Survey or Interview? How to Choose the Right Method for Your Research
Use a survey when you need standardized information that can be compared across participants, such as prevalence, frequency, group differences or statistical associations. Use qualitative interviews when you need to understand experiences, meanings, processes or reasons in depth. Sometimes you need both. The right choice follows from the research question, not from which method is cheaper, faster or more familiar.
“Should I use a survey or an interview?”
Before answering, we need to remove one small methodological trap.
A survey and an interview are not always opposites.
A structured survey can be administered by an interviewer.
What researchers usually mean when they ask this question is:
Should I collect standardized responses from many participants, or conduct open-ended qualitative interviews to explore people's experiences in depth?
Now we have a useful decision to make.
When should I use a survey?
Use a survey when the question requires structured measurement across participants.
For example:
What proportion of university students used a modern contraceptive method during the past six months?
You need a defined variable collected in a consistent way so that responses can be summarized across the sample.
Surveys are particularly useful for questions such as:
- How many?
- How often?
- What proportion?
- What is the average?
- Does one group differ from another?
- Is exposure X associated with outcome Y?
If the sampling design supports it, survey data may also be used to make population-level estimates with quantified sampling uncertainty.
When should I use qualitative interviews?
Now change the question:
How do university students describe the factors that shape their decisions about contraceptive use?
A fixed checkbox list may miss the point.
Open-ended interviews allow participants to describe:
- experiences;
- motivations;
- perceptions;
- meanings;
- decision-making processes;
- unexpected barriers or facilitators;
- social and institutional context.
Interviews are particularly useful for questions such as:
- How?
- Why?
- What is the experience of...?
- How do people understand...?
- What processes shape...?
- How do people navigate...?
If you want a percentage, you need data that can produce a defensible percentage.
If you want to understand meaning or process, give participants room to explain it.
same topic, different questions
Suppose you are studying adolescents' use of youth-friendly health services.
Research question A
What proportion of adolescents used a youth-friendly health service in the past 12 months?
A survey is a natural fit.
You need standardized information on use and a sampling strategy capable of supporting the inference you want to make.
Research question B
How do adolescents decide whether a youth-friendly health service feels safe enough to use?
This is a different question.
Qualitative interviews could allow adolescents to describe confidentiality, provider interactions, family expectations, social visibility and other factors in their own terms.
Research question C
What proportion of adolescents avoid the service, and why?
Now you have two linked questions.
A mixed-methods design may be appropriate:
- survey → estimate the pattern;
- interviews → explore the mechanisms behind it.
You do not use both methods because “mixed methods looks stronger.”
You use both because the study asks for two different kinds of evidence.
Survey versus interview: a quick comparison
| If you need to know... | Survey may fit | Qualitative interview may fit |
|---|---|---|
| Prevalence or proportion | Yes | No, not by itself |
| Frequency | Yes | Usually not the main purpose |
| Statistical association | Yes | No |
| Group differences | Yes | Can explore differences, not estimate them statistically |
| Experience | Limited if predefined | Yes |
| Meaning | Limited | Yes |
| Why a process happens | Sometimes indirectly | Often |
| Unexpected explanations | Limited by instrument | Stronger |
| Population estimate | Possible with appropriate sampling | No |
| Rich contextual detail | Limited | Strong |
This table is a starting point.
It is not a substitute for a research question.
Do not choose based only on sample size
A common shortcut is:
Survey = many participants. Interview = few participants.
That is true often enough to be memorable and incomplete enough to cause trouble.
The key difference is not merely the number of people.
It is the kind of evidence the method is designed to generate.
A poorly designed survey of 2,000 people will not necessarily answer a “why” question.
Thirty rich interviews cannot estimate district prevalence.
Bench Check: the closed questionnaire pretending to answer “why”
Suppose your research question is:
Why do adolescents avoid the clinic?
Your questionnaire asks respondents to tick yes/no for:
- distance;
- cost;
- provider attitude;
- confidentiality;
- lack of knowledge;
- parental disapproval.
You can estimate how frequently respondents endorse those six researcher-defined barriers.
