What Are Screening Questions? How to Design Them in a Survey

What are screening questions in quantitative research

A screening question appears at the beginning of a questionnaire and checks whether the respondent belongs to your target population. For example, a study of e-wallet usage intention may require respondents to have used an e-wallet within the past six months. Someone who has never used one cannot provide suitable data for questions about their experience, perceptions, or intention in the model.

Think of a screening question as an eligibility gate before the respondent reaches the scale items. It may ask about age, location, product-use experience, customer status, or when a behavior last occurred. This is different from a demographic question in a survey. Demographic questions describe the sample, while screening questions determine whether the respondent continues or exits the survey.

A good screening question has three elements: the eligibility condition is stated clearly, the response options do not overlap, and the routing rule has been decided in advance. For example, if your target population consists of people aged 18 or older who work in Ho Chi Minh City, decide what message a 17-year-old or someone working in another location will see when the survey ends.

You can review the structure of a full form in Survey topic, then keep screening questions separate from measurement questions so that your SPSS data file is easier to check.

How to determine screening questions and survey sample size

Before writing the questions, describe your target population in one sentence using four points: who the people are, where they are, what experience they have, and the relevant time period. For example: “Consumers aged 18 or older, currently living in Hanoi, who have purchased cosmetics online at least once in the past six months.” Each important condition in this description usually needs a corresponding screening question.

Screening questions do not mechanically increase the sample size for analysis. Distinguish between the number of people who open the questionnaire, the number who meet the conditions, the number of valid responses, and the final number of observations imported into SPSS. If you expect only 60% of the people reached to meet the eligibility conditions, you need more initial invitations than the number of valid responses required.

For research with a known finite population, the finite-population sample-size formula is often presented as follows according to Yamane (1967):

n = N / (1 + N.e²)

Here, N is the population size, e is the expected sampling error, and n is the minimum sample size produced by the formula. The formula supports the design of your data collection plan, but it does not guarantee the quality of the final dataset. You still need to account for incomplete responses, people who fail the screening questions, and records removed during data cleaning.

If your study uses a scale with multiple items, a rule used in many theses is approximately 5 to 10 observations for each item (Hair et al., 2010). For regression, the rule 50 + 8m, where m is the number of independent variables, is stated by Tabachnick and Fidell in their guidance (Tabachnick and Fidell, 2013). Treat these as reference points and compare them with your supervisor's requirements.

Create a separate tracking table for the number of visits, the number of people who pass each condition, the number who complete the questionnaire, and the number of records removed. For example, if you need 250 valid responses but only 70% of the people who start the form are eligible and complete it, the initial reach must be substantially higher. This is a data-collection plan, not the final analytical sample size.

Designing the questionnaire for this purpose

A quantitative questionnaire usually has four parts: an introduction and consent statement, screening questions, survey demographic questions, and the research scales. The order can vary by topic, but eligibility conditions should appear before long questions about perceptions or behavior.

Write the eligibility condition clearly

Each question should state one condition at a time. For example: “Have you purchased a product on an e-commerce platform within the past six months?” The two response options might be “Yes” and “No.” If the respondent chooses “No,” the form moves to the end page. This is easier to check than one question combining purchase behavior, purchase frequency, and spending level.

When several conditions are required, use several questions in sequence. You might ask about location first and then ask about usage experience. When someone is screened out, you can see which condition they failed instead of finding only one general screening variable with a value of “No.”

Distinguish screening questions from demographic questions

“Which age group do you belong to?” can be a screening question if the study accepts only people aged 18 or older. It can also be a demographic question if every age group is retained for comparison. The content alone does not determine the category. The purpose and the way you process the response do.

What are demographic questions in this context? They describe sample characteristics such as gender, age, education, occupation, income, or residential area. They are usually included in the sample description and may be used for group comparisons, but you should not automatically treat them as independent variables when the theoretical model does not require them.

Choose the scale and response format

After passing the screening questions, respondents reach the items. For an attitude scale, you can use a 5-point Likert scale ranging from “Strongly disagree” to “Strongly agree,” based on the scale format introduced by Likert (1932). If your research design specifies 7 points, use that format consistently rather than mixing 5-point and 7-point scales in the same questionnaire.

You can review the guidance on the 5-point Likert scale to keep the response labels consistent. Each item should express one idea. Avoid wording such as “The product is fast and convenient,” because respondents may evaluate two different attributes in the same statement.

Design the exit route

The exit message for an ineligible respondent should be short and neutral. You do not need to continue asking questions just to reach the required number of responses, because those data do not represent the target population. The message should also avoid making respondents feel that they have been judged.

Before publishing the questionnaire, test it with a few people who resemble the target population. Check whether they understand the eligibility conditions, see the correct exit route, and avoid repeated pages. This tests the operation of the questionnaire. It does not replace the reliability analysis you will run after collecting the data.

Collecting the data

Record how you reached respondents, the collection period, the screening criteria, and how you controlled multiple submissions by one person. If you post the link in several groups, note that the sample may become concentrated in particular communities. In Chapter 3, describe the channels you actually used instead of calling the sample representative of the whole population when you did not use a probability design.

