If your PRISMA flow diagram numbers don’t add up, report the discrepancy transparently, explain the source of the mismatch, and adjust the diagram so the totals are mathematically consistent.
Why do my PRISMA flow diagram numbers not add up?
Most grading rubrics for scholarship essays that require a PRISMA diagram expect three exact totals: records identified, records screened, and studies included. The rubric will often circle the words “total,” “excluded,” and “included” and check that each column adds up correctly. When the numbers don’t line up, the grader sees a methodological red flag before even reading your argument.
Consider a sophomore biology major who pulled 112 articles from PubMed, removed 27 duplicates, screened 85 abstracts, excluded 58 after abstract review, and finally included 27 full‑text studies. If the diagram shows 112 → 85 → 27 but the “records excluded” cell reads 60, the arithmetic is off and the flow looks broken.
Two terminology traps cause this kind of error. First, the PRISMA checklist distinguishes “records identified” (database hits) from “records screened” (after duplicates removed). Mixing those counts inflates or deflates the subsequent columns. Second, the term “excluded” refers to the *cumulative* number removed at each stage, not the *stage‑specific* count. If you list 58 excluded at the abstract stage and then again list 58 at the full‑text stage, the total excluded will be double‑counted.
Grading guides usually ask you to label each box exactly as the PRISMA 2020 template does: “Identification,” “Screening,” “Eligibility,” and “Included.” Any deviation—such as renaming “Eligibility” to “Full‑text review” without a footnote—confuses the reviewer and can be marked down.
“I thought I could just copy the numbers from my spreadsheet, but the professor called out that the ‘excluded’ total didn’t match the sum of the rows.” – a returning student
Bottom line: double‑check that you’re using the precise PRISMA terms and that each total reflects the cumulative count, not a repeat of earlier exclusions.
What technical slip‑ups usually create mismatched counts?
Even when you understand the terminology, spreadsheet quirks and copy‑paste errors can wreck the diagram. One common failure mode is hidden rows in Excel. If you filter out rows after removing duplicates but forget to delete the hidden ones, the “records identified” figure still includes them, leading to a mismatch later.
Another pitfall is rounding or formatting. Some students export the count from a reference manager that shows “112.0” and then manually type “112” into the diagram. If the manager actually counted 113 items (perhaps an extra conference abstract), the discrepancy is hidden until the reviewer tallies the numbers.
Software settings in citation tools can also cause trouble. For example, EndNote’s “Find Duplicates” function may flag only exact matches, leaving near‑duplicates in the list. If you rely on that output for the duplicate‑removal total, the “records after duplicates removed” column will be too high.
Finally, the PRISMA flow diagram itself is often drawn in PowerPoint or a diagramming app. When you move a text box, the number inside may not update automatically. If you edited the underlying data but forgot to edit the diagram, the visual will no longer reflect the spreadsheet.
To avoid these technical snags, always generate the final numbers directly from the same source you’ll cite in the methodology paragraph. Then copy‑paste the values as plain text (Ctrl + Shift + V) into the diagram to prevent hidden formatting.
How can I report the discrepancy without raising integrity concerns?
Transparency is the safest route. The scholarship portal’s plagiarism detector looks for inconsistencies between the narrative and the diagram, not for honest corrections. If you explain the mismatch in the methods section, you demonstrate scholarly honesty.
Write a brief note right after the PRISMA figure: “During the final tally, it became apparent that the ‘records excluded’ total was inflated due to a duplicate‑removal miscount. The corrected total is 85, as shown in the revised diagram.” This sentence satisfies two rubric criteria: (1) acknowledgment of an error, and (2) presentation of the corrected figure.
Do not try to hide the issue by adjusting only the diagram without mentioning the change. Graders often cross‑check the numbers against the text; a silent alteration can be flagged as data manipulation.
When you submit the revised essay, attach a short “Errata” note as a separate PDF or as an appendix. The note should list the original numbers, the corrected numbers, and the reason for the change (e.g., “duplicate records were double‑counted in the initial extraction”). This practice mirrors how journal authors submit corrections and reassures reviewers that you are following ethical standards.
Key integrity tip: keep a dated log of the spreadsheet versions you used. If the portal’s AI scans for sudden number changes, the log shows a documented timeline, reducing the chance of a false flag.
Can I fix everything in 48 hours before the portal locks?
Yes, but you need a focused sprint. Below is a decision matrix you can follow tonight. Each row represents a concrete action, an estimated time block, and the exact deliverable you’ll produce.
| Action | Estimated minutes | Deliverable |
|---|---|---|
| Re‑run duplicate detection in your reference manager and export the final count. | 15 | Updated “records after duplicates removed” number. |
| Filter the master spreadsheet for each PRISMA stage, copy the row counts, and write them in a plain‑text list. | 20 | Three verified totals (identified, screened, included). |
| Open the diagram file, replace the old numbers with the new totals, and save as a PDF. | 10 | Corrected PRISMA flow diagram. |
| Insert a one‑sentence clarification after the figure in the essay body. | 5 | Transparency statement. |
| Draft a one‑page errata appendix that logs original vs. corrected numbers. | 15 | Appendix PDF ready for upload. |
Start with the duplicate check; it’s the most likely source of inflation. If the new count matches the “records identified” minus duplicates, you can trust the downstream numbers. Next, verify each screening stage by applying filters (e.g., “abstract excluded = yes”) and counting the rows. Use the COUNTIF function rather than manual tallies to avoid human error.
When you replace the numbers in the diagram, do it in the source file (PowerPoint, Lucidchart, etc.) and export a fresh PDF. Avoid editing the PDF directly because that can introduce invisible characters that confuse the portal’s file parser.
Finally, write the transparency sentence in the same paragraph that describes your search strategy. Keep it concise: “The initial exclusion count was 62; after correcting a duplicate‑record error, the final excluded total is 58.” Then attach the errata appendix as a supplemental file if the portal allows multiple uploads; otherwise, embed it as an appendix at the end of your essay.
With this matrix, you can realistically finish the corrections well within the 48 hours deadline.
What final‑submission checklist should I run through?
Before you hit “Submit,” run this quick checklist. Each item addresses a different rubric point, so ticking them off ensures you’ve covered the major concerns.
- Confirm that the PRISMA diagram totals add up: identified = screened + duplicates removed; screened = excluded + eligible; eligible = included + excluded at full‑text stage.
- Verify that every box label matches the PRISMA 2020 terminology (Identification, Screening, Eligibility, Included).
- Check that the numbers in the diagram match the plain‑text list you generated from the spreadsheet.
- Read the transparency sentence aloud; it should clearly state the original error and the corrected figure.
- Open the errata appendix and confirm it lists: original total, corrected total, reason for change, and date of revision.
- Save the entire essay as a single PDF (including the diagram and appendix) to avoid mismatched file uploads.
- Run the PDF through a basic text‑extractor (e.g., copy‑paste into a plain‑text editor) to ensure no hidden characters remain in the numbers.
Actionable takeaway: run the checklist once after you think you’re done, then again after a short break to catch anything you missed.