But you have not necessarily designed a study capable of discovering why adolescents avoid the clinic.
You have designed a study capable of measuring agreement with six explanations you chose before data collection.
That may still be useful.
Just be honest about the question it answers.
Can an interview be quantitative?
Yes.
An interviewer can administer a structured questionnaire and record standardized answers. The presence of an interviewer does not automatically make a study qualitative.
Likewise, an online questionnaire can include open-ended questions, although a few free-text boxes do not automatically create a rigorous qualitative study.
The issue is the structure, purpose and analysis of the data, not simply who asked the question.
When should I use both surveys and interviews?
Use both when the research problem genuinely requires complementary evidence.
Common reasons include:
Explanation
A survey shows what is happening; interviews help explain why.
Instrument development
Qualitative work helps identify concepts and language before building a structured survey.
Triangulation
Different methods examine a phenomenon from different angles.
Expansion
One method answers questions the other cannot.
But mixed methods creates additional work:
- integration must be planned;
- samples may need to be linked or intentionally distinct;
- timing matters;
- the qualitative component should not become an appendix added after the “real” quantitative study.
A one-sentence decision test
Complete this sentence:
By the end of this study, I need to know...
If the answer is:
“...the percentage, prevalence, frequency, average or statistical relationship...”
you probably need structured quantitative data.
If it is:
“...how people experience, explain, understand or navigate...”
you probably need qualitative inquiry.
If it genuinely contains both, consider whether a mixed-methods design is justified.
What do I actually write in my proposal?
Quantitative example
A cross-sectional survey will be used because the study aims to estimate the prevalence of modern contraceptive use and assess associations with selected participant and service characteristics.
Qualitative example
In-depth interviews will be used because the study seeks to explore how students understand confidentiality and how these perceptions shape decisions about seeking sexual and reproductive health services.
Mixed-methods example
A sequential mixed-methods design will first use a survey to estimate service use and identify patterns across participant groups, followed by in-depth interviews to explore the mechanisms participants describe as shaping those patterns.
These are examples of reasoning, not boilerplate. Your design language should match what you will actually do.
Write your primary research question at the top of a blank page.
Underneath it, write:
The evidence needed to answer this question is...
If your answer describes numbers, estimates and comparisons, your method should be able to produce them.
If it describes experience, process and meaning, your method should allow them to emerge.
If the method you planned cannot produce the evidence you just described, revise the method before you revise the question to fit it.
Frequently asked questions
- Are interviews better than surveys?
No. They answer different types of questions. A method is “better” when it is better aligned with the research objective.
- Can I use interviews to calculate percentages?
You can count responses in interview data, but a small purposive qualitative sample is generally not designed to estimate population prevalence. Frequency counts can also distract from the purpose of qualitative analysis.
- Can I use a survey to answer why something happens?
A survey can test predefined explanations or associations, but it may miss unanticipated mechanisms. If “why” is central, qualitative inquiry may be needed.
- Is a questionnaire the same as a survey?
A questionnaire is a data-collection instrument. A survey is a broader approach that includes sampling, administration, measurement and analysis.
- When should I use mixed methods?
When the research question genuinely requires more than one kind of evidence and you have a clear plan for how the components will connect.
Use the Methods Bench Study Design Decision Guide to move from your research question to the type of evidence and design that can actually answer it.
References and further reading
- 1.World Health Organization. Health Research Methodology: A Guide for Training in Research Methods. https://iris.who.int/
- 2.CDC. Collecting and Analyzing Qualitative Data, Field Epidemiology Manual. https://www.cdc.gov/field-epi-manual/php/chapters/qualitative-data.html
- 3.CDC/NCHS. National Health Interview Survey Methods. https://www.cdc.gov/nchs/nhis/about/nhis-methods.html
- 4.World Health Organization. Resources on asking young people about sexual and reproductive behaviours. https://www.who.int/news/item/08-05-2014-asking-young-people-about-sexual-and-reproductive-behaviours
Better research, once a week.
Practical methods guidance, research tools and funding opportunities from Methods Bench.
Get practical research notes and new opportunities
Short, useful emails for researchers in Uganda and East Africa. No spam.