You can use fillform.info, DoThesis's own form tool, to create screening questions, set exit routes, and collect responses. Google Forms is a common alternative. Whichever tool you use, inspect the exported data before the official launch, especially the column names, missing values, and coding of response options.

A practical process may include the following steps:

1. Create a draft with the screening questions and demographic questions.
2. Set an exit route for every response option that does not meet the eligibility conditions.
3. Test the form using an account that meets the conditions and an account that does not.
4. Export a test file to see which column each question creates.
5. Publish the questionnaire for the period recorded in advance.
6. Track the number of valid records, and do not change the conditions midway without recording the reason.

If you need to compare the layout, survey questionnaire and survey questionnaire template can help you check the introduction, screening, demographic, and scale sections. Do not copy a template word for word if the context, population, or constructs in your study are different.

Cleaning the data before importing it into SPSS

When you export a .csv or .xlsx file, keep an untouched copy of the original data. Create a duplicate for cleaning and record every removed row, the reason for removal, and the processing date. If the form creates a completion-status column, use it together with the screening variables to identify eligible records.

Before running Cronbach's Alpha or EFA, check the following points:

Straight-lining does not automatically prove that a respondent acted dishonestly. Someone may genuinely choose the same option for similar statements. Combine the response pattern with completion time, the missing-response rate, and other checks only if those checks were designed from the beginning.

In SPSS, open Analyze > Descriptive Statistics > Frequencies to review the distributions of the screening and demographic variables. Then use Data > Select Cases if you need to temporarily select cases that meet the conditions. Remember that Select Cases can hide rows excluded from the analysis, so record the filter condition in syntax or in your processing log.

How to report this in Chapter 3

The methods section should tell the reader who you surveyed, which tool you used, when you collected the data, and how you applied the eligibility conditions. You do not need to include every response in the thesis. Put the complete questionnaire in the appendix and describe the variable groups in the main text.

How to write this in your thesis: “[The study population] was defined as [description of the target population]. Respondents were included in the sample when they met [condition 1], [condition 2], and [condition 3]. Cases that did not meet the screening conditions, did not complete the questionnaire, or showed signs of invalid responding were removed before the data were analyzed in SPSS. After data cleaning, the study used [n] valid responses.”

Replace every bracketed section with your actual information. If you do not yet have the final number of responses, leave the placeholder in the draft rather than entering an attractive number. Your description table may include the number of questionnaires distributed, the number returned, the number meeting the eligibility conditions, the number removed, and the number used for analysis.

When describing demographic questions in the survey, state the variable groups and their purposes. For example, age and occupation may describe the sample, while having purchased a product within the past six months is a screening condition. This distinction helps the committee understand why some variables do not appear in the regression model.

You can also review the form structure in questionnaire design, but your thesis must describe the questionnaire you actually used rather than a different template.

Common mistakes

Placing screening questions after the scale

An ineligible respondent has already answered the entire set of items before being removed. This increases completion time, creates unnecessary data, and makes it easy to forget the filter when running SPSS. Put the important eligibility conditions at the beginning of the form.

Using vague conditions

Phrases such as “use frequently” or “purchased recently” do not mean the same thing to everyone. Specify a period or frequency, such as “at least once in the past six months,” when that fits the research design.

Combining the screening variable with the scale

The variable “has used the product or not” is not an item measuring the same construct as the Likert statements. If you include it in Cronbach's Alpha, the result may be difficult to interpret and inconsistent with the scale structure.

Removing records without recording the reason

Deleting apparently unusual rows without keeping a processing log makes the decision difficult to defend. Create a column or file that records the response code, the failed condition, and the final decision.

Changing the criteria after seeing the sample count

If too few records meet the conditions, you can expand the recruitment channels or extend the collection period according to the approved plan. Lowering the criteria simply to reach the sample-size target changes the target population and should be discussed with your supervisor.

Frequently asked questions

Do screening questions count as items?

Usually, they do not. Screening questions establish eligibility, while items measure a construct in the model. You should still import the screening variables into the data file to check the records and document the process, but do not automatically include them in Cronbach's Alpha or EFA.

Should age and gender be asked in the screening section?

Use age or gender for screening only when the study clearly defines a population limit, such as a study of people aged 18 or older. If all age groups or genders are eligible, place these questions in the demographic section to describe the sample.

How should I handle a respondent who selects “No” on a screening question?

Set the exit route before publishing the form. In the data, retain the indicator showing that the person was ineligible so you can reconcile the counts, then remove the record from the analytical dataset if the research criteria require it. Do not change the response to missing and treat the record as valid.

Do I need to report screening questions in Chapter 3?

Yes. Describe the eligibility conditions, the routing process, and the record-removal criteria. The full wording can go in the appendix, while Chapter 3 gives a concise description of the content and its role in data collection.

Can screening questions bias the sample?

The question itself does not automatically bias the sample, but criteria that are too narrow or recruitment channels concentrated in one group can make the sample differ from the target population. State the criteria, collection channels, and limits on generalization clearly in the thesis.

Now open your questionnaire file, write one sentence describing the target population, list the eligibility conditions, and test every exit route before collecting more data. If you need to run the analysis on your own .sav or .csv file, you can use M4 data analysis by DoThesis.