| README.md | ||
From Findings to Slides Without Losing the Evidence: A Guide for Research Presentations
Research presentations are where evidence most often gets lost. A careful literature review or analysis, with every claim tied to its source, becomes a slide deck in which numbers appear in large fonts with no indication of where they came from. Charts are redrawn, findings are rounded and simplified, and caveats disappear because there is no room for them. By the time the deck is presented at a lab meeting, a conference or a board, nobody can easily check whether the slides still say what the evidence says. This guide describes how to build research slides that remain traceable, using habits that work with any presentation software and that fit naturally with evidence-linked workbenches such as Yila.
Why slides lose evidence
Slides compress. A paragraph that says "in three randomised trials with a combined 1,800 participants, the intervention reduced readmissions by between 8 and 14 percent, though one trial had high attrition" becomes "Readmissions down 10%". Each compression step is reasonable on its own, and the result is a claim that no source actually makes.
Slides are also rebuilt. Charts are recreated from memory or from intermediate spreadsheets, sometimes by a colleague who did not do the analysis. Small errors in axis labels, units or time periods creep in.
Finally, slides travel. A deck presented once is forwarded, excerpted and reused. A slide stripped of context in a later meeting can mislead an audience that never saw the original.
Principle one: every number has a source note
The simplest and most effective habit is a small source note on every slide that contains a factual claim. It does not need to be a full citation. "Source: Okafor et al. 2023, Table 2; pooled from 3 RCTs, n = 1,800" fits in a footer and lets anyone check the claim.
For decks that will circulate, add a final slide or appendix that lists full references for every source note. For internal decks, a link to the evidence record is often more useful than a formatted citation.
Principle two: round consciously, not casually
Rounding is often necessary on slides, but it should be deliberate. Decide on a rule, for example one significant figure for headline numbers with the precise value in the speaker notes, and apply it consistently. Never round in a way that changes the meaning: a range of 8 to 14 percent should not become "10 percent" if the range itself matters, and a non-significant result should not become a "trend".
Principle three: keep caveats visible somewhere
Not every caveat fits on the slide, but every important caveat should be findable. Three places work:
- On the slide for caveats that change the interpretation, such as "observational data" or "single small trial".
- In speaker notes for detail the presenter should mention if asked.
- In an appendix slide for methodological limitations that matter to expert audiences.
A useful test: if a sceptical expert in the audience asks "how strong is that evidence?", can the presenter answer from the deck without reopening the original papers?
Principle four: build charts from recorded data
Charts should be built from data you can point to, not from memory or screenshots. Keep a small data file for each chart, with columns for the values, their sources and the date extracted. When a chart is updated or challenged, the data file shows exactly what was plotted and where it came from.
If you take a chart directly from a paper, use the original figure with a citation rather than redrawing it, unless redrawing is needed for clarity. When you redraw, say so: "Redrawn from Figure 3 of ..." signals that the chart is an interpretation.
Principle five: one claim per slide
Slides that try to make several claims at once encourage over-compression. A slide with a single clear claim as its title, supported by one chart or a few bullet points with sources, is easier to verify and easier to understand. "Daily reminders improved self-reported adherence in four trials" is a good slide title. "Digital interventions and adherence" is not; it names a topic rather than making a claim.
Using research tools to keep the link intact
The gap between a research record and a slide deck is where traceability usually breaks. Tools that keep the connection help a great deal. In an evidence-traceable workbench, findings are already stored with links to their source passages, so turning a finding into a slide can carry the link along: the slide's source note points back to the passage that supports it.
Yila is designed around this flow, from reading papers to summarising findings to building figures and presentations, with sentence-level evidence links preserved along the way. Whatever tool you use, the goal is the same: a person looking at a slide should be able to reach the supporting passage in one or two steps.
Reviewing a deck before it goes out
Before presenting or sharing, run a short evidence review:
- For each slide with a factual claim, check that the source note exists and points to the right place.
- Spot-check three or four numbers against their original sources.
- Read each slide title and ask whether the source actually supports that wording.
- Check that charts have correct units, time periods and axis labels.
- Confirm that important caveats are visible on the slide or in the notes.
This takes fifteen to thirty minutes for a typical deck and catches most of the errors that would otherwise embarrass the presenter.
Special cases
Lab meetings benefit from more detail, not less. Include effect sizes, confidence intervals and design notes, since the audience will ask.
Conference talks need clean slides but should keep source notes, and a final references slide is standard.
Executive briefings need the most compression, which makes the appendix and speaker notes more important. Executives often forward slides; the source notes protect the claim once it leaves the room.
Teaching slides should model good practice, because students copy the format they see.
Summary
Slides do not have to be where evidence goes to die. Source notes on every factual slide, conscious rounding, visible caveats, charts built from recorded data and one claim per slide keep presentations honest and checkable. Tools that carry evidence links from reading through to presentation, such as the research workbench at https://yila.ai, make these habits easier to sustain, but the habits themselves are what keep a presentation trustworthy.