This is important because an aesthetically pleasing form does not guarantee useful results. You may collect many responses yet still be unable to resolve the problem stated in your thesis. The cause is often questions that are too general, a lack of appropriate response options, or directing the survey to people to whom the topic does not apply.
In this guide, you will move from the research idea to testing the finished tool. You will find examples of questions, an introduction template, rules for creating scales, and a checklist before distributing the survey. The examples are educational material: they show how to design questions but do not constitute a validated questionnaire for any topic.
Where should you start when preparing a thesis questionnaire?
Before writing the first question, prepare a brief research concept. It should explain what problem you are addressing, which group it concerns, and what data will be needed to describe it. If you do not yet have a structured concept, start with the guide on the thesis plan and preparation for the meeting with your supervisor.
Write down the following on one page:
- Research objective: what you intend to learn, describe, or compare.
- Main research question and specific questions: what the analysis should answer.
- Study group: who can provide answers and what conditions they must meet.
- Scope: which period, location, or type of experience the questions cover.
- Analysis plan: which summaries or comparisons will allow you to answer the research questions.
- Hypotheses, if derived from the project: what predictions you want to test and on what basis you accept them.
Not every survey requires hypotheses. In research aimed at exploring a little-known phenomenon, research questions can be the starting point. Agree on the requirements with your supervisor and describe them consistently in the methodology.
Also check whether the survey will actually provide the appropriate data. It is suitable for collecting self-reports, opinions and assessments. If you want to know the actual time taken to respond to messages, data from a register may be more accurate than participants' memory. A question about the perceived impact of consultations provides information about the respondent's assessment, but it does not by itself prove that the consultations caused a better thesis outcome.
How do you link the study aim, research questions and questionnaire items?
Each question should have a purpose. It may provide information about the main research question, establish a participant's eligibility for the study or identify a characteristic needed to compare groups. If you cannot say where you will use the answer, consider whether the question is necessary.
Below is a working example about using consultations when writing a thesis. Instead of a wide table, you can prepare similar short cards for your own project.
Example 1: Using consultations
Objective: describing the use of consultations by students preparing a thesis in the current semester.
Specific research question: how often did participants take part in consultations regarding their thesis in the last 30 days?
Survey question: "How many consultations regarding your thesis did you attend in the last 30 days?".
Responses, select one: 0; 1; 2; 3 or more; I do not remember.
Analysis plan: counts and percentages for each response. The "I do not remember" option should be shown separately and not treated as no consultations.
Before the question, explain what you mean by consultation, for example, a planned individual or group meeting dedicated to the thesis. Also decide whether to include email contact. Without this, participants may count different types of activities.
Example 2: Assessing the usefulness of consultations
Specific research question: how do consultation participants assess their usefulness in planning further actions?
Survey question: "How do you rate the usefulness of the last consultation in determining the next step in working on the text?".
Responses, select one: not at all useful; not very useful; neither useful nor useless; fairly useful; very useful; I cannot assess this.
Who answers: only those who attended consultations during the specified period. Others proceed to the next part of the form.
Analysis plan: distribution of ratings among the people to whom the question applies. The denominator will not automatically be the total number of study participants.
This creates a chain: aim, research question, indicator, questionnaire item and method of analysing the answers. You will find a broader explanation of these relationships in the article on research methodology in a bachelor's or master's thesis.
Which question types should you use in a questionnaire?
Choose the question type based on the information you need. The ease of calculating percentages is useful, but it should not determine everything. A closed question with a limited list of answers may omit significant experiences of the respondents.
Single-choice questions
Use these when the respondent must indicate one answer. The categories should allow for an unambiguous choice. In a question about the number of consultations, the ranges "1-2" and "2-3" are incorrect because a person with two consultations fits into both.
In the header or instructions, write: "Select one answer". If you ask about the main difficulty, specify that you mean the most important one, not all experienced difficulties.
Questions allowing multiple selections
Use these when several answers can be true simultaneously. Example: "Which materials did you use when planning a chapter in the last month?". A participant can indicate books, academic articles, faculty guidelines, and notes from consultations.
The instructions should specify whether all matching answers can be selected or only a maximum defined number. These are two different tasks. Limiting to three selections implies selection and may change the interpretation of the results.
The options "none of the above" and "I did not use any materials" should not be selected together with specific materials. If the form does not easily exclude such a combination, split the question into a filter and a list of materials.
