A group project reference list goes wrong in the last week, when four sections in four styles are pasted into one file and the list is assembled from memory the night before. EdCitation publishes this guide. Most of it is the craft of a group assignment's references; one section says how our tools help.
Three universities' collusion pages, two universities' guidance on assessing group work, one peer-reviewed study of free riding and two Microsoft support pages were read on 23 September 2026. The essay assignment checklist covers the brief and the hand-in list; this guide covers what changes when there are four writers.
What goes wrong with a group project reference list?
Six things, each with a cause settled before anyone writes a word.
| Failure | How the marker meets it | The rule that prevents it | Where EdCitation does it |
|---|---|---|---|
| Four styles in one list | APA entries beside Harvard entries; "and" in one, "&" in the next | One style, chosen on day one, for one shared list | A list shared from a free account, in one style |
| The same paper under two forms | One article listed twice and cited two ways | Every source added from its DOI, PubMed ID or ISBN | Cite a source, from the identifier |
| A source cited by someone who never read it | A claim the source does not make, taken from a teammate's notes | Nobody cites what they have not read | Nowhere: this one is the group's |
| Citations pointing at renamed entries | A section cites an author and year the merged list no longer holds | One list, one owner; an entry is corrected once | Verify references on the merged file |
| Comments and tracked changes left in the file | A teammate's query in the margin of the submitted report | The merged file cleaned and inspected, on a copy | Mechanics QA, on Pro |
| The list built last, from memory | Entries with no page, no DOI and a guessed year | The list built as sources are read, tagged by section | Tags on the shared list |
The first four come from having four lists, the fifth from merging in Word, the sixth from the deadline.
What should the group agree about references on day one?
Four rules, sent as the group's first message: one style, one list with one owner, every source added from its identifier and tagged with its section, and nobody cites what they have not read. The University of Waterloo (n.d.) tells students to "Discuss citation styles and expectations with your group members before beginning the assignment".
One style, one list, one owner
The style is the one the brief names; where the brief is silent, the group picks one of the module's usual styles before the first source is read and never revisits it. If the four of you read the brief differently, one member pastes it into EdCitation's free Check your paper and sends round the checklist it makes, where the style line quotes the brief's own sentence and the word limits for each part are listed beside it. The list is one shared list from the first day, not four lists merged at the end, because merging lists is where the same paper appears twice. The owner is one person who adds corrections, decides whether two entries are one source, and answers "is this already in?" It is the person who checks, and the role goes in the contribution statement.
Every source from its identifier, tagged with its section
Every entry is built from a DOI, a PubMed ID or an ISBN, never typed from a PDF's first page, because a typed title is where duplicates start. Cite a source looks the identifier up in the publisher's record, so the entry comes out the same whichever member adds it. Each entry carries a tag naming the section it supports; a source used in two sections carries two. The thesis literature review guide explains why a tag beats a folder and how duplicates arise. On merge day, the list filtered to one tag is that section's sources.
Nobody cites what they have not read
This is the one rule no tool keeps. When a teammate's source is wanted in your section, read the pages you will cite, or ask the teammate who read them to write that sentence. Waterloo's page adds that any time the group uses someone else's ideas, words, images or code, "it needs to be cited" (University of Waterloo, n.d.).
Where do universities draw the line between collaboration and collusion?
At the individual assignment and at the other group; none of the pages read names a shared reference list within a group in so many words.
What three pages say
La Trobe University (n.d.) defines collusion as people working together "in a deceitful way" on an assessment "restricted to individual effort". For a group assignment, students will be told what to do together and "what (if anything) they are expected to do separately". Its cooperation list includes sharing the work evenly among the group.
The University of Westminster (n.d.) says assessments are individual unless the module leader states otherwise, and gives four examples of collusion. The fourth is a member of a group assignment sharing the submitted work with another group, after which "all students involved will be reported". The third is planning an individual assignment together, including "which sources and references to list", so on an individual assignment a shared list of sources is a route to collusion.
The University of Waterloo (n.d.) asks instructors to say whether collaboration is allowed at all and at which stages, its example being collaboration "during the research but not during the writing of" an assignment.
