How to Write a Doctoral Dissertation from Proposal to Complete Manuscript

What is a doctoral dissertation in research

A doctoral dissertation is an independent research project in which you identify a problem with academic or practical significance, build an appropriate model and method, collect data, analyze the results, and defend a reasoned argument. In quantitative research, the results must come from real data in a .sav or .csv file, not from numbers selected because they look good.

A dissertation is usually longer than an undergraduate thesis or a master's thesis because the requirements for the research questions, theoretical foundation, method, data, and contribution are higher. The exact length depends on your university, field, and doctoral regulations. Before setting a page target, use your institution's official template and requirements as the main standard.

If you are planning your academic path, you can also see the Undergraduate thesis topics page to distinguish the scale of an undergraduate project from a doctoral dissertation. The sections below focus on turning a quantitative research proposal into a manuscript that can be checked section by section.

Standard structure of a doctoral dissertation

A quantitative dissertation usually includes an introduction, literature review and theoretical foundation, research method, results, discussion, conclusion, references, and appendices. Some universities separate the literature review from the theoretical foundation, or combine the discussion with the results chapter. Check your university's requirements before finalizing the table of contents.

These percentages only help you divide the work. They do not replace your university's page requirements. The results chapter should not become longer by copying the entire SPSS output into the main text. Keep the necessary tables in the chapter, place detailed tables in the appendix, and point to them clearly in the text.

Your reference list must match the citations in the manuscript. You can also check the guides on how to format references and how to cite references before the final formatting pass.

Steps for writing a doctoral dissertation

Step 1: Finalize the research question and scope

Start by writing the research problem in a short paragraph, then turn it into a general objective, specific objectives, and research questions. A quantitative question needs to identify which variables will be measured, which relationships will be tested, and which context is being delimited.

For example, the question “What factors affect university students' intention to use digital banking in Ho Chi Minh City” is still broad if it does not specify the group of factors, how they will be measured, and when the survey will be conducted. Finalize the scope before building the model, hypotheses, and questionnaire. Each hypothesis needs an independent variable, a dependent variable, and an expected direction of the relationship.

Do not write objectives as tasks such as “run Cronbach's Alpha” or “run regression.” Those are techniques, not research objectives. Your objectives should state what you intend to test or explain.

Step 2: Build a content matrix for each chapter

Create a separate table with four columns: the question to answer, the evidence required, the data source, and the expected location in the dissertation. For example, a question about scale reliability will use the Reliability Statistics and Item-Total Statistics tables in SPSS. A question about a hypothesis will use a regression coefficient, p-value, or path coefficient, depending on the model.

This matrix helps you find paragraphs without evidence. If a conclusion cannot be connected to a table, figure, source, or analysis result, rewrite it or remove it. It also prevents Chapter 2 from presenting one model while Chapter 4 tests another.

Step 3: Write the introduction from problem to contribution

The introduction should follow a clear line of reasoning. The first paragraph introduces the context and practical problem. The next paragraphs show what previous studies have not fully explained. You then present the objectives, questions, subject, scope, general method, and expected contribution.

At doctoral level, the contribution needs to be specific. You may contribute by extending a model, testing it in an under-researched context, clarifying relationships between variables, or providing evidence for a particular group. Avoid writing that “the study has both theoretical and practical significance” unless you explain exactly where that significance lies.

Step 4: Build the literature review, model, and hypotheses

Group the literature by concept or relationship between variables instead of summarizing each paper in publication-year order. For each group, answer three questions: what have previous studies agreed on, where do they disagree, and which gap does your dissertation address.

After the review, define the latent variables and explain the theory used to connect them. The research model must match the hypotheses and scales. If your model includes a mediator or moderator, explain its role before presenting the testing procedure.

Write each hypothesis as a complete sentence, including the direction of the relationship when the theory allows it. For example: “Perceived usefulness has a positive effect on university students' intention to use digital banking.” After each hypothesis, give the reasoning and relevant research source instead of merely placing two variable names next to each other.