Semi-open questions
Combine a list of response options with an 'other, please specify' option. This is useful when you are not sure you have covered every possibility. However, plan to read the written-in responses afterwards. Some may require a new category, while others will fit existing options.
Open questions
These allow describing an experience in one's own words. Instead of a general "What do you think about writing a thesis?", you can ask: "What would most help you plan the next stage of writing your thesis?". Such an answer provides concrete material for analysis.
There is no obligation to limit open questions to the end of the questionnaire. Place them where they are logically needed. However, consider the respondent's time and the work required to code the statements.
How do you write questions that do not lead respondents?
A neutral question allows answering according to experience, even if it differs from the author's expectations. The easiest way to check this is to try giving a negative, ambiguous, or "not applicable" answer.
Remove the assumption that the problem definitely occurred
To improve: "Why does your supervisor not help you enough with writing your thesis?".
Proposal: first ask about using consultations, and then about the evaluation of the support received. A person satisfied with the collaboration should not be forced to indicate the cause of a problem they do not experience.
Separate two matters
To improve: "Were the instructions clear and did you receive them quickly?".
Proposal: prepare a separate question about the clarity of instructions and a separate one about waiting time. The clarity of information and the speed of contact are different features. A combined answer will not show which of them is associated with any potential difficulty.
Establish the period to which the answer refers
To improve: "Do you often work on the text?".
Proposal: "On how many days in the last 7 days did you work on the thesis text?". For a numerical answer, values from 0 to 7 can be allowed, along with a separate option to indicate that the participant does not remember.
Such a question measures the number of days, not the number of hours or the quality of work. If you need different information, prepare a different question. Do not add meanings to the result that the tool does not cover.
Replace a difficult concept with a concrete experience
Instead of asking about "the effectiveness of consultation interactions", ask whether after the last meeting the participant knew what task to perform next. However, first decide whether this experience is the indicator of the concept being studied. Simpler language helps in understanding the question, but it does not replace methodological justification.
How to prepare a response scale and a demographic section?
In a rating scale, each point should have a clear meaning. For the statement "After the last consultation, I knew what to do next", the following variants can be used:
- strongly disagree;
- somewhat disagree;
- neither agree nor disagree;
- somewhat agree;
- strongly agree.
"I don't know" or "I cannot assess" is not the same as the neutral midpoint of a scale. If such an option is needed, add it outside the ordered agreement scale. Treat "not applicable" similarly. Do not automatically assign a value of 3 to these responses during calculations.
A single item with responses ranging from strong disagreement to strong agreement is not yet a validated scale measuring a complex trait. If you intend to study a specific psychological construct, consider an appropriate existing instrument, its properties, and the rules for its use. Do not independently change the content and scoring, and then interpret the result according to the original norms.
In an original form, maintain a clear direction of responses. For an existing instrument, follow its instructions, including for items requiring reverse coding.
The demographic section should follow from the analysis plan. If you are comparing the experiences of bachelor's and master's students, a question about the level of study is justified. A question about the precise town or village, employer or date of birth may be unnecessary. Do not create an extensive demographic section just because you saw a similar one in someone else's thesis.
How to write the introduction to a survey and arrange the order of questions?
The introduction should explain who is conducting the research, what it concerns, whom you are inviting, and how the responses will be used. Provide the actual estimated completion time, determined after a trial of the form. Indicate that participation is voluntary and describe the data handling procedures in accordance with the actual organisation of the research.
A working template that must be completed before sharing:
I am [first name and surname], studying [subject area]. As part of my thesis, I am conducting research on [topic]. I invite people who meet the following criteria: [participation criteria]. Completing the survey takes approximately [time established in the pilot study]. Participation is voluntary. The responses will be used for [actual use of the data]. Information on data handling: [description or relevant participant information]. If you have any questions, please contact: [contact address]. Thank you for your time.
This is an outline for a message, not a ready-made notice suitable for every study. Do not automatically state that the survey is anonymous. Check whether you collect email addresses, names or combinations of detailed information that could identify a participant. Confidential data storage and anonymous response collection are different situations.
A practical questionnaire structure includes:
- introduction and information needed for an informed decision about participation;
- eligibility questions and filters;
- general questions introducing the topic;
- thematic blocks corresponding to the research problems;
- necessary demographic data;
- thank you and conclusion.
The demographic section does not always have to be at the end. If the level of study determines eligibility for the research, ask about it earlier. Do not make people who do not meet the criteria go through the entire survey.
