Survey response rate: calculate, assess and increase it

How to calculate your survey response rate, what a good rate is and how invitations, reminders and incentives increase it. With a worked example.

The survey response rate is the share of completed questionnaires among all invited people who belong to your target group. From the denominator, you only subtract people who demonstrably don't belong to the target group. When in doubt, undeliverable invitations stay in. The response rate is not the same as the completion rate, which is completed questionnaires divided by started ones. There's no universally good rate, and a high rate doesn't prove good data: a result is biased when the people who don't respond differ from those who do (Schnell 2019, pp. 159-160).

I hear this question from students before they hand in their thesis as well as from HR teams after an employee survey: "I invited 1,200 people and got 410 responses. Is that good?" In this article, you'll learn how to calculate the response rate correctly, which values the literature reports, why a high rate doesn't automatically mean good data and what actually increases participation. How many responses you need is covered in Calculating sample size, and where to find respondents in Where do I find respondents for my survey.

How to calculate the response rate: formula and denominator

The response rate answers a simple question: how many of the people who should have responded actually did?

Eligible means the person belongs to the target group of your survey. German textbooks usually call this calculation the Ausschöpfungsquote. Schumann divides the number of participants by the adjusted initial sample. Adjusted means that the so-called sample-neutral losses are subtracted first, for example people who don't belong to the population at all, or addresses that are simply wrong (Schumann 2019, pp. 103-104), and likewise Möhring and Schlütz (2019, pp. 34-35). For undeliverable email invitations, AAPOR advises against subtracting them (see below). In the example further down, Schumann's calculation gives 37.3%, the cautious approach 35.3%. I use the two terms interchangeably here, while keeping undeliverable invitations in the denominator.

The denominator question: invited, reached, eligible

The numerator is rarely disputed. The denominator leaves more room for judgment. Schumann points out that the more losses you count as "neutral" and subtract, the higher the rate becomes, and whether a loss is neutral is often hard to decide in practice (2019, p. 104). Schnell explicitly speaks of ways to manipulate response rates (2019, p. 159).

Three cases come up again and again with email invitations:

Undeliverable emails. It's tempting to simply subtract them. AAPOR, the American Association for Public Opinion Research, advises against it: an address can be wrong or out of date even though the person belongs to the target group. In AAPOR's words, "an undelivered email message essentially would place such cases in the unknown eligibility category", not automatically among the ineligible (AAPOR 2023, p. 27). The cautious approach: leave them in the denominator and also report the rate without them.

People who don't belong to the target group. Someone who writes back "I've never taken a course with you" really doesn't belong in the denominator. What matters is that you can show this, not just assume it.

Break-offs and partial responses. AAPOR's strictest response rate, RR1, counts only complete interviews and leaves partial interviews out of the numerator. RR2 counts partial interviews as well, but the standards don't settle once and for all where a break-off ends and a partial interview begins (AAPOR 2016, cited in Schnell 2019, p. 151). So decide before you start what counts as "completed", and stick to it.

By the way, with an open link, for example on social networks, there's no denominator: you don't know how many people saw the link. AAPOR states that "for non-probability samples, response rate calculations make little sense" (AAPOR 2023, p. 27), and Möhring and Schlütz point out that for online surveys the rate often can't be calculated at all (2019, p. 35). In that case, report the number of responses, the field period and the distribution channel, and don't invent a rate.

If you do know the number of people invited, but didn't draw the list at random from your target group, for example through acquaintances or a group on a social network, AAPOR calls the share of usable responses among all personal invitations a participation rate, not a response rate. That's because there, a high rate doesn't necessarily lower the risk of bias (AAPOR 2023, p. 78).

Response rate vs. completion rate: four metrics, four questions

The completion rate tells you something about your questionnaire: is it too long, is there a question where many people quit? The response rate tells you something about your fieldwork: did you reach and convince the target group? Both numbers are useful, but they're not interchangeable.

