You have designed the perfect questionnaire. The questions are clear, the sample is well chosen, and the topic matters. Then reality hits: people slam the phone down, half-filled forms trickle back, and some answers look suspicious. Survey research is one of the most powerful tools in the social sciences, but it lives or dies by the quality of the responses it collects. Understanding the common problems that surface during fieldwork, and knowing how to tackle each one, is what separates a credible study from a wasted one. This post walks through four of the biggest challenges and the practical fixes that researchers rely on.
Table of Contents
Non-cooperation from respondents
The first wall most researchers hit is simple refusal. People decline to take part, or they start a survey and abandon it midway. This matters because the people who refuse are often systematically different from those who agree, which quietly skews your sample before analysis even begins.
Why do people say no? The reasons are rarely dramatic. Lack of time is the most common one. Filling out a survey now feels like a matter of personal choice and convenience rather than a civic duty, so a long or demanding form competes poorly with everything else in a person’s day. Disinterest is the next big factor; if the topic does not touch a respondent’s life, they see no reason to engage. Privacy concerns also play a growing role. As researchers have noted, a single news story about a data breach can be enough to make people more guarded about sharing anything at all. Suspicion of strangers, fatigue from constant marketing calls, and a fear that the survey is a sales pitch in disguise round out the list.
How to overcome non-cooperation
The core idea is to lower the cost of participating and raise the perceived benefit. Keep the questionnaire short and respect the respondent’s time openly. Explain clearly who you are, why the research matters, and how the findings will be used, because people are far more willing to share when they understand the purpose. Target the right audience so the subject actually resonates with them, and assure them of confidentiality. A friendly, professional first contact does more to convert a hesitant person than any clever wording later in the form.
Low response rates
Closely tied to non-cooperation is the broader problem of low response rates. The response rate is the share of your intended sample that actually completes the survey, and it has long been treated as a headline indicator of survey quality.
The danger of a low rate is non-response bias. If the people who do not respond differ from those who do on the very things you are measuring, your results stop representing the population you care about. Researchers describe this as a perennial concern, because non-respondents may differ from respondents on exactly the key variables a study was built to examine. Interestingly, the same body of research warns that the relationship is more complex than it looks: a higher response rate does not automatically guarantee less biased estimates. Still, a very low rate leaves you unable to defend the representativeness of your sample, which is why it threatens the credibility of the whole study.
Strategies to lift engagement
Decades of experiments point to a reliable toolkit. Reminders are the single most cost-effective fix. A literature review of web surveys found that email pre-notification, an email invitation, and two follow-up reminders all push response rates up, especially when paired with a simple design and a survey short enough to finish in about ten minutes. Incentives work too. A field study reported that prepaid cash incentives combined with follow-up waves were an effective way to improve response and recruitment. Beyond these, personalising the invitation can raise response rates substantially by adding a human touch.
A few design choices help as well. Do not open with demographic questions, which feel intrusive before any trust is built; save them for the end. Make the form mobile-friendly, since most people now answer on a phone. Pilot test the survey on a small group first to catch confusing items, and where possible share what you eventually do with the results, which encourages future participation.
Deliberate wrong information
Even when people cooperate and complete the survey, a subtler problem remains: they may not tell the truth. This is not always malicious. Often it is a predictable human reaction known as social desirability bias, where respondents give the answer they think looks good rather than the honest one.
The classic example is the question about alcohol or unhealthy habits, where someone who drinks four or five a day might report one or two to appear respectable. The bias is strongest on sensitive topics like income, religion, personal habits, and illegal behaviour. It can show up as self-deception, where people genuinely see themselves favourably, or impression management, where they knowingly give exaggerated answers to look good. Either way, if enough people shade their answers, the aggregate data becomes misleading and the conclusions drawn from it become unsafe.
Ways to minimise dishonest answers
The most powerful lever is anonymity. When people trust that their answers cannot be traced back to them, the incentive to lie drops. Self-administered formats tend to produce more honest answers than face-to-face interviews on sensitive subjects, because there is no interviewer to impress.
Beyond anonymity, researchers use clever questioning techniques. Indirect questioning asks people to answer on behalf of others or to compare two questions rather than answer a sensitive one head-on. One team at Tilburg University built a method where respondents only state whether the answers to two questions are the same or different, which makes it easier to answer the sensitive item truthfully while protecting their privacy. The randomised response technique uses a chance device so a respondent can answer truthfully without ever revealing which question they answered. Careful, neutral wording that does not signal a “right” answer also reduces pressure. Finally, you can cross-check responses for internal consistency and flag contradictions that suggest inflated or invented answers.
Data collection outsourcing issues
Large surveys often cannot be run by a single research team. Researchers hire external agencies, field interviewers, or call centres to gather data at scale. Outsourcing brings real advantages, but it also introduces a fresh set of risks to data reliability.
The case for outsourcing
Professional firms bring reach and speed that an individual researcher cannot match. They can handle high-volume fieldwork, manage multilingual studies across regions, and free the core team to focus on design and analysis. Good firms also run quality systems, using tools to catch duplicate respondents, speeders, and other invalid responses, and they monitor quotas in real time to keep the sample balanced. For a multi-state study covering several languages, this expertise can be the difference between finishing on time and not finishing at all.
The hidden risks
The trouble is that the agency’s incentives are not perfectly aligned with yours. To finish faster and protect their margin, a firm may include respondents who do not meet your sampling criteria, monitor enumerators loosely, or underpay field staff who then cut corners. The most serious threat is curbstoning, the outright fabrication of responses by data collectors who never actually conduct the interview. One review describes it as a willful fabrication that threatens the integrity of every inference drawn from the data, and one that persists even at professional levels. Fabrication is not always about bad character; field studies show it can be driven by demoralised workers facing poor management and weak institutional support, treating shortcuts as a way to cope with impossible workloads.
Keeping outsourced data reliable
The fix is verification, not blind trust. Validate that each collector genuinely gathered the data assigned to them. Build constraints into digital data collection tools so that skip logic and range checks prevent surveyors from entering nonsensical or illogical answers in the first place. Conduct back-checks by re-contacting a random share of respondents, audit the timing and location stamps of interviews, and look for statistical patterns that betray invented data. Treat the agency as a partner whose interest in quality you reinforce through clear contracts, fair pay for field staff, and proper training of replacements rather than as a black box you simply hand the project to.
What do you think? If you had to run a survey on a sensitive topic with a tight budget, would you spend more on incentives to raise the response rate or on verification systems to catch dishonest and fabricated answers? And when an external agency delivers data that looks too clean and too consistent, should that reassure you or make you more suspicious?
References
- https://www.demographic-research.org/volumes/vol32/26/32-26.pdf
- https://pointerpro.com/blog/improve-survey-response-rate/
- https://pmc.ncbi.nlm.nih.gov/articles/PMC3681235/
- https://www.sciencedirect.com/science/article/abs/pii/S0020748921002054
- https://link.springer.com/article/10.1186/s12874-019-0868-8
- https://www.qualtrics.com/experience-management/research/tools-increase-response-rate/
- https://atlasti.com/guides/interview-analysis-guide/social-desirability-bias
- https://www.tilburguniversity.edu/current/press-releases/how-prevent-social-desirability-bias-surveys
- https://www.surveycto.com/data-management/tips-survey-firm-mobile-data-collection/
- https://www.researchgate.net/publication/282448585_Curbstoning_and_beyond_Confronting_data_fabrication_in_survey_research
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5034849/
Leave a Reply