What Is a Research Question? How to Develop and Write One

What Is a Research Question in a Thesis?

A research question is the central question that your study must answer using data, analytical methods, and academic reasoning. When you open your proposal file, the topic may look clear, but you may still be unsure what to ask respondents in the questionnaire, which test to run, or which result will answer the question. That is a sign that the research question is not specific enough yet.

For example, “Factors affecting the online purchase intention of students in Ho Chi Minh City” is a topic. The corresponding research questions need to be more specific: “Which factors affect the online purchase intention of students in Ho Chi Minh City?” and “How strong is the effect of each factor on online purchase intention?”

In terms of logic, the research question connects four parts of the thesis: the practical problem, research objectives, hypotheses, and data. A good question usually identifies the respondents, context, variables under consideration, and relationship to be tested. For a quantitative topic, check immediately whether the question can be answered with a questionnaire, .sav or .csv data, Cronbach's Alpha, EFA, regression, or SmartPLS.

You also need to distinguish a research question from a questionnaire item. The research question belongs to the study, while items in the questionnaire measure the model's constructs. One research question may be answered by several items and several sets of results.

If you are developing the full proposal, you can also review the Undergraduate Thesis topic collection to check the scope of your topic before finalizing the questions.

A Standard Structure for Research Questions

A set of research questions usually includes one general question and two to four specific questions. The number does not need to be identical across fields. What matters is that each question connects to at least one objective, one group of variables, and one results section in Chapter 4.

You can develop the research questions after writing the rationale for choosing the topic, because the rationale identifies the gap or problem that needs to be addressed. Do not copy an entire paragraph from the rationale and turn it into a question. The rationale explains why the study is needed, while the research question states exactly what answer you will seek.

A quantitative research question often uses verbs such as “evaluate,” “identify,” “test,” “measure,” “analyze,” or “compare.” Avoid overly broad wording such as “fully understand” or “explore every aspect” when you have not yet identified the variables and sample.

Steps for Writing Research Questions

Step 1: Narrow the Research Problem

Write the problem in one sentence that includes the respondents, location, and time period if time matters. “Service quality affects loyalty” is still broad. “The effect of service quality on the loyalty of customers using digital banking in Hanoi” has a clearer scope.

Open your proposal file and underline three groups of information: who the respondents are, what the outcome variable is, and where the survey will take place. If one group is missing, you should not write the final question yet. A narrow scope tells you who should receive the questionnaire and what type of data you can collect within the time you have left.

Step 2: Write the General Question

The general question should express the central problem accurately without placing too many variables in one long sentence. One usable formula is: “Which factors affect [dependent variable] among [respondents] in [context]?”

For example: “Which factors affect students’ intention to continue using e-wallets in Da Nang?” This opens the way for a model with several independent variables. If you already have a specific theoretical model, you can write: “How do convenience, perceived usefulness, and perceived ease of use affect the intention to continue using e-wallets?”

Step 3: Turn the Research Objectives into Specific Questions

Each specific objective should become a corresponding question. For example, the objective “identify the factors affecting the outcome” becomes a question about the factors. The objective “measure the strength of the effects” becomes a question about the direction and size of the effects. The objective “propose managerial implications” should be included as a question only when the results will genuinely be used to develop those implications.

You can compare your draft with the article on research objectives to avoid having four objectives but only one vague question. A simple comparison table in your working file will help you find missing parts before submitting the proposal.

Step 4: Check Whether the Question Can Be Measured

Read each question and ask: “Which data column will I use to answer this question?” If the answer is one item or a clearly defined group of items, the question can probably be measured. If the question uses a concept such as “comprehensive sustainable development” but the model has no corresponding scale, you need to narrow it.

In quantitative research, a question about a relationship often leads to a regression coefficient test or a path coefficient. A question about the current situation may require descriptive statistics. A question about differences between groups may lead to an Independent-Samples T Test or One-Way ANOVA. The question should come before the SPSS menu choice, but you still need to know what type of result the question can produce.

Step 5: Compare the Questions with the Hypotheses and Method

If the question asks how X affects Y, the model should contain a path from X to Y and a corresponding hypothesis. If the question asks whether there is a difference between men and women, the method section needs a grouping variable and an appropriate test. A question with no related hypothesis, variable, or results table will usually lead to another question from the committee.

You can present the logic in this order: research question, objective, hypothesis, method, and expected result. When writing the opening section, keep the names of the variables consistent with the research questions. The opening section of an essay may be shorter, but the same principle applies when the paper uses survey data.

Step 6: Read the Questions Like a Committee Member

Remove the topic title and read each question on its own. Can the reader tell who the respondents are, where the study takes place, which variable is the outcome, and what data you will use? If you need to explain the question orally before it makes sense, it contains too many unstated ideas.

Next, mark questions that have the same meaning. The questions “Which factors have an effect?” and “How do the factors affect the outcome?” may overlap if you do not clearly distinguish identifying the factors from measuring the strength of their effects. Removing duplicate questions keeps Chapter 4 from repeating the same results table.