Worked example: an email survey with reminders

The numbers are invented and calculated with the script for this article. A training provider emails 1,200 people who took a course in the past year and invites them to a satisfaction survey. After three weeks, the tally looks like this:

• 60 invitations bounce as undeliverable, 1,140 are delivered.
• 40 people write back that they never took a course. They don't belong to the target group.
• Of the remaining 1,100, 520 start the questionnaire and 580 don't react.
• 410 complete it, 110 abandon it.
• During the check, 14 questionnaires are excluded based on quality criteria set in advance. That leaves 396 usable questionnaires.

Depending on the denominator, you get very different percentages:

All numbers: my own calculation with invented data. RR1, which AAPOR also calls the minimum response rate, is its strictest response rate: the denominator also includes all cases of unknown eligibility (AAPOR 2023, p. 85).

What would I report? The cautious response rate of 35.3%, with the sentence: "60 invitations were undeliverable and are included in the denominator. Without them, the rate is 37.3%." Plus the number of usable questionnaires the analysis is based on: 396. That way, anyone can follow what you calculated. Möhring and Schlütz recommend stating the response rate in the publication anyway, where it can be given (2019, p. 192).

And the 78.8%? That's the completion rate. It looks friendlier, but it answers a different question: of 1,100 people in the target group who were reached, only 520 started at all, that is 47.3%. So more than half never even began.

Planning invitations with your own rate

The response rate of one survey is the best basis for planning the next one with the same group. If the provider needs 300 usable responses next year, what counts is the share of usable questionnaires among all invitees, because next time, too, some addresses will be undeliverable and some questionnaires will be excluded: 396 / 1,200 = 33.0%. That gives 300 / 0.33 = 909.1, rounded up to 910 invitations. How many responses you need for a given precision, and why the calculation only works out on average, is covered in From sample to invitations. In my course, response and drop-out rates are among the factors the required sample size depends on (Marx 2026, slide 136).

What is a good survey response rate?

The honest answer: there's no number that holds for all surveys. Response rates depend on the quality of the survey, the care taken in the fieldwork, the length of the questionnaire, the survey mode and the target group (Möhring and Schlütz 2019, p. 35). That's why the values reported in textbooks vary widely:

Please don't read the table as targets. The numbers come from different years, countries and methods, and they're not all defined the same way. Achieved response rates vary widely and are declining overall (Häder 2015, cited in Möhring and Schlütz 2019, p. 35). For online surveys with open recruitment, Möhring and Schlütz expect lower response rates, because people first have to find the link and then take action themselves (2019, p. 145).

My advice: start by comparing your results with your own previous surveys. The rate of your last survey of the same group, with the same invitation and the same field period, tells you more than any outside average. For a thesis: state the rate with its numerator and denominator, explain your denominator decision and discuss who might have been missing. That's more convincing than a high number without explanation.

Response rates in employee surveys

In employee surveys, participation is watched especially closely, because it's also taken as a sign of how well the survey is accepted (Wiertalla 2013, p. 182). The literature names thresholds: according to Bungard (1997), results from response rates below 50% can only be generalized to a limited extent. Görtler and Rosenkranz (2006) consider 70% acceptable, and according to Trost et al. (1999), more than 90% can even be a reason for suspicion (all cited in Wiertalla 2013, p. 182). Linke describes participation of about 94 to 97% as a census, since some employees are always sick or on vacation (Linke 2018, p. 56).

These thresholds are old, and the passages cited don't explain where they come from. I wouldn't use them as a benchmark. More important: here, the denominator often comes from HR data. Before you start, write down who is employed on the reference date and belongs to the target group. And a high rate doesn't help much if it comes from pressure, more on that below and in the article Employee survey.

High response rate, good data?

