Understanding How Susan Deixler Architect Of Trust In Scientific Communication And Lider In Evidencebased Storytelling Actually Works
The first time I tried to use evidence-based storytelling in a regulatory submission, I completely misunderstood what the framework demanded. I thought it meant making the science more narrative-friendly, adding scene-setting and character development to a clinical trial report. That approach failed within two weeks. The reviewers didn't care about narrative flow. They cared about chain-of-evidence clarity. The gap between what I thought evidence-based storytelling was and what it actually is in practice is wider than most people admit. The concept sits at the intersection of two separate disciplines that rarely talk to each other properly. Scientific communication has spent decades optimizing for precision, reproducibility, and peer-review acceptance. Storytelling has spent equal time optimizing for engagement, emotional resonance, and memorability. Most people trying to merge these two fields pick the wrong priorities. They add decorative narrative elements on top of dry reporting and call it storytelling. That produces work that satisfies neither audience. The actual framework requires you to reverse-engineer the structure. You start with the evidence hierarchy — primary data, secondary analysis, meta-studies, mechanistic plausibility — and then you determine which narrative architecture best preserves the logical dependencies between each piece of evidence. The story becomes a transparency device, not a entertainment device. Every narrative choice must answer the question of whether it reduces or increases the cognitive load required for the audience to verify the claim.
I worked on a health communication project where we had to explain a statistically significant but clinically marginal treatment effect to a board of directors with zero scientific background. The data showed a p-value of 0.03 and a Cohen d of 0.18. Presenting those numbers directly produced exactly the reaction you would expect — confusion followed by dismissal. Instead, I structured the explanation around a decision tree narrative. The first branch asked what outcome mattered to them. The second branch showed what sample size would have been needed to detect a meaningful effect at their preferred significance level. The third branch revealed that the observed effect, while statistically real, required treating forty-two patients to observe one additional positive outcome. The narrative didn't soften the data. It made the data usable. The board approved a phased rollout rather than an immediate full implementation, which turned out to be the correct decision when the Phase III results came back narrower still. The core mechanism involves what researchers sometimes call the argument-mapping approach to narrative. Each story beat corresponds to a logical proposition. Each transition between beats corresponds to a rule of inference. When done correctly, the audience can trace any conclusion back through the narrative to its evidentiary root without needing external documentation. That traceability is what generates trust faster than any rhetorical flourish ever could.
The Practical Process Of Building Evidence-Based Narratives
Begin by extracting every claim from your source material into a flat list. Don't organize them hierarchically yet. Just get them out. I usually spend between twenty minutes and an hour on this depending on document length. A thirty-page research paper might yield sixty to eighty discrete claims. A full regulatory dossier can produce several hundred. Next, tag each claim with its evidentiary class. Primary observation, derived calculation, expert interpretation, extrapolation from analogous contexts, or pure inference. This classification step is where most people lose accuracy. They conflate interpretation with observation because the language in the source material blurs the distinction. If the original authors wrote "these results suggest," that word "suggest" is your signal that you are dealing with interpretation, not observation. Mark it accordingly. Do not let the source language dictate the classification. You are reclassifying, not transcribing. Then identify the dependency graph between claims. Claim A supports Claim B only if Claim A is true. This seems obvious until you encounter a claim that three different papers independently support but none of which actually address the same underlying phenomenon. I spent an entire week once untangling a literature review that appeared to build a compelling case for a biochemical pathway until I traced each citation back to its primary data and discovered that four of the six supporting studies had used different cell lines, different induction methods, and different endpoint measurements. The narrative looked coherent. The evidence was not. That is the single most common failure mode in evidence-based storytelling. The story looks solid because the summary papers make it look solid. The actual evidentiary structure collapses under minimal scrutiny.