Step 5: Write a method that others can check

The method chapter needs to describe the research design, sampling method, measurement instrument, data collection procedure, data cleaning, and analysis techniques. State the number of points on the Likert scale, how the items were adapted, the criteria for rejecting responses, and the final sample size.

If you run SPSS, record the procedure names and corresponding output tables. For example, reliability testing uses Analyze > Scale > Reliability Analysis. EFA uses Analyze > Dimension Reduction > Factor. Linear regression uses Analyze > Regression > Linear. If you use SmartPLS, clearly separate assessment of the measurement model from assessment of the structural model.

Do not write that “the data were processed using specialized software.” That sentence does not tell the reader what you did. The reader needs to know which variables were reverse-coded, how missing data were handled, what criteria identified an outlier, and why an observation was removed.

Step 6: Present the results according to the research questions

The results chapter should follow the analysis sequence stated in the method. You can begin with sample characteristics, then assess the scales, evaluate the model, and test the hypotheses. Every table needs a number, title, data source, and a paragraph interpreting it immediately afterward.

When reporting Cronbach's Alpha, state the initial number of items, which item was removed if applicable, the final Alpha value, and the corrected item-total correlation. When reporting regression, present R², the F test, coefficients, p-value, and VIF if those indicators belong to your model.

The table below is illustrative output, not the result of a real study. The numbers only show how a results paragraph can be organized.

An appropriate interpretation is: “In the illustrative output, trust has the largest Beta coefficient among the three variables and a Sig. value below 0,05. This result supports the hypothesis concerning the effect of trust in the illustrative model.” In an actual dissertation, replace every number with the output from your own analysis.

Step 7: Write the discussion by explaining rather than repeating the table

The discussion answers what the results mean. Compare your findings with the studies reviewed earlier, identify similarities or differences, and explain how the context, sample, measurement, or survey period may have contributed to them.

A strong discussion paragraph usually has four parts: the main finding, comparison with previous literature, an explanation, and the implication for theory or practice. Do not turn the discussion into a list of coefficients. The numbers belong in the results chapter. This chapter needs to clarify the story behind those numbers.

Step 8: Write the conclusion, contributions, and limitations

The conclusion should answer the objectives and research questions in order. Then separate the theoretical contribution, any methodological contribution, and the managerial implications. Each implication should connect to a specific result. If a variable has a weak effect, avoid proposing too many solutions based on it.

The limitations section should be honest about the sample scope, cross-sectional design, sampling method, generalizability, and variables omitted from the model. A useful limitation is one that can be explained through the research design, rather than a general statement such as “time and resources were limited.”

Step 9: Cross-check everything before sending it to your supervisor

Check four layers. The first is content: do the objectives, questions, hypotheses, and conclusion match. The second is data: does the sample size in the method equal the number of observations in the output, and are the numbers of items consistent across each round.

The third is citations: every source in the text must appear in the reference list, and every source in the reference list must be cited. The final layer is formatting: check table titles, chapter numbering, table of contents, appendices, and data files. If your university requires a similarity check, you can review the Turnitin check procedure, but fix problems in the argument and citations instead of only trying to reduce the percentage.

Reusable sentence templates

The sentences below are frameworks that you can replace with real information. Do not keep a template unchanged if it does not reflect your actual design and results.

How to write this in your thesis can be adapted as follows: “The analysis shows that [variable X] has a [positive/negative] effect on [variable Y], with a [Beta/path coefficient] = [value] and p-value = [value]. Therefore, hypothesis [H1] is [accepted/not accepted] within the research sample.” Use “accepted” only when your university's testing rules and interpretation guidelines allow it.

A shortened doctoral dissertation example

Suppose the study examines factors affecting university lecturers' intention to use online learning platforms. The model includes perceived usefulness, perceived ease of use, system quality, and trust. Intention to use is the dependent variable. After reviewing the theory, the study proposes four hypotheses corresponding to the four relationships.

