60 Research Assistant interview questions to ask job applicants

Good research assistant interview questions test four things: whether the candidate can run a method, handle data without breaking it, hold a deadline, and say "I don't know" when the evidence runs out. The 60 questions below are grouped so you can pull a set for each of those, score them consistently, and stop relying on whoever in the room talked the most after the candidate left.

Most question lists you'll find are written for the person sitting the interview, not the person running it. This one is built for the hiring side: what to ask, what a strong answer sounds like, and what to test before you spend an hour in a room.

TL;DR

• Ask every candidate the same core questions in the same order. Structured interviews rank among the strongest predictors of job performance; unstructured chats do not.
• Score each answer on a 1 to 5 scale during the interview, then add the scores up. Do not re-weigh everything from memory afterwards.
• Split the 60 questions into four blocks: general, behavioral, technical and methods, and ethics and data integrity. Pull 10 to 12 per round, not all 60.
• Ask about data handling and research misconduct explicitly. Fabrication, falsification and plagiarism are the three categories in the US federal definition, and a research assistant touches all three surfaces.
• Test the mechanical skills (spreadsheet work, statistical software, writing) before the interview, so the interview is spent on judgment instead of a skills audit.
• A research assistant is not a junior version of a researcher. The job is throughput and accuracy under someone else's design.

What does a research assistant actually do?

A research assistant supports a research project that someone else owns: collecting data, cleaning it, running the analysis that was specified, reviewing literature, and writing up what was found. The U.S. Bureau of Labor Statistics tracks social science research assistants as their own occupation code, 19-4061, with 30,890 people employed and a median annual wage of $56,400 as of May 2023.

That framing matters for the interview. You are not hiring someone to invent the research question. You are hiring someone who will execute a design faithfully, notice when the data contradicts it, and tell you rather than quietly smoothing it over. Those are different skills, and the second one is the expensive one to get wrong.

The commercial side of this role gets overlooked. Market research agencies, consultancies, and in-house insight teams hire research assistants in volume, and the work looks different from a university lab: shorter cycles, client deadlines, and a much lower tolerance for a study that runs three weeks over. If you're hiring on that side, weight the deadline and communication questions higher than the methods ones.

General research assistant interview questions

Start here. These 15 establish what the candidate has actually done, as opposed to what appeared on the resume. Ask them in the same order for every candidate so the answers are comparable.

1. Walk us through a research project you worked on end to end. What was your specific piece of it?
2. Which research methods have you used, and on what kind of question?
3. How do you check that a finding is actually reliable before it goes into a report?
4. Describe how you go from a raw dataset to something a reader can use.
5. How do you keep current with methods in your field?
6. Tell us about a project that went wrong. What did you do?
7. How do you run three projects at once when all three want your attention today?
8. Describe a project where you worked inside a team. What was yours to own?
9. How do you handle the ethics paperwork and approvals side of a study?
10. Explain a complicated finding to someone with no background in it. Pick a real one.
11. What do you do when the plan stops working halfway through the fieldwork?
12. Give an example of a time you changed the method because the original one was not going to answer the question.
13. How do you document your work so someone else could repeat it?
14. Describe the largest dataset you have handled. What made it hard?
15. Tell us about a project where you had to pull together sources that disagreed with each other.

Behavioral research assistant interview questions

These 15 look for evidence rather than intention. The useful ones force a specific incident, because a candidate can describe good practice in the abstract without ever having done it.

1. Tell us about a week when every deadline landed at once. What moved and what did not?
2. Describe a time the data came back and it did not support what everyone expected.
3. Give an example of feedback on your work that stung. What did you change?
4. Tell us about a time you found an error in your own analysis after it had gone out.
5. Describe a project where you built the plan from nothing.
6. Tell us about a dataset you had to clean before it was usable. What was wrong with it?
7. Describe a time you handled confidential or identifiable data. What were the rules?
8. Give an example of a survey or questionnaire you designed. What did you get wrong the first time?
9. Tell us about a time you spotted a flaw in a study design. What did you do about it?
10. Describe presenting findings to a room that did not want to hear them.
11. Tell us about a time you had to learn a tool or method fast, mid-project.
12. Describe a situation where two sources contradicted each other and you had to pick.
13. Tell us about a long project where motivation dropped. How did you keep the quality up?
14. Describe interviewing or recruiting participants. What surprised you?
15. Give an example of when you pushed back on a supervisor about a method.

Which technical questions reveal real research skill?

The technical block is where a lot of interviews go soft, because the interviewer often is not the deepest methods person in the building. These 15 are written so a non-specialist can still tell a real answer from a rehearsed one: they ask for choices and tradeoffs, not definitions.