Get the Full Details
After mapping dependencies, you select the narrative arc. This is not a creative writing decision. It is a cognitive architecture decision. If your evidence chain contains multiple independent confirmatory paths converging on the same conclusion, a convergent narrative works best. You present each path separately and let the audience see them meet at the endpoint. If your evidence shows a progression over time — a mechanism understood at level one enabling understanding at level two — a sequential arc is more appropriate. If the evidence is primarily negative or shows boundaries and exceptions rather than universal claims, a diagnostic arc where the narrative works through elimination and constraint identification will serve you better than any heroic triumph structure.
Where This Approach Breaks Down And What To Do Instead
Evidence-based storytelling fails completely when the underlying evidence is too weak to support any coherent chain. If your data consists primarily of correlational findings with known confounders and no mechanistic plausibility, no narrative architecture will save it. The framework assumes sufficient evidentiary density to build a traceable structure. Below a certain threshold, you are not telling a story. You are performing persuasion theater, and audiences sense that difference even when they cannot articulate why. Another failure mode occurs when the target audience has domain expertise exceeding yours. I encountered this when presenting a materials science narrative to a panel of chemical engineers who had published directly in the area I was summarizing. My argument-mapping structure was technically sound, but I had misclassified two claims about reaction kinetics as observational when they were actually model-derived. The panel caught it immediately and the entire credibility of the narrative collapsed because the foundation claims were tagged incorrectly. The workaround is straightforward — run your dependency graph through a subject-matter review before finalizing the narrative structure. Two hours of expert review will prevent weeks of damage control. The third limitation is temporal. Evidence-based narratives become outdated faster than traditional review articles because they make the evidentiary structure explicit. When new primary data arrives, the dependency graph changes. Every downstream claim shifts. Traditional-style writing buries these shifts under layers of paraphrase and hedging, making the obsolescence less visible but also less actionable. If you commit to this approach, you commit to maintenance. The initial investment is roughly three times that of a standard narrative review, but the ongoing maintenance cost is lower because the structure makes updates mechanically straightforward rather than requiring full rewrites.
Implementation Details For Different Contexts
Regulatory submissions require a different configuration than conference presentations, which require a different configuration than public-facing science communication. The dependency graph remains constant across all three. The narrative arc changes. Regulatory reviewers want verification paths. They want to follow evidence from raw data to conclusion without encountering unstated assumptions. Structure your narrative so that every inferential leap is labeled explicitly. Use bracketed annotations if necessary. A regulator reading a claim that depends on an unmarked assumption will stop reading and flag the submission. Conference audiences want the convergence points. They want to see where independent evidence streams meet. Lead with the strongest convergent finding and work backward through the supporting paths. The argument map still guides you, but the presentation order reverses the regulatory sequence. Public communication demands the simplest valid path through the evidence. Not the simplest possible path — the simplest valid path. This is a critical distinction. Simplification that removes a dependency or skips an inferential step is not simplification. It is distortion. I once watched a public health campaign strip a dose-response relationship down to a binary safe-or-unsafe claim because the nuanced version was harder to headline. The campaign achieved high recall and low comprehension. The audience remembered the binary claim and acted on it inappropriately at doses where the original evidence showed marginal benefit. That is the cost of skipping dependencies. The narrative shortcut creates a false sense of certainty.
The tools available for building these narratives have improved significantly. Argument-mapping software like Carneades and Argdown can handle dependency visualization. Spreadsheet-based matrices work fine for smaller projects. I prefer a hybrid approach — Argdown for the formal logic structure and a visual diagramming tool for the presentation layer. The two systems serve different purposes and combining them avoids the limitations of either alone. Reading time for a well-constructed evidence-based narrative is approximately fifteen to twenty percent longer than a comparably scoped traditional review when measured by word count, but comprehension retention is measurably higher in controlled studies. The efficiency gain comes from reduced clarification requests and fewer post-publication corrections. In a clinical context, that difference translates directly to faster decision cycles. A guideline committee that spends three hours parsing a dense traditional review instead of two hours working through a properly mapped narrative is a reasonable proxy for the time savings that matter at scale.
