How to Promote a Survey Study Without Overstating the Numbers
A practical guide to sharing survey findings so the headline number travels with the context it needs. It covers what to disclose, how margins of error really work, opt-in samples, a worked example and a checklist.
Your survey paper is out, and one number in it is the kind people share. Forty percent of something. Twice as many of one group as another. You can already picture the post.
That number is the most quotable thing you have ever published, and also the easiest to misread. Once it leaves the paper, it travels without the sample, the dates and the question wording that gave it meaning. Promoting a survey study well means sending the context along with the number.
Why Survey Findings Need Extra Care
Most findings are hard to quote out of context. A single percentage is the opposite: it fits in a headline, a slide or a reply, and readers fill in the rest themselves. They will usually assume it describes everyone, measured precisely, right now.
Survey researchers already have a standard for this. The AAPOR Code of Professional Ethics and Practice asks researchers to disclose enough about how the research was done to allow independent review and verification of research claims, at the time results are released. Its FAQ says the disclosure requirement applies to surveys whose results are released to the public, and that the Code represents ethical standards for all public opinion researchers, members or not.
The good news for a busy author: you do not have to fit the whole method into a post. The same AAPOR FAQ says summaries and shorter reports can refer to the full report for the details of the method, and that public reports should indicate, whenever practical, where the description of method can be obtained. For you, that full report is the paper. Link it every time.
What a Reader Needs Next to the Number
The table below pairs the kinds of claims people post about surveys with the context that keeps them honest. Each row comes from a source opened while writing this guide.
| What you want to post | What should travel with it | Where that comes from |
|---|---|---|
| A headline percentage | Who was surveyed, when, how many answered, and whether the sample was probability-based or opt-in | The AAPOR Code lists population, sample method, dates and sample sizes for immediate disclosure |
| A margin of error | For probability samples, whether it was adjusted for the design effect of weighting; for opt-in samples, no margin unless the paper defines and validates one | The AAPOR Code sets both conditions |
| "Group A is higher than group B" | Whether the gap itself is larger than its own margin of error, which is bigger than the one for each group | Pew Research Center shows the margin for a difference is larger than for each estimate (about twice, for two candidates in the same poll) |
| A finding about a subgroup | How many people the subgroup estimate rests on | Pew shows how much wider subgroup margins get |
| A quote of what people "agree" with | The exact question wording and any text before it that could shape answers | The AAPOR Code asks for exact wording and preceding context |
| A surprising rare attitude or behavior | What the paper did to screen out careless or fraudulent responses | Pew Research Center on bogus respondents in opt-in polls |
Three of those rows trip up careful researchers most often.
The margin of error is smaller than people think it covers
Pew Research Center's explainer on the margin of error describes a margin of plus or minus 3 percentage points at the 95% confidence level as meaning that, if the same survey were fielded 100 times, you would expect the result to be within 3 points of the true population value 95 of those times. The margin for the difference between two candidates' levels of support in the same poll is generally about twice as large, and any gap between two estimates has a larger margin than either estimate alone. Subgroups have fewer cases, so their margins are larger, in some cases much larger. In Pew's example, a simple random sample of 1,067 cases has a margin of plus or minus 3 points overall, while a subgroup of about 160 cases has a margin of plus or minus 8 points, and plus or minus 16 points for the difference between two candidates within that subgroup.
The same piece notes that weighting, while crucial for avoiding biased results, also makes the margin of error larger, and that sampling error is only one of many possible sources of error. Question wording, mode and who could not be reached all matter too.
Opt-in samples need their own caveat
The AAPOR Code asks researchers to state explicitly whether the sample was probability-based or came from opt-in, volunteer or other sources. It also says reports of non-probability surveys should only give measures of precision if they are defined and accompanied by a detailed description of how the underlying model was specified, its assumptions validated and the measure calculated. If your paper used an opt-in panel and did not do that, a "plus or minus 3 points" in your post claims more than the paper does.
