Scientific Research Proposal: Structure and Detailed Writing Guide

What is a scientific research proposal in a thesis

A scientific research proposal is a plan that explains what you will study, why the issue is worth studying, what data you will use, how you will analyze it, and what contribution you expect to make. Your supervisor uses this document to check whether the topic is reasonable before you design the questionnaire, collect data, and run SPSS or SmartPLS.

You can also read what is a research proposal to distinguish a research proposal from a course outline or presentation outline. In a quantitative thesis, the proposal must be specific enough for the reader to understand which variables will be measured, who will be surveyed, what each row in the data file represents, and which result will answer each research question.

A proposal is often around 8 to 20 pages, depending on your department's template, the complexity of the model, and your supervisor's requirements. This is only a reference range. A 20-page proposal that lacks a model, scales, and an analysis plan is still weaker than a 10-page proposal with clear logic.

Check feasibility from the beginning. Can you collect the data within the time you have left, can you reach the target group, is there a suitable scale for each research variable, and can the analysis method answer the research questions? If you skip these four questions, you may finish the proposal and only then discover that you cannot survey the right respondents.

Standard structure of a scientific research proposal

Your university may use different names for the sections, but a quantitative proposal usually follows the structure below. The suggested length helps you allocate effort. It is not a mandatory rule.

You can also compare how to write a thesis proposal if your department requires a specific layout. If the proposal is being used to obtain topic approval, the rationale and feasibility need to be clear. If it will guide the full thesis, the model, scales, sample, and analysis plan need more detail.

Steps for writing a scientific research proposal

Step 1: Fix the problem, respondents, and scope

Start with one sentence describing the problem using this structure: “This study examines the factors affecting [outcome] among [respondents] in [context] within [scope].” The sentence helps you see whether the topic is too broad or too vague.

For example, “the factors affecting university students' intention to use e-wallets in Ho Chi Minh City” already identifies the outcome as usage intention, the respondents as students, and the context as e-wallets. You still need to decide whether you mean students generally or students who have used an e-wallet, because these two groups would complete different questionnaires.

The unit of analysis must remain consistent from the title through the SPSS file. If each row represents one respondent, your conclusion must also refer to respondents or groups of respondents. Do not move to conclusions about all businesses when your sample consists only of customers from one location.

Step 2: Write the rationale from the problem to the gap

The rationale should move through four paragraphs. The first describes the specific context. The second states the problem that needs explanation. The third identifies what previous studies have not fully resolved in your chosen context. The final paragraph connects that gap to your proposed study.

A gap does not have to mean that “no one has studied this.” You can identify a different context, a different respondent group, a different mediator, or inconsistent findings across previous studies. Then explain why the relationship needs to be tested in your context.

Avoid opening with general statements such as “society is increasingly developing.” The reader needs to know which concrete change creates the research problem. If you do not have a reliable source for a number, describe the trend without creating a percentage.

Step 3: Turn the rationale into objectives and questions

The general objective states what the study aims to achieve. Specific objectives must connect to operations that can be checked with data, such as identifying factors, assessing effects, testing differences, or evaluating model fit.

Each specific objective should have a corresponding research question. If the objective is to “assess the effect of service quality on satisfaction,” the question could be: “Does service quality affect customer satisfaction, and how strong is the effect?” This question points toward regression or PLS-SEM.

Do not include an objective for which you have no data. If the questionnaire measures customer perceptions at one point in time, use terms such as “absolute cause,” “long-term forecast,” or “prove a causal relationship” with caution.

Step 4: Build the theoretical basis, model, and hypotheses

For each concept, record its working definition, measurement approach, and proposed scale source. This tracking table helps you detect when one variable is being used with several different meanings. Models such as TAM, TPB, or UTAUT should be selected only when they fit the research question and context.

The research model should show the independent variables, dependent variable, and, where relevant, mediator or moderator. Each arrow needs a theoretical reason and should lead to a hypothesis. For example, if “perceived usefulness” affects “usage intention,” H1 should be written as a sentence showing the direction, or at least clearly stating the relationship to be tested.

If you use SmartPLS, separate assessment of the measurement model from assessment of the structural model. This two-step approach is presented in (Anderson and Gerbing, 1988). In the proposal, you can state in advance that you will assess reliability, convergent validity, and discriminant validity before evaluating relationships between variables.

Step 5: Design the method and sample size

The methodology section must answer four questions: who will be surveyed, which instrument will be used, how many observations are needed, and which analysis sequence will be followed. State the screening criteria, collection period, questionnaire distribution method, and rule for removing invalid responses.

If you use a Likert questionnaire, state the number of scale points and the meaning of the endpoints. The Likert scale originates from the attitude measurement technique presented in (Likert, 1932). Code items in advance, such as PU1, PU2, and PU3, so you do not have to rename columns halfway through data entry in SPSS.

For regression, the minimum sample-size rule of n equal to 50 + 8m, where m is the number of independent variables, is commonly cited from (Tabachnick and Fidell, 2013). For studies with many items, the rule of 5 to 10 observations per item is also commonly used (Hair et al., 2010). These rules support planning. They do not replace checking sample quality.

You can write the analysis sequence as follows: clean the data, produce descriptive statistics, assess scale reliability, run EFA where appropriate, and then run regression or PLS-SEM. For SPSS, name outputs such as Reliability Statistics, KMO and Bartlett's Test, Rotated Component Matrix, and Coefficients. For SmartPLS, name Outer Loadings, Construct Reliability and Validity, Discriminant Validity, and Path Coefficients.

Step 6: Prepare the schedule and check feasibility

Divide the schedule by concrete deliverables: finalize the model, complete the questionnaire, test the data, collect valid responses, clean the file, run the analysis, and write each chapter. Each milestone should have a start date, end date, and completion condition.