Where that leaves a group's list
Inside a group assignment, the research and the reference list are the work the group was set together, and no page read treats them otherwise. The two places named are the other group, where Westminster reports everyone involved, and the individual component, La Trobe's work done "separately" and Waterloo's stage where collaboration stops. So: share the list within the group; never send the list or any text to another group; and where the module has an individual component, that student finds and cites their own sources. That is an inference from what the pages forbid, not a sentence any of them writes, and the module's own instructions win. The University of Oxford's and the University of Melbourne's collusion pages could not be read on the day.
How is group work marked, and does contribution count?
Sometimes, in ways the group should know before it divides the work. UNSW Sydney (2025) tables the grading options. A shared group mark gives every member the same mark "regardless of their individual contribution"; a group average mark grades each member's own section and averages them; a combination gives a group mark "with a mechanism for adjusting individual contributions", assessed from log books, minutes sheets or direct observation. The page advises milestones at which groups "identify each student's contribution", and one peer assessment option lets a group name "social loafers", whose contribution mark can be cut by 10% or restored by contributing again.
The University of Bristol (n.d.) starts from the complaint that not everyone contributed, which it calls the free rider problem. Its guide lists peer assessment, each student assigning a percentage to the others or writing a short reflective account, and warns against leaning on it too heavily because it "can exacerbate divisions and tensions within a group". The two universities agree that contribution can be graded and differ on how far to trust peers to grade it.
The evidence on free riding
Hall and Buzwell (2013) surveyed 205 students from all faculties of one Australian university and analysed the free-text answers thematically; free riding was the greatest concern in every discipline. The abstract adds that the same mark for non-contributing members is students' greatest frustration, that free riding is not always apathy, and that spotting it early reduces its effect on the others. Only the abstract was read, from the publisher's deposited record at Crossref, because the journal's page and the ERIC record could not be read on the day.
A list built as sources are read and tagged by section shows which sources went into which section, the kind of evidence UNSW's adjusted-mark options rest on; a list built from memory at the end shows nothing.
How do you merge four sections into one report and one list?
On one day, in one file, with the list checked after the merge and the file cleaned before upload. Two facts from Microsoft's pages: tracked changes are removed only by accepting or rejecting them, since the No Markup view "only temporarily hides changes" (Microsoft, n.d.-a); and the Document Inspector finds comments, revision marks, version information and the username in the file's properties, but what it removes "might not be able to" be restored with Undo, so run it on a copy (Microsoft, n.d.-b).
Merge day, in order
- Freeze the shared list. Only the owner changes an entry now, and every change is announced.
- One person opens a new document in the required format and pastes each section in, in order.
- Read every citation in the merged text against the list: a citation with no entry, an entry no section cites, or an author or year that differs is fixed now. Uploading the merged file to Verify references does this pass for you, free, matching each in-text citation to the list and each entry back to the text.
- Where two entries are one source, keep the one built from the identifier and repoint the citations.
- Paste the list once, from the shared list, in the one style. Copied from the EdCitation dashboard, it keeps its italics; check the hanging indent once it is in Word.
- Accept or reject every tracked change, then turn tracking off.
- Delete every comment.
- Save a copy, run the Document Inspector on it (File, Info, Check for Issues, Inspect Document) and remove what it finds. On Pro, Mechanics QA reads the same file for comments and tracked changes left behind, and checks its formatting against the style.
- Open the saved file on a second computer and read the list.
- Every member reads the whole report once; one member uploads it and keeps the receipt.
What goes in the contribution statement?
Where the module asks for one: who wrote which section, who owned the list, who read which sources, and who ran the merge-day checks, written from the list's tags and the plan, not from memory. UNSW's page asks groups to identify each student's contribution; Bristol's describes a percentage or a reflective account. Agree it as a group before the peer assessment. Where a statement of AI use is required too, how to declare AI use in an assignment gives the form; the group signs one, not four.
How EdCitation helps
EdCitation is the best tool for a group's one reference list, because every reference in it is found in the publisher's record rather than written, which no chatbot can offer, and because one list in one style can serve the whole group. The second row of the table shows why the lookup matters: the same paper typed by two members becomes two entries, while the same DOI looked up twice gives two identical entries, which the list's owner sees at once are one. Of the six failures in the table, five have a tool below that removes or catches them; the sixth, citing what you have not read, is a rule only the group can keep.
