CHERRIES 清单：全部 30 项内容应用于真实的开放获取网络调查

发布时间 2026 年 11 月 1 日

此 CHERRIES checklist, short for Checklist for Reporting Results of Internet E-Surveys, is the reporting guideline for questionnaires that people complete on the web. Gunther Eysenbach published it in the Journal of Medical Internet Research in 2004 as a table of 30 items in eight categories, and a 2012 correction fixed the formulas for three response measures.^1,2 We found no revised or extended version, so those two papers remain the working record.

Most guides stop at listing the items. This guide goes further. After the checklist itself, we score all 30 items against a real open-access web survey, published in JMIR Cancer in 2026, that states it followed CHERRIES, with the article section beside every verdict.³ The scoring shows where a paper that cites the CHERRIES checklist still leaves readers unable to find an answer, how easily the three response rates are mislabelled, and how many answers sit only in an appendix.

What the CHERRIES Checklist Is and What Changed Since 2004

Eysenbach defined an e-survey as an electronic questionnaire administered on the Internet or an Intranet. The CHERRIES checklist focuses on web-based surveys, although many items also suit e-mail surveys.¹ Its aim is to let reviewers and readers see how the sample was selected and how it may differ from a representative one. Bias from a non-representative internet population and from self-selection is the problem the CHERRIES checklist is built around, and the paper says that conclusions drawn from a convenience sample need to be qualified in the discussion.¹ When the statement appeared, JMIR told authors that its editors and peer reviewers expected web survey reports to follow it.¹

CHERRIES is a reporting checklist and not a quality score. The CHERRIES checklist asks whether a reader can find out how the survey ran, and it does not judge whether the survey was good. The original paper is blunt about response measures, noting that "there is no single response rate" in online surveys, and it advises authors to avoid the term altogether.¹ The checklist is listed on the EQUATOR 网络 网站。

The 2012 correction and other versions

In 2012 the author corrected Table 1 of the original paper. The correction notice says that, in the recommendations on how view rate, participation rate and completion rate should be calculated, "denominators and numerators were flipped".² It also warns that "incorrect versions may exist on other sites". The Europe PMC copy of the 2004 paper that we read carries the corrected formulas, and Table 2 below uses them. Authors who copied the CHERRIES checklist from a repository or a slide deck should check which version they hold.

We searched PubMed, Europe PMC and Crossref on 1 November 2026 and found no revised, extended or numbered second version of CHERRIES. Related items include two that matter here. A Spanish adaptation of the checklist appeared in 2019, and a broader consensus-based checklist for survey studies, CROSS, was published in 2021.^4,5 Neither replaces the original CHERRIES checklist, so authors can still cite CHERRIES, as the example in this guide does. A 2026 survey of European residents in JMIR mHealth and uHealth, for instance, reports following CHERRIES for the online survey and STROBE for the cross-sectional design.⁶

The 30 CHERRIES Checklist Items in Eight Categories

Table 1 groups the CHERRIES checklist items as the original table does, with shortened category names. We paraphrase here, and the full wording of each item with its explanation sits in Table 1 of the open-access original, which is licensed CC BY 2.0.¹

Table 1. The eight CHERRIES categories and what a reviewer checks. Content summarised from Eysenbach, 2004.

Category 5 carries 11 of the 30 items, so most of the work sits in describing the administration of the survey. Two items in the CHERRIES checklist hold advice that authors often miss. The completeness check item recommends that every question offer a non-response option such as not applicable or rather not say, with one option required. The unique site visitor item says that if view or participation rates are given, the way a unique visitor was identified must be defined, whether by IP address, cookie or both.¹ Figure 1 shows the eight categories in one view.

Response Rates in the CHERRIES Checklist After the 2012 Correction

The response-rate block is the part of the CHERRIES checklist most often misapplied, because online surveys have no single denominator. The CHERRIES checklist defines three measures instead, and Table 2 gives them in their corrected form.^1,2

Table 2. View, participation and completion rates as corrected in 2012.

A short hypothetical shows the arithmetic. Suppose 10,000 unique visitors reach a site, 800 open the first survey page, 500 agree to take part and 400 submit the last page. The view rate is 800 / 10,000, or 8.0%. The participation rate is 500 / 800, or 62.5%. The completion rate is 400 / 500, or 80.0%. The original paper adds that a view rate below 0.1% is not unusual for a voluntary survey, and that completion can involve leaving items blank, so authors who want to describe how fully forms were answered should call that a completeness rate.¹

Closed surveys raise a practical problem. If invitations go only to a panel or a mailing list, there may be no public site visitors to count, and a view rate may be undefined. The sensible course is to say so in the Methods, give the counts that do exist, and name each ratio by what its numerator and denominator really are. The worked example below shows what happens when that discipline slips.