Usable Research Question Templates

The following sentences are frameworks that you can adapt to your actual variables and context. Do not copy a template word for word if your model, respondents, or data are different.

How to write this in your thesis: “This study is conducted to answer the following question: [general question]. To make this question specific, the study focuses on the following questions: (1) [question 1]; (2) [question 2]; and (3) [question 3].” Replace every bracketed section with content that you have checked against your objectives and model.

One point to preserve is that the subject in the question must match the sample. If you survey final-year students, do not write “Vietnamese consumers” because the conclusion would cover a wider population than your data. If the data were collected at only one university, do not write a question about every student in the city.

Short Example of Research Questions

Suppose the topic is “Factors affecting students’ intention to use online English-learning applications in Can Tho.” The illustrative model includes perceived usefulness, perceived ease of use, social influence, and trust as predictors of intention to use.

The research questions could be written as follows:

1. To what extent do students in Can Tho evaluate perceived usefulness, perceived ease of use, social influence, and trust regarding online English-learning applications?
2. Which factors affect students’ intention to use online English-learning applications in Can Tho?
3. What are the direction and strength of each factor’s effect on intention to use?
4. Is there a difference in intention to use between student groups based on Internet usage frequency or online learning experience?

From these four questions, the research objectives could respectively be to describe the sample’s evaluations, identify the factors, measure the strength of the effects, and test differences between groups. If you use only linear regression, question 4 needs to be reviewed because group comparison may require a separate test. If you use SmartPLS, you can examine outer loading, AVE, HTMT, path coefficient, and R² to answer different parts of the model.

A research question does not prove that the model is correct. It defines what you need to test. The final conclusion must be based on the actual results in the data file. The figures in the example only illustrate the structure of the questions and are not results from a completed study.

Common Problems That Lead to More Questions from the Committee

The first problem is writing a question that is too broad for the data. “Which factors affect the quality of higher education in Vietnam?” requires a very large scope and dataset. If you survey students from only one department, narrow both the respondents and the context.

The second problem is writing a question without a dependent variable. “How do technological, social, and personal factors affect the outcome?” does not say what is being affected. You need to specify intention to use, satisfaction, loyalty, or another outcome variable.

The third problem is confusing a research question with a questionnaire item. “How do you rate the application’s speed?” is an item in a questionnaire. A research question belongs at the study level, such as “How does application speed affect user satisfaction?”

The fourth problem is a mismatch between the questions, objectives, and hypotheses. You can create a checking table in Word with the columns “question,” “objective,” “hypothesis,” “variable,” and “analysis.” Any blank cell needs to be addressed before you send the draft to your supervisor.

The fifth problem is adding a question that the data cannot answer. If the questionnaire has no grouping variable, you cannot draw a conclusion about differences by income. If there is no scale for a construct, adding that construct to the questions after data collection will make the study inconsistent.

The final problem is using wording that assumes the result before the analysis, such as “Factor X has a positive effect on Y.” In the research questions, use “does it have an effect” or “in what direction does it affect the outcome?” The direction must be concluded from the coefficient and p-value. Do not write it in advance to make the model appear stronger.

Frequently asked questions

What is a research question, and how is it different from a research objective?

A research question states what the study needs to answer. A research objective states what you plan to do in order to produce that answer. For example, the question asks, “Which factors affect purchase intention?” while the objective may be “To identify the factors affecting purchase intention.”

The two parts need to correspond, but they should not be copied mechanically. A question is usually written in interrogative form, while an objective usually begins with a verb such as identify, measure, evaluate, or test.

How many research questions should a thesis have?

You can use one general question and two to four specific questions. The number depends on the model, objectives, and data. Each question should produce one result or one group of results that can be presented in the thesis.

If you have eight questions but use one regression table to answer all of them, the questions have probably been split too finely or repeated. Combine questions that use the same variables and the same analytical method.

Do research questions need to be written as questions?

They should be written as questions so that the reader can immediately see the issue to be answered. Some institutions may ask for a “research problem” or a statement instead, but the content still needs to identify the respondents, variables, and relationship to be tested.

Prioritize sentence forms containing question words such as “which factors,” “to what extent,” “is there a difference,” and “how does it affect the outcome?”

Can research questions be changed after data collection?

You can adjust the wording if the original question was too broad, but you should not add a new variable or relationship simply because the available data produce an attractive result. Any change needs to be checked against the questionnaire and model and discussed with your supervisor.

If the data do not contain the required variable, software cannot repair a design problem. In that situation, narrow the question to what the dataset actually measures.

Which analysis should a quantitative research question be linked to?

A question about evaluation levels is usually linked to descriptive statistics. A question about effects may be linked to regression or a path coefficient in SmartPLS. A question about differences between groups may require a T Test or ANOVA, depending on the number of groups and the variable design.

You do not need to name the software in the question itself, but you must know which results table will answer it. This prevents you from writing a question that sounds good but cannot be run on your .sav or .csv file.

Open your proposal file, create a table linking every question to its objective, hypothesis, variable, and expected results table, then revise the blank cells. If you need help running and interpreting real data in SPSS or SmartPLS, you can use DoThesis M4 data analysis.