About the ARRIVE compliance checker - how it works and FAQs
The ARRIVE compliance checker is a tool to identify when information is missing from your manuscript, helping to ensure that you are describing you research transparently and comprehensively. This increases the impact of your studies and improves the reproducibility of animal research as a whole.
On this page
How the ARRIVE compliance checker works
Step 1. Uploading text
First you need to create an ARRIVE compliance checker account. Once this is done you can quickly assess a manuscript for adherence to the ARRIVE Essential 10 by either copying and pasting the methods and results text from a manuscript into the text upload box or by uploading a PDF.
If you are copying and pasting text we recommend after copying your text that you paste by right clicking in the text upload box and select 'paste as plain text'. The tool can only process plain text, so text such as formulas and special characters (e.g. the degree symbol) will not be processed.
If a PDF is uploaded a text extractor, known as a parser, identified the methods and results sections of the manuscripts and reformats the information in these sections as plain text so it can be analysed by the tool. You must verify that the relevant text has been extracted correctly from the methods and results sections of the paper. Text in tables, figures and figure legends is not extracted. The PDF parser cannot extract mathematical formulas and, depending on how the PDF was created, may struggle with special characters such as Greek letters and mathematical symbols. Special characters such as the degree symbol are not extracted.
Step 2. Submitting and processing
Once your manuscript text has been pasted in or extracted the text is sent to the SciScore server in the USA where it will be analysed for statements that indicate reporting of the ARRIVE Essential 10 items. The tool first checks for the use of animals in the text by searching for either and RRIS, an NCBI taxonomy ID, or an ethical review statement from an IACUC, AWERB or equivalent. If the text does not include these statements, a report may not be generated. The tool then uses natural language processing as well as specific phrases to identify compliance. ARRIVE sub-items are then marked as either 'detected' or 'not detected'. The tool has been trained on hundreds of statements from across the scientific literature, but some information is not prevalent enough in animal research (e.g. reporting of the confidence interval and effect size) to train the tool to accurately detect it. These sub-items require manual review to assess compliance.
Step 3. Report
Once the ARRIVE compliance checker has finished assessing your paper, a report is generated. This is available from withing the tool and a link to it is delivered to your email inbox (to the email address you used to create your account). You can read the summary on the first page of the report to quickly identify areas where reporting can be improved and then use the full report for advice on how to fulfil the requirements or each item and move towards more comprehensive and transparent reporting.
Limitations
Not all sub-items are able to be assessed with the tool and these are clearly marked as 'Not analysed' in the report. The tool will mark something as 'detected' so long as there is at least a single instance of this information being reported. Researchers need to verify that this information is reported for every experiment in the manuscript. To enable the binary responses used in the report (detected versus not detected), only part of the information is needed for some sub-items to be labelled as detected. For example, when checking whether the manuscript states if randomisation was or was not used to allocate experimental units to groups (item 4a), if there is a statement that mentions that randomisation was, explicitly was not, used to allocate animals to groups the tool marks this as information 'detected', even if there is no description of the method used to create the randomisation sequence.
Frequently Asked Questions
The tool is designed to assess scientific manuscripts that describe in vivo animal research. It may not function correctly with other types of papers, for example manuscripts describing human or in vitro research, or for non-scientific publications.
Yes. Submitted text is stored only as long as needed to generate your report, and the report generated by the ARRIVE compliance checker is deleted from the system after 10 days (giving you a chance to download it). It is not used for training the tool or shared to parties outside of NC3Rs and SciCrunch (SciCrunch is the organisation that developed and maintains the tool). See our Privacy Policy for details.
The tool uses plain text to perform the compliance check. You can submit plain text by pasting or writing text directly into the submission box. You can also upload a PDF document, the tool will extract text from the methods and results sections as plain text. Check the extracted text before submission to verify that the methods and results information has been extracted. The tool can more accurately detect information copied and pasted in to the text up load box compared to text extracted from a PDF. This is because different PDFs are coded different ways, the can affect the performance of the tool that extracts text from the PDF.
See question on text formats for more information on plain text.