Schnell calls it misleading to see the response rate as the sole quality indicator of a study. For most statistics, bias due to nonresponse, known as nonresponse bias, results from two ingredients: the share of nonrespondents and the difference between respondents and nonrespondents (Schnell 2019, p. 159). For a mean or a proportion, nonresponse bias can be expressed like this (after Schnell 2019, p. 160):

Bias here means the difference between the value among respondents and the value in the entire group you're looking at. What matters is the difference in the characteristic you're studying. The formula shows: if there's no difference between the two groups, there's no bias, no matter how many are missing. And a high rate doesn't protect you if the few who are missing think very differently.

Let's look at our example. Of the 1,160 invitees who are eligible or whose eligibility is unknown, 396 usable questionnaires are available. For 764 cases, that is 65.9%, there's no usable data, including the 14 excluded questionnaires. For this simplified example, we treat all 1,160 people as belonging to the target group. Suppose 70% of the people with usable responses are satisfied, measured as the share of "somewhat satisfied" or "very satisfied" on a five-point scale with a neutral midpoint. With this fixed share of missing data, how far the reported share deviates from the value for the whole group depends on how satisfied the other people are, and that's something you don't know:

My own calculation with invented data, formula after Schnell 2019, p. 160. The 80% row is the same worked example as in the article Employee survey.

The last row is the uncomfortable one: an 80% usable response share, and yet the result is further off than with 34.1% and a small difference. In a meta-analysis by Groves and Peytcheva, the nonresponse rate explained only 11% of the variation in bias between the individual estimates (Groves and Peytcheva 2008, cited in Schnell 2019, pp. 160-161). Möhring and Schlütz report studies according to which follow-ups increase the response rate but hardly improve the quality of the sample (2019, p. 47).

Does that mean the rate doesn't matter? No. Schnell stresses that this can't be used as an argument against higher response rates (2019, p. 161), and the possible error grows with the share of those missing. Jacob et al. consider it more important to achieve the highest possible response rate than to explain the refusals afterwards. Correction methods applied after the fact rest on assumptions that are hard to justify (2019, p. 227). So: get as many responses as you reasonably can, but check who's missing.

What you can check

Compare known characteristics. If you know the structure of your target group, for example departments, locations or course types, compare the respondents with it. By their nature, little is known about people who refuse, and often all that's left is a rough comparison of the composition with known data (Jacob et al. 2019, p. 226). Möhring and Schlütz advise keeping a record of the selection and collecting even rudimentary information about the nonrespondents (2019, p. 35).

Compare early and late responses. People who only respond after the second reminder may be closer to the nonrespondents than the quick ones. Such a comparison can provide a clue, but it is not proof. Borg found no systematic differences in opinions and attitudes between early and late respondents (Borg 2000, cited in Möhring and Schlütz 2019, p. 139).

Weighting only with caution. Giving more weight to responses from underrepresented groups after data collection is highly controversial (Schumann 2019, p. 106). The method assumes that the losses within the weighting classes are random, and it often only hides the problem (Jacob et al. 2019, p. 227).

By the way, the question of the response rate at which a study becomes "representative" can't be answered this way (Schnell 2019, p. 163). What makes a sample representative is covered in the article Representative sample: what it is and how to check yours.

How to increase your survey response rate

There's no single trick. Schnell stresses that isolated measures are usually not enough to reduce nonresponse substantially. It takes a combination, and respondents want to be taken seriously as people with their own concerns (2019, p. 172). Easier said than done, but the individual building blocks are well known.

The invitation: who, why, how long

The invitation is the first contact with your survey and often decides whether people take part. It should be personal, name the topic, explain why each person's opinion counts, state how long it takes and include a note on data protection (Möhring and Schlütz 2019, p. 46). Also name the person responsible and provide their contact details, say what the results will be used for, and add a thank-you (Weinreich and von Lindern 2008, p. 78).

The sender matters. People are more willing to cooperate with academic surveys than with commercial ones (Stocké and Becker 2004, cited in Möhring and Schlütz 2019, p. 46). So if a university is behind the survey, say so in the invitation (Möhring and Schlütz 2019, p. 46). Schnell puts it in a nutshell: a serious letter from a university hospital, even in drab official grey, is likely to bring in more responses than a colorful envelope from an unknown market research firm (2019, p. 249).