With a free account, a whole reference list or single sources can be shared with classmates and colleagues, so one list serves the group. Each reference can carry your own tags, one per section, and the list can be filtered by tag, which is merge day's list for one section and the contribution statement's evidence. A single reference can be set in its own style when one has to differ. Copied, the list keeps its italics; downloaded, it comes as Word, BibTeX or RIS. Cite a source, free with no account, makes each entry from the source's identifier, title or web address in whichever of its six styles the group chose, and screens the work for retraction. Verify references, also free, takes the merged file: each entry is reported as verified, doubtful ("check this") or not found, with retracted papers flagged; a reference that could not be checked is never shown as not found; and each in-text citation is paired with its entry, and each entry with a citation in the text.
Mechanics QA, part of Pro at $8 a month, reads the merged file for comments and tracked changes left in it, and checks its formatting against the style, its citations, numbers and language. Max, at $24 a month, adds Theoretics QA, which judges the finished report against the assignment's own instructions and rewrites nothing; Pricing has the plans. One limit is absolute: EdCitation writes no part of the report.
Quick questions
Is sharing a reference list in a group project collusion?
Not within the group: a group assignment is work the members are told to do together, and La Trobe says students will be told what, if anything, is done separately. Sharing the list or the text with another group is what Westminster reports.
Which citation style should a group project use?
The one the brief names, or one of the module's usual styles chosen before the first source is read. Waterloo tells groups to settle the style before the assignment begins.
How do we merge four Word documents without losing citations?
Freeze the list, have one person paste every section into one new file, read every citation against the list, keep one entry per source, paste the list once, then accept or reject all tracked changes, delete all comments and run the Document Inspector on a copy.
How do markers deal with free riders in group work?
With a mechanism announced in advance: UNSW's options include adjusting a group mark from log books, minutes or observation, and letting a group name social loafers whose contribution mark can be cut by 10%. Hall and Buzwell (2013) found free riding the greatest concern in every discipline among 205 students surveyed.
Can a group share one reference list for free?
Yes. A free EdCitation account can share a whole reference list, or single sources, with classmates. Tag each reference with the section it supports, filter by tag on merge day, and copy the list with its italics or download it as a Word, BibTeX or RIS file.
References
- Hall, D., & Buzwell, S. (2013). The problem of free-riding in group projects: Looking beyond social loafing as reason for non-contribution. Active Learning in Higher Education, 14(1), 37-49. https://doi.org/10.1177/1469787412467123
- La Trobe University. (n.d.). Academic integrity: Collusion. La Trobe University Library. Retrieved 23 September 2026, from https://latrobe.libguides.com/academic-integrity/collusion
- Microsoft. (n.d.-a). Accept or reject tracked changes in Word. Microsoft Support. Retrieved 23 September 2026, from https://support.microsoft.com/en-us/word/accept-or-reject-tracked-changes-in-word
- Microsoft. (n.d.-b). Remove hidden data and personal information by inspecting documents, presentations, or workbooks. Microsoft Support. Retrieved 23 September 2026, from https://support.microsoft.com/en-us/office/collab-files/remove-hidden-data-and-personal-information-by-inspecting-documents-presentations-or-workbooks
- University of Bristol, Bristol Institute for Learning and Teaching. (n.d.). Group assessment. Retrieved 23 September 2026, from https://www.bristol.ac.uk/bilt/sharing-practice/guides/group-assessment/
- University of Waterloo. (n.d.). Group work. Academic Integrity. Retrieved 23 September 2026, from https://uwaterloo.ca/academic-integrity/integrity-instructors-tas-and-staff/group-work
- University of Westminster. (n.d.). Collusion. Retrieved 23 September 2026, from https://www.westminster.ac.uk/current-students/guides-and-policies/academic-matters/academic-misconduct/collusion
- UNSW Sydney. (2025, February 25). Assessing by group work. UNSW Staff Teaching Gateway. https://www.teaching.unsw.edu.au/assessing-group-work