentence states what the output shows. The next sentence explains how that result relates to the hypothesis or objective. This helps you avoid words such as “very large” or “very important” when you have no basis for comparison.

A shortened master's thesis example

Assume that the topic examines factors affecting students' intention to use e-wallets. The model includes perceived usefulness, perceived ease of use, trust, and usage intention. The questionnaire is built from established scales, and data are collected and cleaned before SPSS is run.

In the theoretical foundation chapter, define each concept, present technology acceptance theory, and develop the rationale for three hypotheses. In the methods chapter, describe the student sample, Likert scale, questionnaire distribution, response exclusion criteria, and analysis sequence.

In the results chapter, report the sample characteristics first, followed by Cronbach's Alpha and EFA. If one item in the trust scale is removed because its Corrected Item-Total Correlation is low, state the item code and reason clearly. Then run regression or a PLS-SEM model according to the design you selected.

The discussion could follow this logic: perceived usefulness has a positive effect on usage intention, consistent with the theory and some previous studies. Trust is not statistically significant in the current sample, which may relate to respondents already being familiar with electronic payments. This is a conditional interpretation. It does not claim a cause that the research design cannot establish.

This shortened example only illustrates how to connect the chapters. In the actual thesis, replace the variable names, scale sources, sample size, output tables, and interpretations with information from your own study. If the software reports a different result, keep the real result and revise the wording, rather than changing the numbers to fit the model sentence.

Common mistakes that make the committee ask follow-up questions

The first mistake is a mismatch between the objectives, questions, hypotheses, and results tables. Create an internal cross-check table in which each research question has a corresponding hypothesis, measured variable, test method, and result location in Chapter 4.

The second mistake is citing a source without reading the original content. The committee may ask where the scale came from, how the wording was adapted, and why it fits the target group. Record the author, year, context, number of items, and scale usage as soon as you read the source.

The third mistake is copying the entire SPSS output into the thesis. Keep the complete output in the appendix. In the results chapter, select the necessary tables, format them for readability, and explain their meaning. Keep the original output file so you can check it when questioned.

The fourth mistake is applying thresholds mechanically. A coefficient meeting a threshold does not automatically prove that the model is good, just as an unsupported hypothesis does not make the study worthless. Present the context, data limitations, and reasons for selecting the criteria.

The fifth mistake is writing the discussion before the results are final. Check the number of data rows, variable codes, reverse-coded items, valid sample size, and output version one last time. If you edited the data, keep a change log so you can explain what was done.

The final mistake is inconsistent formatting. Variable names in the model, questionnaire, SPSS file, and thesis must match. Check table numbers, figure numbers, table of contents, captions, and references before sending the document to your supervisor. You can then run a Turnitin check, but the similarity result does not replace reading the content and checking the citations.

Frequently asked questions

What are the requirements for studying for a master's degree?

Admission requirements depend on the university, discipline, programme, and intake period. Universities commonly specify a relevant degree, application documents, language requirements, work experience, or other programme-specific conditions. Check the official admission notice for the programme you plan to apply to so that you have the current criteria.

Admission requirements are also different from thesis completion requirements. After admission, you may still need to complete coursework, a proposal, research, a defence, and other academic requirements under the university's regulations.

How many pages should a master's thesis have?

There is no single page count for every programme. A department may set different ranges, formatting templates, and appendix limits. Ask your supervisor or use the latest approved thesis from your university as a reference.

Rather than extending the text artificially, make sure each chapter performs its intended job. A long methods chapter that does not clearly explain the sample, scales, and analysis procedure still needs revision, even if the total page count meets the requirement.

Is running SPSS compulsory for a master's thesis?

Not every topic requires SPSS. SPSS fits many quantitative analyses, including Cronbach's Alpha, EFA, correlation, regression, and some hypothesis tests. SmartPLS fits PLS-SEM when the model includes latent variables and requires simultaneous assessment of the measurement and structural models.

You may also encounter AMOS, JASP, or R in different programmes. Software is only a tool. The important point is that the method must fit the research question, data, and supervisor's requirements.

How should you write a hypothesis that is not supported?

Report the actual coefficient and p-value, then state that the hypothesis is not supported in the research sample. The discussion may suggest possibilities related to the context, sample, measurement, or differences from previous studies, but do not claim a cause that the data did not test.

A non-significant result still has reporting value when the research process is transparent. Changing data, changing the hypothesis after seeing the result, or keeping only favourable results increases the risk during the defence.

Should you hire a thesis-writing or SPSS-running service for a master's thesis?

Consider the cost, time, and risk of being asked directly about your data. If someone else runs the analysis, you still need to understand the data file, the meaning of each output table, the reason for selecting each test, and the interpretation of the results. A polished file that you cannot explain becomes a weakness during the defence.

You can run the analysis yourself in SPSS, SmartPLS, JASP, or R, or ask for technical support and then check the complete procedure yourself. Do not use generated numbers in place of real survey data, because a quantitative thesis must be traceable to the data file and questionnaire.

Open your thesis file today, create the headings using the structure above, build a cross-check table linking the objectives, hypotheses, and output, and write the section for which you already have reliable data. If you need to run the analysis on your .sav or .csv file, use DoThesis's M4 analysis module.