State the length honestly. Give the duration your pretest showed, not the one you'd like. In the pretest, also check how far the test participants have got after half the planned time. Anyone who has only finished a quarter by then will notice that the stated time isn't right, and the invitation loses credibility. In that case, shorten the questionnaire or correct the stated time.

Announce it where you can. Malhotra and Birks put it plainly: "Prior notification increases response rates", because it draws attention to the purpose and benefits of the survey (Malhotra and Birks 2007, p. 446). If you have postal addresses, an advance letter sent beforehand clearly increases willingness to take part, even in online surveys (Zerr and Müller-Schneider 2002, cited in Möhring and Schlütz 2019, p. 46, n. 18).

Adapt the sentence about anonymity to what you've actually set up. Only make promises you can actually keep when analyzing the responses.

Reminders: how many, when, to whom

Reminders are among the most effective tools. In mail surveys, they reach precisely the people who otherwise take part less often, for example the less educated and those with little interest in politics (Reuband 2001, cited in Möhring and Schlütz 2019, p. 139). Malhotra and Birks consider follow-up particularly effective in mail and internet surveys: "Two or three mailings may be needed in addition to the original one." (Malhotra and Birks 2007, p. 446).

How many? For online surveys, Jacob et al. mention one or two reminder letters, which lets you complete the whole survey in two to three weeks (2019, p. 121). For customer surveys, Weinreich and von Lindern recommend more than two reminders only in exceptional cases, because each additional one increases the number of annoyed recipients (2008, pp. 80-81).

When? In the Dutch Health Survey, a web survey of the Dutch statistics office with invitations by letter, the first reminder went out after one week and the second after the second week (Klingwort et al. 2018, cited in Schnell 2019, p. 298). Online, you can monitor responses continuously and choose the time for the reminder accordingly (Jacob et al. 2019, p. 122). A rule of thumb, not a research finding: send a reminder as soon as the curve of daily responses flattens out.

To whom? Ideally, only to those who haven't responded yet. That requires the system to recognize who has already responded. Jacob et al. advise weighing up whether personalized invitations might, in the eyes of respondents, call the promise of anonymous analysis into question, and doing without them if in doubt (2019, p. 122). If you do invite people personally, briefly explain your response tracking procedure in the invitation, as the same authors recommend for mail surveys (2019, p. 228).

In our example, 230 of the 410 completed questionnaires came in after the invitation, 110 after the first reminder and 70 after the second. The response rate rose from 19.8% to 29.3% and then to 35.3%. Invented numbers, but a typical pattern: each wave brings in less than the one before. In written customer surveys, about 70 to 80% of those willing to respond have usually done so after about ten days (Weinreich and von Lindern 2008, p. 80).

How long? With open recruitment, an online survey should run for at least one full week, so that people who are only online occasionally are reached as well (Lanninger 2001, cited in Möhring and Schlütz 2019, p. 144). With an invitation list and one or two reminders, an online survey can be completed within two to three weeks (Jacob et al. 2019, p. 121). State the end date in the invitation and in every reminder.

Length and design

Ideally, an online questionnaire should take less than 15 to 20 minutes to complete, and the shorter it is, the lower the risk of drop-out (Möhring and Schlütz 2019, p. 147). An attractive layout increases the response rate only slightly, but it does so precisely among people who otherwise wouldn't respond at all (ibid., p. 177). Online, the software shows you after which questions people drop out particularly often, a reason to change the question, delete it or move it further back (Jacob et al. 2019, p. 122). How to plan structure, length and required questions is covered in the article Questionnaire design.

Mobile access and reaching people offline

Many people fill in questionnaires on their smartphones today. According to Möhring and Schlütz, data quality hardly differs by device, even though sliders and date fields are harder to use on smartphones, and completing the questionnaire takes longer (2019, p. 147). Still, test the questionnaire on a phone before you send the invitations.