Choosing the CHERRIES Checklist Alongside STROBE and Other Guidelines

A cross-sectional web survey is also an observational study, so journals may ask for the STROBE 指南 as well. The two cover different ground. STROBE addresses design, variables, bias and statistical methods, while CHERRIES addresses how the electronic questionnaire reached people and how entries were counted. If you are unsure which guideline fits your design, our 选择报告准则的决策树 提出选择。

Other guides in this series use the same row-by-row method on other designs, for example the PRISMA-DTA 工作示例 for diagnostic accuracy reviews. If patients helped to shape the questionnaire, the GRIPP2 short form covers how to report that involvement. Before fielding, two Editverse guides cover the instrument itself, namely survey design and questionnaire writing 和 questionnaire design in epidemiological studies. The consent and approval items link naturally to our guide on informed consent for transparent research. Students whose survey sits inside a thesis can see how guidelines map onto chapters in our 医学论文报告指南 指南。

Worked Example of the CHERRIES Checklist on a Real Open-Access Web Survey

我们评分的文章

We chose Hirayama, Kasai and Tabei, published in JMIR Cancer on 28 May 2026 under the title “Information-Seeking Patterns and Communication Preferences Among Japanese Survivors With Cancer”, with the subtitle Cross-Sectional Analysis.³ The authors surveyed 350 Japanese cancer survivors recruited from two online panels in December 2024, then used exploratory factor analysis and latent class analysis to describe three groups. The questionnaire ran over 21 screens and 52 items. The abstract calls it a CHERRIES-compliant web survey, the Methods say it was "conducted on the basis of the CHERRIES", and the authors supply their completed form as Checklist 1.

The licence is CC BY 4.0. The Europe PMC licence field reads cc by, the PMC open-data record gives the licence code CC BY, and the article XML states that the article is distributed under the Creative Commons Attribution License 4.0. Crossref returns no licence element for this DOI, so that source is silent and we relied on the other three. The JMIR page we read points to the 2.0 version in its general licence wording, which differs in version from the 4.0 statement in the article XML. Both are CC BY, and we quote at most 25 words at a time with attribution.

We reproduce no figure from the article, because its Figure 1 is a 799-pixel image. Our Figure 2 redraws the counts from the text and the flowchart instead.

我们如何评分

Two points about fit. We score what a reader can find in the article and in Checklist 1, the authors’ completed CHERRIES checklist. We confirmed that the Checklist 1 file we read is identical, by checksum, to the supplement held in PMC. Where an answer sits only in Checklist 1, the table says so. A row marked Not reported also means we could not find the information, and the authors may have recorded it elsewhere.

每一行代表四种判决之一。 报道 意味着信息完整清晰。 部分报道 means some of it is present. 判断 这意味着理性的读者可能会对这一行做出两种不同的评价，我们会解释我们的解读。 未报告 means we found nothing, and in this example both such rows are disclosed by the authors as unavailable.

Table 3. CHERRIES checklist applied to Hirayama et al., JMIR Cancer 2026 (CC BY 4.0). Quotations are verbatim and no longer than 25 words.

Across 30 rows the tally is 23 Reported (9 of them found only in Checklist 1), 2 Partly reported, 3 Judgement call and 2 Not reported. The paper is strong on design, ethics, questionnaire mechanics and the incentive schedule. The weak spots are the response measures and the labels attached to them, and a good deal of the evidence sits in an appendix and not in the article body.

Where the paper and its checklist do not line up

The points below come from our own arithmetic and reading, and several are probably typographical. We list them because a reviewer or reader would meet the same questions.

