What Is a Survey Questionnaire and How to Design a Quantitative Questionnaire

You have a research model and need to turn it into a survey questionnaire that respondents can actually complete. This step is easy to underestimate, but a questionnaire aimed at the wrong population, missing indicators, or using leading questions will make the data difficult to analyse later. The workflow here follows the steps you need for a quantitative study: define the objective, determine the sample size, design the questionnaire, collect responses, and clean the file before importing it into SPSS.

What is a survey questionnaire in quantitative research

A survey questionnaire is a structured instrument for collecting responses from a group of research participants. In a quantitative study, it usually includes screening questions, sample demographics, and indicators that measure the constructs in your research model.

You can use a paper questionnaire, an online form, or an electronic file. The format is only the outer layer. What matters more is that each item serves a specific research objective and that the exported data has a structure suitable for SPSS or SmartPLS.

For example, if your model includes the construct “continuance intention,” you should not ask one question and treat it as a complete scale. A construct is usually measured with multiple indicators coded as YD1, YD2, and YD3. Respondents rate each indicator on a Likert scale, usually from strongly disagree to strongly agree. The Likert scale originated from Likert’s attitude measurement technique (Likert, 1932).

Survey questionnaire and questionnaire are different labels, but in a thesis they are often used almost interchangeably. You can also look at this survey questionnaire template to understand the layout, then adjust the content to your own model, target population, and research context.

How to determine the sample size for a survey questionnaire

Sample size is the number of valid responses you need for your analysis. Determine this number before distributing the questionnaire and allow for invalid responses. Do not use the final number of people who responded as your only justification after data collection is complete.

If your study uses multiple indicators and you plan to run EFA, a commonly cited rule is 5 to 10 observations for each indicator (Hair et al., 2010). For example, a questionnaire with 30 indicators would produce a reference range of 150 to 300 responses under this rule. This is a planning guideline, not an automatic guarantee that your data will suit every analysis.

If you run regression with m independent variables, you can refer to the sample-size requirement of 50 + 8m (Tabachnick and Fidell, 2013). These two approaches may produce different results. Choose the approach that fits the main analysis in your study and explain the choice clearly in the methodology chapter.

For a finite population with a known size N, you can present the Yamane formula as follows:

n = N / (1 + N × e²)

Here, n is the required sample size, N is the population size, and e is the acceptable error. This formula is commonly used in survey theses with a defined population (Yamane, 1967). If the population size is unknown, do not attach this formula to your study simply because it appears in many sample papers.

Aim to collect more than the expected minimum because some responses may be excluded for missing data, unusually fast completion, or selecting the same option throughout the questionnaire. A larger number cannot compensate for choosing the wrong participants. One thousand responses from people outside your target population still leave you with a sampling problem.

Design the questionnaire around the research objective

How to design a survey questionnaire should begin with the model and hypotheses, not with opening Google Forms. Make a list of the constructs in the model, the proposed indicators, the scale sources, and the types of questions you need. If you have not clearly separated independent, dependent, and control variables, the questionnaire can become long without becoming focused.

Write the introduction and screening questions

The opening section should briefly state the survey purpose, the eligible participants, the expected completion time, and how the data will be used. Commit only to what you can actually control, such as using the data for research purposes and reporting the results in aggregate.

Screening questions establish whether a respondent belongs to the target population. For example, if your study concerns people who have purchased from a platform, the first question might ask whether the respondent has ever purchased from that platform. Someone who selects “never” should be directed to the end of the questionnaire or to the appropriate section.

Do not use a screening question to make respondents feel that they should choose the answer most helpful to your research. The question should reflect the sample criteria stated in your methodology, such as age, location, usage experience, or the time when the behaviour occurred.

Choose a scale for the indicators

For constructs such as perceived usefulness, service quality, satisfaction, or usage intention, you will usually use multiple indicators and a Likert scale. A 5-point Likert scale works well when you want a questionnaire that is short, easy to answer, and includes a clear neutral option. A 7-point scale can provide finer distinctions, but respondents also need to understand the difference between adjacent points. Review the 7-point Likert scale before choosing.