And think of the people who don't work at a screen. In a case study of an employee survey, a conference room was set up as a kind of polling station for warehouse staff without computers (Wiertalla 2013, p. 181). If some people in your target group lack access to the technology needed to take part, you have an access gap, and that can lower the response rate too.

Trust and anonymity

A promise of anonymity can noticeably lower the refusal rate, especially for sensitive topics (Singer et al. 2003, cited in Möhring and Schlütz 2019, p. 46). In employee surveys, this includes making sure no one finds out who took part and who didn't (Wiertalla 2013, p. 185). Say concretely in the invitation who sees the answers and how they're analyzed. "Your data is safe" on its own convinces no one.

Share the results

Offer to share the results, and then actually do it. Announcing this during the survey increases credibility (Weinreich and von Lindern 2008, p. 81). For recurring surveys, this is the most important investment in participation in the next survey: people who learn that nothing happens often don't take part next time (Linke 2018, p. 21).

What I wouldn't do

I wouldn't publish rankings of participation by team, and I wouldn't make the rate the actual goal. Wiertalla describes a case in which employees complained that achieving a high participation rate was the company's only goal for the survey (2013, pp. 184-185). I wouldn't make participation mandatory: people who wouldn't have taken part out of a lack of trust are then more likely to give socially desirable, overly positive answers. And I wouldn't ask anyone on vacation to take part, because that can do more harm than the marginally higher participation is worth (Linke 2018, p. 56).

Incentives: when a reward pays off

An incentive is something given in return for taking part. Research results are mixed, but the direction is clear: money works better than gifts, incentives provided in advance are the most promising, and the value has to be appropriate so that people don't feel they're being paid (Möhring and Schlütz 2019, p. 47). In the words of Malhotra and Birks: "Prepaid incentives have been shown to increase response rates to a greater extent than promised incentives." (Malhotra and Birks 2007, p. 446). Schnell sums up the state of research with the title of a paper: "Charities, No; Lotteries, No; Cash, Yes" (Warriner et al. 1996, cited in Schnell 2019, p. 173). Unusually high rewards can cause suspicion, and everyone who might be in contact with one another has to be treated the same (Schnell 2019, p. 173).

With customers and employees, caution is advised. Attractive prizes may mainly attract people who want to win, and some may then give more favorable answers (Weinreich and von Lindern 2008, p. 81). An employee survey is not a "prize draw" (Trost et al. 1999, cited in Wiertalla 2013, p. 184).

In my course, I recommend offering incentives in the questionnaire introduction, preferably non-monetary ones, such as sharing the results (Marx 2026, slide 114). For surveys of the general population, on the other hand, the research cited above speaks fairly clearly in favor of money in advance. This is how I read the two together: customers, members or employees who have a connection to the topic are more likely to be won over by purpose and feedback, strangers without a connection more likely by a small, guaranteed reward.

Whether an incentive pays off with your target group is best shown by a trial with two randomly assigned groups. In an invented example with 400 students in each group, the share of completed questionnaires rises from 25% to 35% with a guaranteed voucher worth 5 euros per completed questionnaire, that is, from 100 to 140 completions. After the quality check, 95 and 130 usable questionnaires remain. With the same base costs, this means 140 · 5 = 700 euros in additional costs, that is, 700 / (130 minus 95) = 20.00 euros per additional usable case. Whether that's money well spent depends on who the incentive brings in.

If you run a prize draw, collect the contact details for the draw separately from the survey responses and prevent multiple participation, because without an access code respondents can take part more than once (Möhring and Schlütz 2019, p. 146). How to obtain consent for this is covered in the article Data protection in online surveys.

How to measure responses in QUESTIONSTAR

A word on our own product, but with concrete features. QUESTIONSTAR doesn't increase your response rate by itself. It helps you put the measures above into practice and interpret your participation figures correctly.