• Age range. The Methods give eligibility as 20 to 89 years and Table 1 lists five respondents aged 80 to 89, whereas Checklist 1 says 20 to 80.
• 公司名。 The panel company appears as Asmarc, Asmark and Asmarq across the Methods, Figure 1 and Checklist 1. The survey system is called ORCA in Checklist 1 and is not named in the article, which names the D-style web and Monitas platforms.
• Final yield. 350 / 1,095 is 31.96%, which rounds to 32.0% on our arithmetic, while the article prints 31.9%. The difference is probably truncation.
• Table 1 columns. In the Area rows, Chugoku and Shikoku shows 22 in total with 19 Yes and 2 No, and the No column then sums to 67 and not 68. A No count of 3 would also match the printed 0.9%. In the device rows, Other shows 32 in total with 14 Yes and 8 No, and 18 would match both the printed 5.1% and a No column that sums to 68. We compared the Area rows on the publisher page, which match the Europe PMC copy.
• Figure 1 exclusions. The exclusion counts in the flowchart are 17, 5, 49, 174 and 223, which sum to 468. That equals 1,095 minus 627, and the Results say that 350 of the 627 screening respondents met the criteria, which leaves 277. The flowchart labels also read within 10 years and more than 11 years, whereas the Methods exclude those more than 10 years after treatment and the abstract requires treatment to have ended within 5 years. A reader cannot tell from the figure which step the exclusion counts belong to.

这个例子告诉我们什么

Name each rate by its CHERRIES definition. The paper reports 627 of 1,095 (57.3%) as a participation rate, 350 of 627 (55.8%) as a transition rate and 350 of 1,095 as a final yield. None of the three is the corrected view, participation or completion rate, because the panel list is not a count of unique site visitors and the first-page count is missing. Figure 2 lays the counts out. Relabelling the first ratio a screening response rate would have kept the arithmetic and avoided a clash with the checklist.

Say in the body when a number cannot be known. The view rate and unique site visitors are missing, and the reason, an expired data retention period, appears only in Checklist 1. A single sentence in the Methods would let a reader judge the sample without opening an appendix.

Bring the CHERRIES checklist answers into the paper. Nine rows were answered only in Checklist 1, among them the closed design, the contact mode, the review step and the absence of cookies. Editors and readers who see only the article cannot find them. A short survey-methods paragraph or a small table with the item names costs little and removes the problem.

Keep numbers and names consistent. The three spellings of the panel company, the two age ranges and the table cells that do not sum are the kind of detail reviewers check first. Reconcile every count, date and name across the abstract, text, tables, figure and checklist before submission, and see our guides on 调查数据分析 and on how survey weights change an estimate for what comes after fielding.

How to Write Your Own CHERRIES Checklist Statement

A CHERRIES checklist statement can be a single short paragraph per category, placed in the Methods or in a supplementary table, with a pointer in the Results where a rate is reported. Adapt the skeleton below to any web survey.

• 设计。 State the target population and sample frame, and say plainly if the sample is a convenience sample.
• IRB approval and consent. Give the approving body and number, what respondents were told on the consent screen and how personal data were protected.
• Development and pre-testing. Say how items were chosen, who reviewed them and what was tested before launch, such as display, skip logic and usability.
• 招聘。 Say whether the survey was open or closed, how first contact was made and where it was announced, and put the announcement text in an appendix.
• Survey administration. Give the platform, dates, incentives, items per page, number of screens, randomisation, adaptive questions, completeness checks and whether respondents could go back.
• 响应率。 Define a unique visitor, then give view, participation and completion rates with their numerators and denominators, or state that a measure is unavailable and why.
• Preventing multiple entries. Report cookies, IP checks, log file analysis and registration, the time window used and which entry was kept.
• 分析。 Say how incomplete forms and very fast submissions were handled and whether any weighting or other correction was applied.

Common mistakes with the CHERRIES checklist

• Calling the share of a panel list who answered a participation rate when the denominator is not first-page visitors.
• Using the flipped 2004 formulas from an uncorrected copy of the checklist.
• Leaving answers in a supplementary checklist that the article body never mentions.
• Reporting a duplicate-entry rule without the time window or the entry that was kept.
• Letting counts, dates and names drift apart between the abstract, the text, the tables and the checklist.