The tool extracts text from the methods and results sections only. It does not extract text from other parts of the paper, such as the introduction, discussion or references, to ensure that the information it detects when looking for adherence with the ARRIVE Essential 10 relates to the experiment being described in this paper rather than information from other experiments which might be included to put the current work in context.
Reports are typically generated withing two to three minutes. A link to your report will appear in your browser and the system will also send you an email containing a link to the report. During peak times, it may take longer.
Check your junk of spam folder in case the email has been filtered into there.
If the email is not in your junk or spam folder and a link to the report has not arrived in your email inbox within 30 minutes, try resubmitting. Ensure your text includes indicators of animal research (e.g. the full species name and a species NCBI taxonomy ID e.g. Mus musculus NCBI:txID10090, an RRID for the species/line or an ethical approval statement from an IACUC, AWERB or equivalent).
If you still do not receive an email it could be blocked by your institutional firewall or filters. Please contact your IT department and ask them to whitelist emails from arrive@nc3rs.org.uk.
The report highlights areas where the manuscript could be improved based on the ARRIVE Essential 10. Use this guidance to revise the text and include clear statements on the experimental design used in all in vivo studies described. Guidance given in the report covers both when information was detected in your manuscript and when it was not. This will help you to maximise the quality of your descriptions of experiments and results. More examples of transparent reporting can be found on the ARRIVE guidelines webpages for individual items and sub-items.
In the current literature there are not enough examples of good reporting for some ARRIVE items to enable the tool to be trained, such as reporting of effect size and confidence interval (ARRIVE item 10b). Future updates to the tool will aim to expand coverage to enable quick and efficient assessment of ARRIVE compliance for more ARRIVE guidelines items.
If you uploaded a PDF, the extracted text might not have included the relevant information (we recommend you check text extracted from PDFs before submitting to the ARRIVE compliance checker). The tool may also miss information if phrasing is unusual or if the information is only in tables, figures or figure legends.
If you have copied and pasted text from your manuscript, try rewording using explicit terminology (e.g. the experimental unit was...) and visit the examples pages for the relevant ARRIVE item(s) for example statements from the published literature.
No. Only text from one manuscript at a time is supported at this time. For best results, each text submission should represent one animal experiment. Where multiple experiments are described in the text, you should check that reporting is comprehensive for all expeirments.
Plain text is text including letters, numbers and punctuation but no formatting (such as bold, italics, colours or specific fonts). Plain text includes some 'invisible' characters that define how the text is arranged (such as space, line breaks and tabulation characters).
The ARRIVE compliance checker can only process plain text. If you submit text that is a combination of plain and non-plain text the compliance checker will not process the non-plain text and the report may not be accurate. If you submit text that does not contain any plain text a report will not be generated.
In most programs or documents you can select text and copy it (for example using the keyboard shortcut CTRL + C on a Windows device or CMD + C on a Mac) and then paste it into the text upload box in the ARRIVE compliance checker using right click and selecting ‘Paste as plain text’ or by using keyboard shortcuts (for example CTRL + SHIFT + V on a Windows device or OPTION + SHIFT + CMD + V on a Mac).
Alternatively, you can copy text from your document and paste it into a plain text document (such pasting into a document in ‘Notepad’ on a Windows device or ‘TextEdit’ on a Mac and using Choose Format > Make Plain Text), this strips out the formatting. The text can then be simply copied and pasted from the text document into the ARRIVE compliance checker text upload box.
Different journals build PDFs in different ways and sometimes the PDF extractor will extract the beginning of a new column of text before it extracts all of the text from the previous column. This is unlikely to affect the accuracy of the compliance check.
The ARRIVE compliance checker does not need the information to be in an order logical to a human reader to be able to check for ARRIVE information, so this is unlikely to affect the results of the compliance check. If text is extracted with a break in a key phrase that the tool is looking for, this could mean the tool does not detect that the item was reported.
The ARRIVE compliance checker tool cannot process symbols, this means they will not be present in the report.
Email us at arrive@nc3rs.org.uk for further assistance.
Feedback welcome
This is a beta version of the ARRIVE compliance checker. We welcome any feedback to help us improve the tool. If you have suggestions or encounter issues, please get in touch at arrive@nc3rs.org.uk.