Reserve time for revising the questionnaire or removing responses. EFA rarely produces the expected structure on the first run. Save each file version and record why an item was removed instead of deleting it directly and later forgetting what you did.

Usable model sentences

The sentences below are frameworks for replacing the bracketed content. Do not copy them word for word if your context, respondents, or method differs from the proposed study.

How to write this in your thesis

You can adapt the following paragraph directly: “This study uses a quantitative method to test the effects of [independent variables] on [dependent variable] among [respondents and location]. Data will be collected through a questionnaire and analyzed using [software], following these steps: [briefly list the procedure].” When you have real data, replace the bracketed content with accurate information from your study. Do not enter projected figures in the results chapter.

A shortened scientific research proposal example

Suppose you choose the topic “The factors affecting university students' intention to continue using e-wallets in Ho Chi Minh City.” This example only demonstrates how the sections connect. It is not the result of a completed study.

The research problem is that some students know about and have used e-wallets but may switch to another platform or stop using them. The general objective is to analyze the factors affecting continuance intention. The proposed independent variables are perceived usefulness, perceived ease of use, and trust. The dependent variable is continuance intention.

The research questions are: Does perceived usefulness affect continuance intention, does perceived ease of use have an effect, and how does trust affect continuance intention? H1, H2, and H3 connect the three independent variables to the dependent variable, respectively. The respondents are students who have used an e-wallet during a specified period before answering the questionnaire.

In the proposal, explain why you selected regression or PLS-SEM. If the model is simple and the variables have been combined into scores, regression may be suitable. If the model includes latent variables, several items, and mediation relationships, PLS-SEM may be considered. Do not list both methods merely to make the proposal look more complete, because the committee may ask how you chose between them.

The expected-results section should describe the type of result that will be produced, such as identifying which variables have statistically significant relationships and how much variance the model explains. Do not write that a hypothesis will definitely be accepted before collecting data.

Common mistakes that make the committee ask follow-up questions

The first mistake is a title that is too broad. “The factors affecting consumer behavior” does not identify which behavior, which consumers, or which context. Narrow the outcome, respondents, and location so that the data-collection plan can be carried out.

The second mistake is a mismatch between objectives, questions, and hypotheses. If the objective concerns the effects of three factors but the model contains five factors, the committee will ask where the actual scope lies. Create a checking table in which each objective has a corresponding question, variable, and analysis method.

The third mistake is using an unclear scale source. A variable's name in English does not prove that its scale is appropriate. Record the source, proposed number of items, translation method, and plan for adapting the scale to the Vietnamese context.

The fourth mistake is promising causal conclusions from a one-time survey. With this design, more careful wording is “test the relationship” or “assess the effect within the model.” Use the word “cause” only when the design and reasoning genuinely support it.

The fifth mistake is naming an analysis method without explaining how the results will be read. If you use PLS-SEM, mention outer loading, CR, AVE, HTMT, VIF, R², f², Q², and bootstrapping at a level appropriate to the model. CFI, TLI, RMSEA, and SRMR belong to CB-SEM and should not be included in a SmartPLS procedure. For PLS-SEM, the current procedure commonly uses 5,000 bootstrap subsamples according to (Hair et al., 2022).

The final mistake is a schedule that ignores data cleaning. Plan time to check missing responses, patterned answers, unusual values, and coding errors. Dirty data cannot be fixed simply by clicking Run again in SPSS.

You can also read what is a proposal in a thesis, what is a thesis defense slide deck, and the Undergraduate Thesis topic collection to check the connection between the proposal, topic, and later presentation.

Frequently asked questions

How many pages should a scientific research proposal have

There is no single length that applies to every university. Follow your department's template and your supervisor's instructions first. For a typical quantitative topic, the proposal may contain around 8 to 20 pages, excluding references and appendices.

If the proposal is short, keep the problem, objectives, model, hypotheses, method, sample size, and schedule. The committee needs to assess feasibility, so page count cannot replace specific information.

Does a research proposal need research results

A proposal prepared before data collection does not need actual results. You only need to state the expected types of results, the analysis tables that will be produced, and how those results will answer the research questions.

Do not enter a p-value, R², regression coefficient, or respondent count in the proposal if you do not yet have valid data. Those figures should appear after you run the actual file.

Can you change the research title after submitting the proposal

You may be able to, but the extent of the change depends on your department's rules and on whether the change affects the objectives, model, or respondents. Changing the wording of the title is usually easier to manage than changing the dependent variable or the entire sample.

If you need to make a change, prepare a comparison table showing the old title, new title, reason, affected sections, and update plan. Ask your supervisor to approve it before revising the questionnaire or collecting additional data.

Should a quantitative proposal use SPSS or SmartPLS

SPSS is suitable for descriptive statistics, Cronbach's Alpha, EFA, correlations, regression, and many common hypothesis tests. SmartPLS is suitable when the model includes latent variables measured with multiple items and you want to assess the measurement model and structural model together.

The choice must come from the model, data type, and supervisor's requirements. Software cannot repair an incorrect model, an unsuitable scale, or a sample that does not represent the target respondents.

Do you need to cite sources in a research proposal

Yes. Definitions, theories, models, scales, and the basis for selecting the sample size all need appropriate sources. The item-based sample-size rule can be cited with (Hair et al., 2010), while the basis for the regression method can be cited with (Tabachnick and Fidell, 2013).

Create your reference list as you read and check consistency between in-text citations and the final proposal bibliography. Do not wait until the submission date to search for the source behind each paragraph.

When the proposal is complete, open your planning file and check each section with three questions: which data will answer the objective, which variable measures the question, and which analysis produces the evidence? If you need to run the analysis on your own .sav or .csv file, use M4 data analysis to move from real data to result tables and interpretations that match your proposal.