Professional Support for Web Survey Reporting

Most CHERRIES items are decisions made before the first invitation goes out, such as how visitors will be counted, whether IP addresses will be logged and which entry is kept when two arrive. Two Editverse services map onto that. A methodology consultation suits the planning stage of a survey study, covering sampling, data collection and validity. Substantive editing suits the write-up, including the survey methods paragraph. Editverse editing and writing support aligns with ICMJE，COPE 和 GPP 2022并遵守您所在机构和国家的道德规范要求。^7,8

结语

The CHERRIES checklist asks 30 things of any paper that reports a web survey, grouped in eight categories from design and consent to response rates and analysis. Use the 2012 corrected formulas for the three response measures, and say plainly when one of them cannot be calculated. Our worked example shows what scoring against the CHERRIES checklist looks like in practice. A recent open-access paper can describe its questionnaire mechanics, incentives and skip logic clearly and still leave a reader unsure how its rates were defined, where its exclusions went and why its tables do not sum. Plan the survey with the eight CHERRIES checklist categories in view, write one short paragraph for each, and run a consistency pass before you submit.

常見問題解答

How many items does the CHERRIES checklist have?

Table 1 of the 2004 paper lists 30 items in eight categories, on our count. Category 5, survey administration, holds 11 of them.

What does CHERRIES stand for?

Checklist for Reporting Results of Internet E-Surveys. It was published by Gunther Eysenbach in the Journal of Medical Internet Research in 2004.

Has the CHERRIES checklist been updated?

A 2012 correction fixed the formulas for view rate, participation rate and completion rate. In our searches of PubMed, Europe PMC and Crossref on 1 November 2026 we found no revised or extended version. A Spanish adaptation (2019) and the broader CROSS checklist (2021) exist as separate items.

How are view rate, participation rate and completion rate calculated?

View rate is unique survey visitors divided by unique site visitors. Participation rate is those who agreed to take part divided by unique first-page visitors. Completion rate is those who finished divided by those who agreed.

Do authors need CHERRIES and STROBE together?

There is no single rule, because the two cover different ground. CHERRIES covers the mechanics of the web survey and STROBE covers observational design, so some authors report both. Check the journal instructions.

What should authors do when the view rate cannot be calculated?

State in the Methods that the measure is unavailable and give the reason, for example a closed panel with no public site visitors. Report the counts that do exist and label each ratio by its real numerator and denominator.

Where is the official CHERRIES checklist published?

It is Table 1 of the open-access 2004 paper in the Journal of Medical Internet Research, with the 2012 correction published alongside it, and it is listed on the EQUATOR Network website.

案例

1. Eysenbach G. Improving the quality of web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). J Med Internet Res. 2004;6(3):e34. doi:10.2196/jmir.6.3.e34. PMID: 15471760. PMCID: PMC1550605. CC BY 2.0. PMC全文.
2. Eysenbach G. Correction: Improving the Quality of Web Surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). J Med Internet Res. 2012;14(1):e8. doi:10.2196/jmir.2042. PMCID: PMC4220447. PMC全文.
3. Hirayama M, Kasai D, Tabei K. Information-Seeking Patterns and Communication Preferences Among Japanese Survivors With Cancer: Cross-Sectional Analysis. JMIR Cancer. 2026;12:e79065. doi:10.2196/79065. PMID: 42207161. PMCID: PMC13218279. CC BY 4.0. PMC全文.
4. Sharma A, Minh Duc NT, Luu Lam Thang T, Nam NH, Ng SJ, Abbas KS, et al. A Consensus-Based Checklist for Reporting of Survey Studies (CROSS). J Gen Intern Med. 2021;36(10):3179-3187. doi:10.1007/s11606-021-06737-1. PMID: 33886027. PMCID: PMC8481359. PMC全文.
5. Lopez-Rodriguez JA. Improving the quality of Spanish web surveys: Spanish adaptation of the checklist for reporting results of internet e-surveys (CHERRIES) to the Spanish context. Aten Primaria. 2019;51(9):586-589. doi:10.1016/j.aprim.2019.03.005. PMID: 31060751.
6. Shokralla M, Willemsen RF, Kasteleyn MJ, Chavannes N, Talboom-Kamp E. Trust and Use of Recommendations for Health Apps Among European Residents: Cross-Sectional Survey. JMIR Mhealth Uhealth. 2026;14:e64468. doi:10.2196/64468. PMCID: PMC13064958. Facts only are cited.
7. 国际医学期刊编辑委员会。《医学期刊学术著作的撰写、报告、编辑和出版建议》。可访问 icmje.org/recommendations 获取。
8. DeTora LM、Toroser D、Sykes A、Vanderlinden C、Plunkett FJ、Lane T 等。公司赞助生物医学研究的良好出版规范 (GPP) 指南，2022 年更新。《内科年鉴》。2022;175(9):1298-1304。doi:10.7326/M22-1460。PMID: 36037471。