In the method chapter, the study uses a five-point Likert questionnaire, surveys lecturers who have previously used an online learning platform, and analyzes the data with SPSS. After cleaning, the study retains [number of valid observations] valid questionnaires. Cronbach's Alpha is used to assess reliability, EFA is used to examine the factor structure, and linear regression is then used to test the hypotheses.

In the results chapter, the author first describes the sample by gender, age, seniority, and frequency of use. The author then presents Reliability Statistics, KMO and Bartlett's Test, Total Variance Explained, Rotated Component Matrix, and Coefficients. Each table includes an explanation of the relevant conditions, any item removed, and the effect on the final model.

In the discussion, if system quality has the strongest effect, the author needs to explain why this feature fits the lecturer context and compare the result with previous studies. The implications should then focus on improving stability, response speed, or platform integration instead of presenting a list of solutions unrelated to the results.

If the dissertation includes an English abstract, write it after the Vietnamese chapters have stabilized. You can review the abstract structure to check whether the objective, method, results, and contribution appear consistently.

Common errors that make the committee ask follow-up questions

The first error is that the title refers to one population while the sample was collected from another. The committee will usually ask immediately why that sample was selected and how the findings can be generalized. Define the research subject from the beginning and keep it consistent in the objectives, questionnaire, method, and conclusion.

The second error is a hypothesis appearing in Chapter 2 without a corresponding row in the results table. The reverse also happens, with a variable appearing in the output without a theoretical basis. Create a comparison table covering hypotheses, variables, scales, tests, and conclusions before closing the manuscript.

The third error is copying results without explaining them. A small p-value does not automatically answer the research question. You must state the direction and size of the relationship and whether it matches the theoretical expectation.

The fourth error is changing numbers between chapters. The sample size, number of items, number of factors, and variable names must remain the same unless you clearly explain each cleaning round or model respecification. Save each version of the data file and output instead of overwriting the old file.

The final error is using unchecked citations or relying on one source for every claim. A source must genuinely support the definition, model, or criterion you are presenting. When you are unsure about a number, return to the original source and your program's requirements instead of filling in a familiar threshold.

Frequently asked questions

How many pages is a doctoral dissertation usually?

There is no single page count for every field and university. Use the doctoral regulations, dissertation template, and supervisor's requirements as your basis. Page count is only an indicator. The quality of the argument, data, and contribution determines whether the manuscript meets the requirement.

Can I use a master's thesis structure for a doctoral dissertation?

You can use the basic structure of introduction, literature review, method, results, and conclusion, but the level of development must suit doctoral research. A dissertation needs to clarify the gap, contribution, research design, and limitations of the findings in greater depth. To compare the presentation of the adjacent degree level, see the content about master's study.

Does a quantitative doctoral dissertation have to use SPSS?

SPSS is a common choice for data cleaning, Cronbach's Alpha, EFA, correlation, and regression. You can also use SmartPLS for PLS-SEM, AMOS for CB-SEM, or JASP and R if you can use them competently and your university accepts them. The method section should state the tool, version, and procedure. Choose the software based on the analysis, not simply because its name is familiar.

How should I write up results that do not support a hypothesis?

Report the result accurately, state the coefficient and significance level, and then explain possible causes based on the context, sample, measurement, or theory. Do not remove a hypothesis only because its p-value is not significant, and do not alter the data to make the model look better. A non-supported result can still be valuable when it is presented transparently.

Should I include the entire SPSS output in the dissertation?

Keep only the tables that directly serve the research questions in the main dissertation. Detailed output can go in the appendix, with a table title and a cross-reference from the results chapter. This lets the committee check the procedure while keeping the main argument readable.

Open your data file again, match the objectives to each results table, and revise the chapter using the real numbers before sending it to your supervisor. If you need to run the analysis on your own .sav or .csv file, you can use M4 analysis for this step.