The completions widget. In the dashboard, the completions widget shows for every launched survey how many people have started and how many have completed it, for example "6,504 of 8,257 (78%)", along with the trend, channels and devices. Important: the widget lists this value under "Completions", and the percentage is completed out of started, that is, the completion rate, not the response rate. In our example, it would show 410 of 520. You calculate the response rate yourself, with your own denominator: the number of people invited, minus those who demonstrably aren't eligible.

Invitations and reminders. You can invite people by email with personal links and send reminders only to those who haven't responded yet. This also works for anonymous surveys, at both anonymity levels explained in the article Employee survey. Explain the procedure in the invitation, as described above.

Channels for people without an invitation. You can publish a survey directly on social networks, embed it on a website or distribute it by QR code. For employees without a PC, there's kiosk mode on the iPad. For open channels, what's said above applies: there's no response rate there, so report the number of responses per channel.

Preventing multiple participation. In the Access settings, you can make sure the same person can't take part more than once. If you offer incentives, you should definitely switch this on.

With an AI assistant. If QUESTIONSTAR is connected to an AI assistant through the MCP interface, for example in ChatGPT or Claude, the assistant can retrieve the number of responses by date, source and device, and use them to show, for example, the trend after each reminder. A request could look like this:

For my survey "Course satisfaction 2026", show me the completed
responses per day and per source. Mark the days of the invitation and
the reminders (March 3, 10 and 17). I invited 1,200 people, 40 of whom
don't belong to the target group. Calculate the response rate after
each wave with 1,160 in the denominator and show me the calculation.

Only you know the denominator, so include it. If you also let the assistant read responses: questions you've marked as personal data in the survey aren't passed on to the assistant when the setting is switched on. Responses without this marking, especially free text, can still contain personal data. So check beforehand what the assistant will see, and double-check the result.

What QUESTIONSTAR doesn't do: it doesn't pay out incentives. You organize vouchers and prize draws yourself.

Checklist: planning and reporting your response rate

1. Before you start, write down the target group and the denominator: who belongs, who doesn't?
2. Decide what counts as "completed" and what counts as "usable".
3. Write an invitation with the sender, purpose, honest duration, end date, data protection and anonymity.
4. Test the questionnaire on a smartphone and measure the duration in the pretest.
5. Plan two reminders, only to those who haven't responded yet, and explain the procedure in the invitation.
6. Reach groups without email or a screen separately.
7. Decide on an incentive deliberately and, if possible, test it in a small trial.
8. In the report, state the response rate with numerator and denominator, how you handled undeliverables, the completion rate and the number of usable questionnaires.
9. Compare respondents with known characteristics of the target group.
10. Share the results with the respondents.

Frequently asked questions about survey response rates

What is a survey response rate?

The survey response rate is the share of completed questionnaires among all invited people who belong to the target group. German textbooks usually call it the Ausschöpfungsquote, but they often subtract wrong addresses beforehand as neutral losses (Schumann 2019, p. 104). It's more cautious to leave undeliverable invitations in the denominator (AAPOR 2023, p. 27).

How do I calculate the response rate?

Divide the number of completed questionnaires by the number of people invited, minus those who demonstrably don't belong to the target group, and multiply by 100. To be on the safe side, leave undeliverable emails in the denominator, because their eligibility is unknown (AAPOR 2023, p. 27). Example: 410 completed questionnaires with 1,160 invitees of confirmed or unknown eligibility give 35.3%.

What is a good response rate for online surveys?

There's no universal threshold. The rate depends on factors such as the survey mode, the target group, the fieldwork and the length of the questionnaire (Möhring and Schlütz 2019, p. 35). The literature values cited above come from different types of surveys and aren't general online benchmarks. Where possible, compare surveys of the same group with similar recruitment and field period.

What is the difference between response rate and completion rate?

The response rate relates completed questionnaires to all eligible invitees. The completion rate relates them only to those who started the questionnaire. The first tells you something about your fieldwork, the second about your questionnaire. The percentage in the QUESTIONSTAR completions widget is a completion rate as well.