1. Which statistical software do you reach for first, and why that one?
2. Walk us through how you'd structure a literature review on an unfamiliar topic.
3. How do you decide on a sample size when the budget is already fixed?
4. What is a control group for, in plain terms, and when have you used one?
5. How do you handle missing data? Talk about a real case.
6. Describe your process for cleaning a messy dataset.
7. Which reference manager do you use, and how do you keep it from falling apart?
8. How do you test whether a survey question is actually measuring what you think?
9. Explain a time you used qualitative and quantitative methods on the same question.
10. How do you keep an analysis reproducible six months later?
11. What does your file and folder structure look like on a live project?
12. Have you written or adapted code for research work? What for?
13. How do you sanity-check a result that looks too good?
14. Describe transcribing and coding qualitative data. How did you keep it consistent?
15. What would you do if the analysis the study specified turned out to be the wrong one?

Pro tip: ask question 13 to everyone. A candidate who has never been suspicious of a clean result has usually never owned an analysis that mattered.

Research ethics and data integrity questions

This block gets skipped most often and costs the most when it goes wrong. Fabrication, falsification and plagiarism are the three categories in the U.S. federal definition of research misconduct, and a research assistant has hands on all three surfaces: the raw data, the record of it, and the write-up.

1. What counts as research misconduct, in your own words?
2. Describe how you store and protect identifiable participant data.
3. Have you worked under an ethics board or approval process? Walk us through it.
4. What would you do if a supervisor asked you to drop inconvenient cases from a dataset?
5. How do you handle informed consent when the study design changes mid-way?
6. Describe a time you had to say no to a request about data.
7. How do you credit other people's work in a literature review?
8. What is your process when you realize a published figure of yours was wrong?
9. How do you anonymize a dataset that still needs to be useful?
10. What are your rules for sharing data with a collaborator outside the team?
11. Describe how you keep an audit trail of changes to a dataset.
12. What would you do if you suspected a colleague was cutting corners?
13. How do you handle a participant who withdraws after data collection?
14. What do you do with data that falls outside the approved scope of the study?
15. How would you explain the study's risks to a participant who is nervous?

What does a strong answer look like?

Five worked examples. For each one, the thing to listen for is specificity: a named method, a real constraint, and an outcome the candidate does not oversell.

Question: How do you check that a finding is actually reliable?
Look for: a process, not a virtue. Strong answers name something concrete, like re-running the analysis on a held-out slice, checking whether the effect survives a different specification, or having a second person re-code a sample of the qualitative data. Weak answers say "I'm very detail-oriented" and stop.

Question: Tell us about a time you found an error in your own analysis after it had gone out.
Look for: the candidate reported it quickly and specifically. The tell is how fast they say the word "wrong" without softening it. If they cannot produce an example at all, that is worth probing rather than accepting, because everyone who has done volume analysis has shipped a mistake.

Question: How do you handle missing data?
Look for: a decision tied to why the data is missing, not a default. A candidate who says "I drop incomplete rows" and can explain when that biases the result is stronger than one who names three imputation methods without a reason to pick any.

Question: What would you do if a supervisor asked you to drop inconvenient cases?
Look for: a first move that is neither compliance nor confrontation. Good answers ask what the exclusion rule is, ask for it to be written down, and check whether it was specified before anyone saw the results. That is the honest version of the question, and it separates people who have been near a real study from people who have read about one.

Question: How do you run three projects at once?
Look for: an explicit ranking rule and an admission of what gets dropped. "I make a list and work through it" is not a method. "I sort by whose deadline is external and tell the internal ones they're slipping on Monday" is.

How do you score research assistant interviews?

Write the scores down during the interview, on a fixed scale, and add them up afterwards. Adding scores mechanically predicts outcomes at least as well as experts who re-weigh the same information in their heads, which is the finding from Kuncel and colleagues' 2013 meta-analysis of mechanical versus clinical data combination. The debrief where everyone talks until a consensus appears is the part that adds noise.

The scorecard below is one application of the Testlify Competency-to-Evidence Matrix, which starts with the role rather than the test: map the role to the competencies that matter, then connect each competency to evidence you can actually collect, through assessments, work samples, interviews and structured reviewer feedback. For a research assistant, four competencies carry most of the weight.

Score each answer 1 to 5. Anything at 3 or below on integrity is a stop, regardless of the total, because that is the one competency where a strong average does not compensate. Two reviewers scoring independently, then comparing, beats one reviewer scoring carefully.

Structured interviews sit among the strongest predictors of job performance in the 2022 reanalysis of selection-method validity by Sackett and colleagues, while years of education and general years of experience sit among the weaker ones. The exact coefficients are still argued over in print. The ranking is not, and it is the part that should change how you run the hour.