A common set of labels for a 5-point scale is: 1, strongly disagree; 2, disagree; 3, undecided or neutral; 4, agree; and 5, strongly agree. Keep the interpretation direction consistent across indicators. If you include a reverse-coded item, note it clearly in the data file so that you can recode it before running Cronbach's Alpha.

Do not rewrite the original scale in a way that changes its meaning. Read the construct definition, the scale context, and the wording of each indicator. The questionnaire must be easy for respondents to understand, but it must still preserve the original measurement content. This section on questionnaire design can serve as a separate checklist for this step.

Arrange questions in a logical response flow

A practical layout usually has four sections: introduction and screening, measurement items for the constructs in the model, demographic information, and a thank-you message. Put general and easy questions first, then move to groups of indicators. Questions about income or personal information should come last if the study genuinely requires them.

Each indicator should have its own code in the file, such as DV1, DV2, and DV3 for one construct. Do not combine two ideas in a single statement such as “This application is easy to use and useful,” because a respondent may agree with one part but not the other.

Pilot the questionnaire with several people from the target population before distributing it widely. The purpose is to identify unclear wording, duplicate items, page-transition errors, and unusual completion times. If you revise the questionnaire after the pilot, save the new version and record which variables changed.

Implement data collection

When distributing the questionnaire, follow the sample criteria stated in chapter 3. Record when the survey opened and closed, which channels you used, how many responses you received, and how many were excluded. These details allow you to explain the process instead of presenting only one final number.

You can distribute the questionnaire through class groups, professional communities, customer lists that you are permitted to access, or networks that match the target population. The most convenient channel is not necessarily the correct one. If your study concerns company employees but you distribute the questionnaire only to students, the results will not represent the target population even if the questionnaire itself is well designed.

DoThesis has fillform.info, our own form tool for creating and collecting responses to Vietnamese questionnaires. It is useful when you need to manage the survey link and returned data in a separate workflow. Google Forms is the familiar alternative, especially when you need to share a form quickly and export the data to Google Sheets. Whichever tool you use, check the column structure before importing the file into SPSS.

Before sending the link, complete the entire questionnaire yourself on both a phone and a computer. Check required questions, page logic, the “other” option, date formats, and file export. A small error in a required field can cause an eligible respondent to abandon the questionnaire halfway through.

You should also standardize the wording of the invitation. State who should respond, how long the questionnaire takes, and what experience participants need to have. Do not send a generic invitation to every group when the study has specific sampling criteria.

Clean the data before importing it into SPSS

The file exported from a survey tool is usually not ready for analysis. First, keep an untouched original copy. Then create a separate cleaned version and record every decision to remove a row or variable. This allows you to check the process when your supervisor asks why the number of responses decreased.

Remove ineligible responses

Remove rows that fail the screening question, fall outside the target population, or leave the main measurement section incomplete. If a respondent skips an optional demographic item but completes the indicators, do not automatically remove the entire row unless your research criteria require it.

Check the response ID, start time, completion time, and completion status if the survey tool exports these columns. A record that was abandoned halfway through should be reviewed separately rather than combined with completed responses.

Check straight-lining

Straight-lining occurs when a respondent selects the same rating for nearly all indicators. This is a reason to investigate the row, but identical answers alone do not justify immediate removal. A respondent may genuinely agree with several statements.

Review completion time, responses to reverse-coded items, and any attention-check question you included. If a row was completed extremely quickly, uses one rating from beginning to end, and conflicts with the screening response, the reason for exclusion is more defensible.

Standardize columns before importing into SPSS

Each row should represent one respondent, and each column should represent one variable. Column names should not contain special characters, long spaces, or the full question wording. Codes such as GT, TUOI, DV1, and DV2 are easier to manage when you set up Variable View in SPSS.

Remove technical columns that you will not use in the analysis, such as access-session codes or timestamps if you do not need them for quality checks. Do not delete original columns before saving a backup. Text responses such as “Male” and “Female” can be coded as numbers using a codebook, but record the meaning of every code.