How often should I send survey reminders?

For online surveys, one textbook mentions one or two reminders within a field period of two to three weeks (Jacob et al. 2019, p. 121). Weinreich and von Lindern recommend more than two reminders only in exceptional cases (Weinreich and von Lindern 2008, p. 81). Where possible, remind only those who haven't responded yet.

Are survey incentives worth it?

They often increase the response rate, especially as a small amount of money in advance (Malhotra and Birks 2007, p. 446; Schnell 2019, p. 173). According to the research cited, prize draws have a weaker effect. Whether an incentive pays off depends on the cost per additional usable case and on who it brings in.

Is a survey with a high response rate representative?

Not automatically. A result is biased when respondents and nonrespondents differ, and a response rate alone can't show that (Schnell 2019, pp. 159-164). With the same difference between respondents and nonrespondents, a higher rate makes the bias smaller. Whether that difference actually stays the same is something the rate alone doesn't show.

Sources

AAPOR, American Association for Public Opinion Research (2016). Standard Definitions. Cited in Schnell (2019), p. 151.

AAPOR, American Association for Public Opinion Research (2023). 2023 Standard Definitions. Final Dispositions of Case Codes and Outcome Rates for Surveys (10th ed.). Online, accessed November 10, 2026.

Borg, I. (2000). Früh- versus Spätantworter [Early versus late respondents]. ZUMA-Nachrichten, 24(47), 7-19. In German. Cited in Möhring and Schlütz (2019), p. 139, n. 57.

Bungard, W. (1997). Mitarbeiterbefragung als Instrument modernen Innovations- und Qualitätsmanagements [The employee survey as a tool of modern innovation and quality management]. In W. Bungard and I. Jöns (eds.), Mitarbeiterbefragung [Employee surveys] (pp. 5-14). Weinheim: Beltz PVU. In German. Cited in Wiertalla (2013), p. 182.

Görtler, E. and Rosenkranz, D. (2006). Mitarbeiter- und Kundenbefragung. Methoden und praktische Umsetzung [Employee and customer surveys: methods and practical implementation]. München: Carl Hanser. In German. Cited in Wiertalla (2013), p. 182.

Groves, R. and Peytcheva, E. (2008). The Impact of Nonresponse Rates on Nonresponse Bias: A Meta-Analysis. Public Opinion Quarterly, 72(2), 167-189. Cited in Schnell (2019), pp. 160-161.

Häder, M. (2015). Empirische Sozialforschung. Eine Einführung [Empirical social research: an introduction] (3rd ed.). Wiesbaden: Springer VS. In German. Cited in Möhring and Schlütz (2019), p. 35.

Jacob, R., Heinz, A. and Décieux, J. P. (2019). Umfrage. Einführung in die Methoden der Umfrageforschung [Surveys: an introduction to the methods of survey research] (4th ed.). Berlin/Boston: De Gruyter Oldenbourg. In German.

Klingwort, J., Buelens, B. and Schnell, R. (2018). Early Versus Late Respondents in Web Surveys: Evidence from a National Health Survey. Statistical Journal of the IAOS, 34(3), 461-471. Cited in Schnell (2019), p. 298.

Lanninger, T. (2001). Methodentests als Basisvoraussetzung für Online-Forschung: Stichprobeneffekte im Split-Half-Design CATI - Online [Method tests as a basic requirement for online research: sample effects in a split-half design, CATI vs. online]. planung & analyse, (6), 50-54. In German. Cited in Möhring and Schlütz (2019), p. 144.

Linke, R. (2018). Mitarbeiterbefragungen optimieren. Von der Befragung zum wirksamen Management-Instrument [Optimizing employee surveys: from survey to effective management tool]. Wiesbaden: Springer Gabler. In German.

Malhotra, N. K. and Birks, D. F. (2007). Marketing Research. An Applied Approach (3rd European edition). Harlow: Pearson Education.