When should you test skills before the interview?

Before, not after, and only for the things an interview measures badly. Nobody can tell from a conversation whether a candidate can actually clean a 40,000-row export or write a readable 300-word summary of a result. Those are work samples, and a work sample is the one part of the process a conversation cannot stand in for.

A practical split for a research assistant role: test the mechanical skills up front, interview for judgment. That usually means a short spreadsheet or statistical software exercise, a writing sample, and an aptitude or analytical-reasoning test, then a single structured interview round instead of three conversational ones.

Testlify covers that pre-interview layer. The test library includes cognitive ability, analytical reasoning, situational judgement, language and software-skills categories, plus practical and hands-on question types where candidates submit work by file upload or URL, and office-app questions bound to the real products, including Google Sheets and Microsoft Excel. The aptitude test for researchers is the closest single fit for this role. Assessments can be weighted from x1 to x5 so the scorecard above maps onto the platform rather than sitting in a separate document.

One caveat worth stating: personality and cultural tests are qualitative and do not produce a total score, so do not try to rank candidates on them. They're useful as interview material, not as a cutoff.

If the role leans more analytical than operational, compare this list against research analyst interview questions, and if you're hiring a market research assistant the commercial questions matter more than the academic ones. Teams staffing a lab or product-development function often want the research and development assistant set instead.

Hiring an onderzoeksassistent in the Netherlands?

This page gets a steady stream of Dutch-language searches, and the hiring problem is the same one in a different label. The competency map does not change across a border; the language of the interview and the assessment does.

What does onderzoeksassistent mean?

Onderzoeksassistent is the Dutch job title for a research assistant. It covers the same work: supporting a study someone else designed, collecting and cleaning data, reviewing literature, and writing up findings. Dutch and Belgian employers use it for both university and commercial research roles, so read the job description rather than the title to work out which one you're staffing.

Onderzoeksassistent job interview questions to ask applicants

Use the same four blocks above. Two things shift in the Dutch market. Ethics and data-protection questions carry more weight because GDPR enforcement sits closer to the work, so the questions on identifiable data, consent and anonymization are worth asking in full rather than sampling. And academic and commercial research sit closer together than they do in the US, so ask directly which of the two the candidate has actually worked in.

Onderzoeksassistent interview questions to ask candidates in Dutch or English

Run the interview in the language the job is done in. If the research is written up in Dutch, a candidate who interviews well in English but writes stiffly in Dutch is the wrong hire, and you will not find that out in an English-only round. Where the team is mixed, a common approach is an English structured interview plus a short written task in Dutch.

For the assessment layer, Testlify translates tests and custom questions into the candidate's preferred language in real time, with the caveat the product itself states: quality varies slightly by language pair. For a scored written task, set it in the target language directly rather than relying on translation.

Hire your next research assistant with Testlify

The 60 questions on this page are free to copy into your own scorecard. What they cannot do on their own is tell you whether the candidate can handle the data before you spend the hour. That is the part to move earlier.

Set up the assessments that match the four competencies above, run one structured round against a written scorecard, and you'll have a defensible record of why the person you hired scored higher than the person you didn't. Book a 30-minute demo and we'll build the research assistant assessment with you on the call.

Key takeaways

• Structure beats rapport. Structured interviews rank among the strongest predictors of job performance, and unstructured ones sit well below. Ask every candidate the same core questions in the same order. The cost is that the hour feels less friendly; the benefit is that the comparison between two candidates is real rather than a memory of who was more pleasant.
• Write scores down live, then add them. Mechanically combining scores predicts outcomes at least as well as experts reworking the same information holistically. Practically: score 1 to 5 during the interview, total afterwards, and treat the debrief as a check on the numbers rather than a replacement for them.
• Test the mechanical skills before the interview. Spreadsheet work, statistical software and written summaries are measured badly by conversation and well by a short work sample. Moving them earlier buys back the interview hour for judgment questions, which is where an interview genuinely outperforms a test.
• Ask the ethics block in full. Fabrication, falsification and plagiarism are the three federal categories of research misconduct, and a research assistant touches the raw data, the record and the write-up. A candidate who scores 3 or below here should not be advanced on a strong average elsewhere, because integrity is the one competency the rest cannot compensate for.
• Weight the questions to the actual job. A commercial insight team hiring for client deadlines should score communication and prioritization higher; a lab running long studies should weight method and reproducibility. Using one fixed scorecard across both is how teams end up hiring for the wrong half of the role.
• Do not over-read the resume. Academic grades predict job performance far more weakly than most screening assumes, and general years of experience is among the weaker predictors in the modern validity evidence. Screen on evidence you collect yourself instead.

FAQs