In the world of quantitative research, numbers are only as trustworthy as the people behind them. Every percentage, graph, or data table depends on one simple thing: people answering surveys. But what happens when a large group of people doesn’t respond at all? That’s where nonresponse error comes in—a subtle yet powerful threat to research accuracy.
Nonresponse error occurs when the people who choose not to participate in a survey are systematically different from those who do. This isn’t just about a few missing answers—it’s about how missing voices can shift the results, often in misleading ways.
Let’s take a real-world example in Vietnam. Imagine a beverage company launching a new low-sugar iced tea, targeting health-conscious urban millennials. They distribute an online survey through social media and mobile apps to gather opinions about packaging, pricing, and taste preferences. The results come back with highly positive feedback—most respondents say they would buy the product, love the branding, and are ready to try it. Encouraged by the data, the company rolls out the product widely across convenience stores in Hanoi and Ho Chi Minh City.
But after launch, sales are disappointing. What went wrong?
The survey failed to capture a critical group—young office workers who avoid unnecessary digital interactions during working hours or aren't active on social media. These people may have had less interest in the product or stronger opinions about pricing, but they were underrepresented because they didn’t respond. Meanwhile, the highly engaged health-food fans were overrepresented. The result? A distorted view of the broader market.
Nonresponse error doesn’t just affect opinions. It affects any data point, from brand awareness to purchase frequency. For instance, in rural areas of Vietnam, older respondents may be less willing to engage in mobile-based surveys. If these individuals consistently opt out, a survey measuring awareness of a health campaign might overstate success, simply because the most skeptical or less-informed individuals are missing from the data.
One of the trickiest aspects of nonresponse error is that it’s often invisible. Researchers may assume that low response rates are just a sign of disinterest or bad timing. But the danger lies in who exactly is not responding. For example, if a financial services brand sends a customer satisfaction survey via email, and only loyal or satisfied customers reply, it may conclude that satisfaction is high. However, dissatisfied customers—perhaps less motivated or more frustrated—might avoid responding altogether. Their absence gives a falsely positive picture.
To deal with this issue, researchers often try multiple strategies. One is follow-up reminders, which help draw in those who ignored the first invitation. In Vietnam, where people often skim over promotional emails or ignore unfamiliar phone calls, a polite second or third message—sent at the right time—can increase participation. However, even with reminders, some groups remain unreachable.
Incentives can also help boost response rates. For example, offering a small phone card credit or entry into a prize draw often works well in Vietnam, especially for younger or price-sensitive respondents. But even incentives can introduce bias. Those with more free time, or those more responsive to rewards, may still skew the sample.
Sometimes, the method of survey delivery is part of the problem. Online panels tend to overrepresent urban, tech-savvy consumers. Telephone interviews may miss out on younger people who rarely pick up unknown numbers. In-person surveys are more inclusive but time-consuming and expensive. Each method comes with trade-offs, and poor choices can amplify nonresponse error.
There was a case during a retail study conducted across three regions of Vietnam. A large chain wanted to understand why shoppers in Da Nang weren’t visiting their stores as frequently. They distributed surveys through their app and website. Unsurprisingly, the respondents were those who had already visited the store at least once. What they didn’t get was feedback from the people who stopped going—or never went at all. The company assumed satisfaction levels were high and awareness was strong. In reality, many consumers in surrounding neighborhoods hadn’t even heard of the store because the marketing had been heavily digital and failed to reach offline audiences. The survey didn’t pick up these gaps because non-visitors simply didn’t respond.
Nonresponse error doesn’t just affect external research—it also affects internal company surveys. For instance, in employee engagement research, nonresponse may indicate deeper problems. If a significant portion of employees skips a satisfaction survey, especially in specific departments or roles, it could be a sign of mistrust, dissatisfaction, or fear of backlash. Ignoring these missing voices can lead to management making decisions that further alienate already disengaged staff.
One way researchers try to estimate nonresponse error is by comparing known population data to sample data. For example, if a survey of university students shows a very low number of first-year respondents, but enrollment data shows that first-years make up 40% of the population, that’s a red flag. It tells researchers that they may be missing important input from a specific group.
Statistical adjustments like weighting can sometimes help, by giving more influence to underrepresented responses. But this only works if the missing group is known and understood. If nonrespondents are totally unknown, or if their attitudes are drastically different from those who responded, weighting won’t fix the issue—it’ll just mask it.
Cultural context also plays a role. In Vietnam, respect for authority, desire to avoid conflict, and a tendency toward indirect communication can all affect who answers a survey and how. Respondents may feel more comfortable ignoring a question than giving a negative answer, especially in phone or in-person surveys. In such cases, nonresponse may not apply to the entire survey but to specific questions. This partial nonresponse can still distort findings, especially in sensitive areas like politics, personal finance, or health.
Nonresponse error reminds us that data is not neutral. It is shaped by who speaks up, who stays silent, and why. A survey’s success isn’t measured only by how many people respond—but by how representative those responses are of the people the survey aims to understand.