Marx, P. (2026). Principles of Survey Research. Online course, slide 114 (chapter 3) and slide 136 (chapter 4). QUESTIONSTAR. Course, chapter 3.

Möhring, W. and Schlütz, D. (2019). Die Befragung in der Medien- und Kommunikationswissenschaft. Eine praxisorientierte Einführung [Surveys in media and communication studies: a practical introduction] (3rd ed.). Wiesbaden: Springer VS. In German.

Reuband, K.-H. (2001). Möglichkeiten und Probleme des Einsatzes postalischer Befragungen [Possibilities and problems of using mail surveys]. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 53, 307-333. In German. Cited in Möhring and Schlütz (2019), p. 139.

Schnell, R. (2019). Survey-Interviews. Methoden standardisierter Befragungen [Survey interviews: methods of standardized surveys] (2nd ed.). Wiesbaden: Springer VS. In German. DOI 10.1007/978-3-531-19901-6.

Schumann, S. (2019). Repräsentative Umfrage. Praxisorientierte Einführung in empirische Methoden und statistische Analyseverfahren [Representative surveys: a practical introduction to empirical methods and statistical analysis] (7th ed.). Berlin/Boston: De Gruyter Oldenbourg. In German.

Singer, E., van Hoewyk, J. and Neugebauer, R. J. (2003). Attitudes and behavior: The impact of privacy and confidentiality concerns on participation in the 2000 Census. Public Opinion Quarterly, 67, 368-384. Cited in Möhring and Schlütz (2019), p. 46.

Stocké, V. and Becker, B. (2004). Determinanten und Konsequenzen der Umfrageeinstellung: Bewertungsdimensionen unterschiedlicher Umfragesponsoren und die Antwortbereitschaft der Befragten [Determinants and consequences of survey attitudes: how respondents evaluate different survey sponsors, and their willingness to respond]. Universität Mannheim. In German. Online. Cited in Möhring and Schlütz (2019), p. 46.

Trost, A., Jöns, I. and Bungard, W. (1999). Mitarbeiterbefragung [Employee surveys]. Augsburg: WEKA. In German. Cited in Wiertalla (2013), pp. 182 and 184.

Warriner, K., Goyder, J., Gjertsen, H., Hohner, P. and McSpurren, K. (1996). Charities, No; Lotteries, No; Cash, Yes: Main Effects and Interactions in a Canadian Incentives Experiment. Public Opinion Quarterly, 60(4), 542-562. Cited in Schnell (2019), p. 173.

Weinreich, U. and von Lindern, E. (2008). Praxisbuch Kundenbefragungen [Customer surveys: a practical handbook]. München: mi-Fachverlag, FinanzBuch Verlag. In German.

Wiertalla, K. (2013). Der Einsatz von Mitarbeiterbefragungen als Lerninstrument in Unternehmen. Dargestellt anhand des Praxisbeispiels der Mitarbeiterbefragung bei der Ingram Micro Distribution GmbH [Employee surveys as a learning tool in companies, illustrated by the employee survey at Ingram Micro Distribution GmbH]. Dissertation, University of Siegen. In German.

Zerr, K. and Müller-Schneider, T. (2002). Postalische Vorankündigung bei Online-Befragungen? Ein projektbegleitendes Methodenexperiment zur Erhöhung der Ausschöpfungsquote [Advance letters for online surveys? A methods experiment alongside a project to increase the response rate]. planung & analyse, (4), 46-59. In German. Cited in Möhring and Schlütz (2019), p. 46, n. 18.

I hope each reminder brings another rise in responses, and that you feel confident reporting even a modest rate with a clearly defined denominator.

Related articles:

• Calculating sample size: technical details explained simply
• Where do I find respondents for my survey
• Representative sample: what it is and how to check yours

Dr. Paul Marx

Author: Dr. Paul Marx
Dr. Paul Marx is the founder and Managing Director of QUESTIONSTAR and was a professor of marketing at the University of Siegen.