For reverse-coded variables, determine the recoding formula before running the scale analysis. If the scale runs from 1 to 5, the new value is usually calculated by reversing the two ends of the scale. Check the value labels after recoding, because reversing the wrong variable can reduce the item-total correlation and affect Cronbach's Alpha.

How to present the survey questionnaire in chapter 3

In chapter 3, present the survey questionnaire as part of the research methodology, not merely as a form link. The reader needs to know where the questionnaire came from, which sections it contains, which scale it uses, and which criteria governed data collection.

You can describe the following in order: the target population, the sampling method, the collection period and channels, the questionnaire structure, the scales used, the coding procedure, and the criteria for excluding responses. If you adjusted wording from a referenced scale, explain that the adjustment was made at the wording level for the research context.

How to write this in your thesis

You can adapt the paragraph as follows: “The survey questionnaire was developed to measure [number] constructs in the model, including [names of variables]. The questionnaire consisted of [number of sections] sections and used a [5/7]-point Likert scale ranging from [lowest point] to [highest point]. After excluding responses that did not meet [criterion], the study obtained [number] valid observations for analysis.”

Create a separate appendix containing the full questionnaire, including the introduction, screening questions, indicators, and classification information. In the scale description table, record the variable code, Vietnamese wording, reference source, and measurement method. This survey questionnaire can help you check whether the appendix contains all the required sections.

If you exclude many responses after data collection, the figures in chapter 3 or chapter 4 must remain consistent. The number of questionnaires distributed, responses received, excluded responses, and valid responses must reconcile. Do not report one number in the methodology section and a different number in the sample description table.

Common survey questionnaire mistakes

The first mistake is adding too many questions unrelated to the hypotheses. Each indicator needs a reason for being included. A long questionnaire increases the likelihood of abandonment and inattentive responses, while adding more questions does not automatically produce a better model.

The second mistake is using leading questions, such as describing a product as “convenient and modern” before asking respondents to evaluate it. The introduction should remain neutral to reduce its influence on the answers.

The third mistake is mixing several populations in one sample without a grouping criterion. Before distributing the questionnaire, you should be able to answer three questions: who is being surveyed, what experience must they have, and which responses count as valid.

The fourth mistake is changing the scale between construct groups without explaining why. If one group uses 1 to 5 and another uses 1 to 7, you need a clear methodological reason. In most student theses, keeping one scale type reduces respondent confusion.

The final mistake is importing data directly into SPSS without a codebook. Save the variable names, variable labels, value labels, reverse-coded variables, and sample-exclusion rules in a separate file. When you rerun the analysis, you will know exactly what changed.

Frequently asked questions

What is a survey questionnaire, and is it different from a questionnaire?

A survey questionnaire is a form used to collect responses from research participants. “Questionnaire” is often used with the same meaning, but a survey questionnaire may also include instructions, screening questions, and classification information in addition to measurement items.

How many questions should a survey questionnaire contain?

There is no fixed number that applies to every topic. Base the length on the number of constructs in the model, the number of indicators for each scale, and the screening questions you need. The questionnaire must contain enough items to measure the constructs, but it should not include questions merely to make it longer.

Should I use Google Forms or another tool to create the survey questionnaire?

Google Forms is suitable when you need to create and share a form quickly. DoThesis's fillform.info is suitable for a Vietnamese questionnaire collection workflow with separate data management. Whichever tool you choose, check the final export, variable codes, and incomplete responses.

What sample size is adequate for a survey questionnaire?

The required sample size depends on the analysis, the number of indicators, the number of independent variables, and the population size. You can refer to 5 to 10 observations for each indicator (Hair et al., 2010), or the 50 + 8m formula for regression (Tabachnick and Fidell, 2013). Choose the basis that fits your design and state it before collecting data.

Do I need to remove responses that use the same rating throughout the questionnaire?

Do not remove a response solely because one person selected the same rating throughout. Also check completion time, reverse-coded items, screening questions, and completion status. Exclude a response only when there is a clear data-quality basis, and record the rule you applied.

Open your data file now, create a codebook for each column, check the sample criteria, and export one test version of the questionnaire before distributing it widely. If you need to run the analysis on your own .sav or .csv file, you can use M4 data analysis by DoThesis.