Rare, surprising results deserve a second look
In a 2024 analysis, Pew Research Center described bogus respondents in online opt-in polls, who try to complete surveys with as little effort as possible to earn rewards, and cited studies showing they can cause opt-in surveys to overestimate rare attitudes and behaviors. Pew found that this kind of overreporting tends to be especially concentrated in estimates for adults under 30 and Hispanic adults. In one Pew experiment, 12% of opt-in respondents under 30 claimed to be licensed to operate a nuclear submarine. When Pew repeated a widely reported opt-in question on its probability-based panel, 3% of adults under 30 agreed that "The Holocaust is a myth", rather than the 20% the opt-in poll had reported.
None of this means opt-in surveys are useless; Pew notes they can perform similarly to traditional approaches for some purposes. It means your most surprising number is the one to post most carefully, alongside the quality checks your paper reports.
A Worked Example
Say, as a hypothetical, that you published a psychology paper from an online opt-in survey of 1,200 university students, and one result is that 41% of first-year students reported sleeping fewer than six hours on most weeknights.
- Weak post: "41% of students sleep less than six hours a night (±3%). Universities need to act." It drops the population, implies a probability-sample margin the paper may not support, and stretches first-year students to all students.
- Better post: "In our survey of 1,200 students at four universities (an online opt-in sample, spring term), 41% of first-year respondents said they slept fewer than six hours on most weeknights. It is not a national estimate, and the paper explains how we screened out inattentive responses. Question wording and full methods are in the paper."
The better post is a little longer and much harder to misquote. It also gives a journalist or a policy reader exactly what they need to decide whether to use it. If your audience includes policymakers, the guide on social science research for policymakers covers that side, and the LinkedIn post templates help with structure.
A Checklist Before You Share
- Pick the one number you most want people to remember, and write it with its population and dates.
- Say whether the sample was probability-based or opt-in.
- Include a margin of error only if the paper reports one that fits that design.
- Check that any gap between groups holds up on its own, not just against each group's margin.
- Give the size of any subgroup you quote.
- Quote the question wording when the exact words could change how the result reads.
- Link the paper in every post, as the place where the full method lives.
Where Amplify Fits
The hard part of promoting a survey study is keeping the context attached to the number in every format people actually see. Amplify turns the PDF you upload into a video abstract (16:9, about 3 minutes), a vertical short (9:16, 45 to 60 seconds), an audio brief and a one-page print-ready infographic, generated from your paper and nothing else, with no open-web filler. Every artifact carries the title, the authors and a link back to the paper of record, so the full method is always one step away.
Each format arrives as three variants you can edit line by line or regenerate, so you can make sure the sample, the dates and the right caveat sit next to the headline number. A subject-matter reviewer reads every draft before it can leave your workspace, every claim is traced to a page in the PDF, and nothing publishes until you approve it.
The embeddable widget puts every format beside the paper on your faculty profile or lab site, where readers can read, listen and chat with the paper without leaving the page. Analytics show reads, listens, questions asked, languages and which format moved people, which is useful for grant reports and broader-impact sections. You keep ownership of your work, ResearchBunny does not train on it, and you can delete a paper and its artifacts at any time.
A survey number will be quoted without you in the room. Give it its context before it leaves, and it will hold up wherever it lands.
Quick answers
What should I include when I share a survey result on social media?
Give the number with its context: who was surveyed, when, how many people answered, and whether the sample was probability-based or opt-in. AAPOR's disclosure FAQ says short reports can refer to a full report for method details, and that public reports should indicate, whenever practical, where the description of method can be obtained, so link to the paper.
Can I report a margin of error for an opt-in online survey?
Only with care. The AAPOR Code says reports of non-probability sample surveys will only provide measures of precision if they are defined and accompanied by a detailed description of how the underlying model was specified, its assumptions validated, and the measure calculated. If your paper does not do that, leave the margin of error out of your posts.
Is a difference between two groups real if it is bigger than the margin of error?
Not necessarily. Pew Research Center explains that the margin of error for a difference is larger than the margin for each estimate; for two candidates in the same poll it is generally about twice as large. Subgroups have fewer cases, so their margins are larger, sometimes much larger. Check the comparison in your paper before you post a gap as a finding.
Why do opt-in surveys sometimes produce surprising results?
One reason is bogus respondents, who answer with as little effort as possible to earn a reward. Pew Research Center found that this kind of overreporting tends to be concentrated among adults under 30 and Hispanic adults, and notes that studies have shown bogus respondents can make opt-in surveys overestimate rare attitudes and behaviors.