How many questions and how many respondents are needed for a thesis?
There is no universal number of questions or responses that guarantees the validity of the research. A short questionnaire may be sufficient for a narrow objective, while a longer one may be justified for a more complex project. Merely exceeding 20 questions does not improve the value of the instrument.
When assessing length, consider not only the number of items but also the effort required to answer. Ten extensive open-ended questions may take more time than twenty simple closed questions. Conversely, questions hidden behind filters do not apply to every participant.
Sample size depends on the objective, planned analyses, the nature of the population, the sampling method, and available resources. If you want to compare groups, you need appropriate sample sizes in each of them. A large total number of responses will not solve the problem if only a few people ended up in one group.
A survey shared with friends and in an open internet group does not become representative just because it collected 100 or 200 responses. Describe how you reached the participants, who could take part, and what limitations arise from recruitment. When planning statistical tests, determine the required sample size before starting data collection, preferably with a person competent in the planned analysis.
How to conduct a pilot survey?
A pilot tests how the instrument works before the main study. Invite people similar to your intended respondents. The author's own check of the form is not enough, because the author knows what they meant even when a question is unclear.
Ask pilot participants to explain several questions in their own words. Ask how they chose their answers, whether any options were missing and which expressions caused uncertainty. Record the actual completion time. This practical stage is also recommended in the Polish 2024 guide 'Jak przygotować i przeprowadzić badanie ankietowe?' from the Statistical Office in Poznań.
Also check the technical aspects during the pilot:
- whether each option of a filtering question leads to the correct section;
- whether the form works on a phone;
- whether mandatory questions do not block people to whom they do not apply;
- whether the instructions match the number of possible selections;
- whether the responses export to a readable spreadsheet;
- whether you can prepare the planned summary using the trial data.
After making changes, re-check the corrected areas. Keep the date and label of the final version. If you change the meaning of questions or responses, do not combine the pilot results with the main research without explanation. They will not automatically be comparable measurements.
A short pilot helps detect problems, but it is not a full validation of a scale. Do not describe an original questionnaire as an instrument with confirmed validity and reliability solely because a few people found the questions understandable.
How to prepare a survey in Google Forms?
First, refine the content and logic of the questionnaire, and then transfer them to the form. Saving it in a document earlier facilitates consultation with the supervisor and version control, although it is not a condition for the validity of the research.
Choose the field type to match the task. Single choice should allow selecting one option, checkboxes several, and a text response should indeed be necessary. Do not replace all questions with extensive grids: on a small screen, it is easy to mix up a row or miss the scale description.
Check the transitions between sections. Google Forms allows directing participants to sections based on answers in single-choice questions and dropdown lists. Test every path, including the end of the survey for a person who does not meet the participation conditions.
Verify the response settings. Collecting email addresses and access restrictions affect how participants engage. Open the link intended for the respondent on a different device or in a separate browser window and check what the participant actually sees. Do not treat the mere disabling of the email field as proof of full anonymity.
Send test responses and review the spreadsheet. Ensure that you distinguish between no response, skipping after a filter, and the answer “not applicable”. Delete or clearly separate test entries before analysing the main study. Keep a copy of the questionnaire that you actually distribute.
How to plan the analysis of results before sending the questionnaire?
1. Choose how to present the answers
For each question, note provisionally how you will present the answers. For single-choice questions, this may be the number of people and the percentage; for multiple-choice questions, the number of selections for each option; and for open questions, thematic categories along with a description of how they were created.
2. Test calculations on trial data
Practise this on a small set of fictitious answers, clearly marked as a test. If a question about difficulties allows selecting several options, the sum of percentages of people may exceed 100%. If a question applies only to participants of consultations, percentages must be calculated for that group, not for everyone who fills in the questionnaire.
3. Set rules for missing answers
Also record how missing data will be handled. No answer to the question “How many times did you use consultations?” does not mean zero consultations. Meanwhile, “I do not remember” is a provided answer with a different meaning than an empty field.
4. Plan comparisons and scoring
If you plan to compare groups, ensure that you have a variable allowing you to identify them. If you want to calculate a composite score from several items, establish the justification for such a combination and the scoring rules in advance. Do not sum random questions just because each has answers marked with numbers from 1 to 5.
We described the next stage in the guide on processing research results, percentages and comments on the questionnaire. It is worth looking at it before the study to check whether the designed questions will provide the data needed in the empirical chapter.
Are you planning group comparisons or tests of associations and unsure what data you will need? Before distributing the questionnaire, you can discuss with me selecting statistical analyses for research questions. I will help determine whether the answers will allow the planned calculations to be performed, before you collect material that cannot be used for the intended analysis.
Checklist before distributing the questionnaire
Go through the following points on the final version of the form. Evaluate what the respondent sees, not an earlier draft.
- I can state the research objective and participation criteria.
- Each question is related to the objective, recruitment, or planned analysis.
- Questions do not assume an answer or the occurrence of a specific problem.
- One item addresses one issue.
- The period to which the answers refer is clearly described.
- Single-choice categories do not overlap.
- For each question, it is clear how many answers can be selected.
- A neutral rating is distinguished from lack of knowledge and “not applicable”.
- Filter questions lead to the appropriate sections.
- The demographic section covers the necessary information.
- The introduction corresponds to the actual settings and organisation of the study.
- The pilot checked question clarity, completion time, use on a phone and response export.
- After corrections, I checked the form again.
- The final version is consistent with the methodology and agreements with the supervisor.
Most common questions about questionnaires for bachelor's and master's theses
Must a survey have 20 or 30 questions?
No. The number of questions is justified by the scope of the research, not a universal limit. Remove unnecessary items, but check if the remaining ones still allow you to answer all research problems. Agree on thesis requirements with your supervisor.
Does every thesis or survey require hypotheses?
No. Whether hypotheses are appropriate depends on the objective and nature of the research. Testing relationships requires a different approach from research focused on description or exploring experiences. We adapt the approach to the project and the requirements of the relevant degree programme.
If you formulate hypotheses, plan the data and analysis that will allow you to actually test them. Do not add them solely to fill out the chapter scheme.
Can I use a ready-made survey?
Yes, if it fits the objective and the group being studied, and its terms of use permit this. Check the source, conditions of use, and method of interpretation. A questionnaire found online does not become a validated instrument just because another student has used it.
Should all questions be mandatory?
Not automatically. A mandatory question must allow for a sensible answer by the person it is directed to. If the topic does not concern them, apply a filter or an appropriate variant. Forcing a random answer increases the completeness of the sheet but reduces the usefulness of the data.
Can I modify questions after starting the research?
Changing the meaning of a question or the set of answers may make results before and after the change non-comparable. Record the scope and date of the change and agree on further action with your supervisor. It is best to catch such problems in the pilot phase.
Is an online survey always anonymous?
No. It depends on the information collected and how the study is organised. Even without a name, detailed demographic information and the content of responses may make it possible to identify a person in a small group. The survey introduction must reflect how data are actually collected.
Check the questionnaire before sending it to respondents
You have a research objective and questions, but need to prepare the questionnaire or improve its draft version? We can develop questions, answer options, instructions, and the demographic section, and check their order and relevance to the study assumptions. If the questionnaire already exists, send it for evaluation before distributing it to respondents.
If you first need to establish the objective, problems, and research method, check the service “Help with your thesis methodology chapter”. If you plan statistical tests or group comparisons, it is also worth discussing in advance the service “Statistical analysis for bachelor’s and master’s theses”, so that the questionnaire provides the data you need.
- Attach the objective, research questions, and the current version of the questionnaire.
- Describe the group of respondents and pass on the supervisor's comments.
- Provide the deadline and indicate whether the form has already been shared.
Bibliography and sources
- Urząd Statystyczny w Poznaniu, Wielkopolski Ośrodek Badań Regionalnych (2024). Jak przygotować i przeprowadzić badanie ankietowe? Compiled by: Magdalena Bryza, Wanda Nowara, Arleta Olbrot, Artur Owczarkowski. Polish guide used in the discussion of piloting and form organisation.
- American Association for Public Opinion Research. Best Practices for Survey Research. Guidelines used to verify notes on sample selection and research limitations.
- Pew Research Center. Writing Survey Questions. Material used to verify principles of constructing questions and answers.
- Google. Wyświetlanie określonych pytań w zależności od odpowiedzi; Wyświetlanie odpowiedzi na pytania w formularzu i zarządzanie nimi. Google Forms documentation used to describe the tool's settings.
Polish literature expanding on the topic: Antoni Sułek (2002), Ogród metodologii socjologicznej, Warsaw: Wydawnictwo Naukowe Scholar. The book discusses, among other things, the relationship between the form of a question and the content of the response obtained.
