Wednesday, October 15, 2025

How To not Mislead with Your Knowledge-Pushed Story


is all over the place. There are numerous books, articles, tutorials, and movies, a few of which I’ve written or created.

In my expertise, most of those assets are inclined to current information storytelling in an overwhelmingly constructive mild. However currently, one concern has been on my thoughts:

What if our tales, as an alternative of clarifying, mislead?

Picture 1. Change the attitude, and also you see a wholly completely different story. Pictures by the creator

The picture above exhibits one of many condominium buildings in my neighborhood. Now, check out the photograph on the left and picture one of many flats within the white constructing is up on the market. You might be contemplating shopping for it. You’d doubtless concentrate on the quick environment, particularly as introduced within the vendor’s photographs. Discover something uncommon? In all probability not, no less than not instantly.

Ought to the quick setting be a dealbreaker? In my view, not essentially. It’s not essentially the most picturesque or charming spot—only a typical block in a mean neighborhood in Warsaw. Or is it?

Let’s take a brief stroll round to the again of the constructing. And… shock: there’s a public bathroom proper there. Nonetheless be ok with the placement? Possibly sure, perhaps no. One factor is evident: you’ll wish to know {that a} public rest room sits slightly below your future balcony.

Moreover, the condominium is situated within the decrease a part of the constructing, whereas the remainder of the towers rise above it. That is one other issue that could be vital. Each these “points” for certain will be introduced up in worth negotiations.

This straightforward instance illustrates how simply tales (on this case, utilizing photographs) will be misinterpreted. From one angle, all the things seems advantageous, even inviting. Take a number of steps to the proper, and… whoops.

The identical scenario can occur in our “skilled” lives. What if audiences, satisfied they’re making knowledgeable, data-backed choices, are being subtly steered within the improper path—not by false information, however by the best way it’s introduced?

This put up builds on an article I wrote in 2024 about deceptive visualizations [1]. Right here, I wish to take a bit broader perspective, exploring how the construction and move of a narrative itself can unintentionally (or intentionally) lead folks to incorrect conclusions, and the way we will keep away from that.

Knowledge storytelling is subjective

We frequently wish to imagine that “information speaks for itself.” However in actuality, it hardly ever does. Each chart, dashboard, or headline constructed round a dataset is formed by human decisions:

  • what to incorporate,
  • what to go away out,
  • methods to body the message?

This highlights a core problem of data-driven storytelling: it’s inherently subjective. That subjectivity comes from the discretion now we have in proving the purpose we wish to make:

  • selecting which information to current,
  • choosing applicable evaluation approach,
  • deciding on arguments to emphasize,
  • and even what to to make use of.

Subjectivity additionally lies in interpretation — each ours and our viewers’s — and of their willingness to behave on the knowledge. This opens the door to biases. If we aren’t cautious, we will simply cross the road from subjectivity into unethical storytelling.

This text examines the hidden biases embedded in information storytelling and the way we will transition from manipulation to significant insights.

We’d like tales

Subjective or not, we’d like tales. Tales are important to us as a result of they assist make sense of the world. They carry our values, protect our historical past, and spark our creativeness. By way of tales, we join with others, be taught from previous experiences, and discover what it means to be human. Regardless of your nationality, tradition, or faith, now we have all heard numerous tales which have formed us. Informed us by our grandparents, mother and father, academics, pals, and colleagues at work. Tales evoke emotion, encourage motion, and form our id, each individually and collectively. In each tradition and throughout all ages, storytelling has been a robust technique of understanding life, sharing data, and constructing group.

However whereas tales can enlighten, they will additionally mislead. A compelling narrative has the facility to form notion, even when it distorts details or oversimplifies advanced points. Tales typically depend on emotion, selective element, and a transparent message, which may make them persuasive, but additionally dangerously reductive. When used carelessly or manipulatively, storytelling can reinforce biases, obscure reality, or drive choices based mostly extra on feeling than motive.

Within the subsequent a part of this text, I’ll discover the potential issues with tales — particularly in data-driven contexts — and the way their energy can unintentionally (or deliberately) misguide our understanding.

Picture 2. Tales have at all times been a necessary a part of our lives. Picture generated by the creator in ChatGPT.

Narrative biases in data-driven storytelling

Bias 1. Knowledge is much, far-off from interpretation

Right here’s an instance of a visible from a report titled “Kentucky Juvenile Justice Reform Analysis: Assessing the Results of SB 200 on Youth Dispositional Outcomes and Racial and Ethnic Disparities.”

Picture 3. Picture from “Kentucky Juvenile Justice Reform Analysis…”, web page 18 of the report.

The graph exhibits that younger offenders in Kentucky are much less prone to reoffend if, after their first offense, they’re routed by a diversion program. This program connects them with group assist, comparable to social employees and therapists, to deal with deeper life challenges. That’s a robust narrative with real-world implications: it helps lowering our reliance on an costly legal justice system, justifies elevated funding for non-profits, and factors towards significant methods to enhance lives.

However right here’s the issue: until you have already got sturdy information literacy and topic data, these conclusions aren’t instantly apparent from the graph. Whereas the report does make this level, it doesn’t accomplish that till almost 20 pages later. It is a traditional instance of how the construction of educational reporting can mute the story’s affect. It outcomes from the truth that information is introduced visually in a single part and interpreted textually in several (and typically distant) sections of the doc.

Bias 2. The Story of the Lacking Map: Choice Bias

Picture 4. Photograph Ashleigh Shea, Unsplash

Selecting which information factors (cherries 😊) to incorporate (and which to disregard) is among the strongest — and infrequently most ignored — acts of bias. And maybe no business illustrated this higher than Huge Tobacco.

The now-famous abstract of their authorized technique says all of it:

Sure, smoking causes lung most cancers, however not in individuals who sue us.

That quote completely captures the tone of tobacco litigation within the late twentieth century, the place firms confronted a wave of lawsuits from prospects affected by illnesses linked to smoking. Regardless of overwhelming medical and scientific consensus, tobacco companies routinely deflected accountability utilizing a collection of arguments that, whereas typically legally strategic, have been scientifically absurd.

Listed here are 4 of essentially the most egregious cherry-picking ways they utilized in court docket, based mostly on this text [2].

Cherry-pick tactic 1: use “exception fallacy” tactic in authorized or rhetorical contexts.

Sure, smoking causes most cancers — however not this one.

  • The plaintiff had a uncommon type of most cancers, like bronchioloalveolar carcinoma (BAC) or mucoepidermoid carcinoma, which they claimed weren’t conclusively linked to smoking.
  • In a single case, they argued the most cancers was from the thymus, not the lungs, regardless of overwhelming medical proof.

Cherry-pick tactic 2: Spotlight obscure exceptions or uncommon most cancers sorts to problem normal epidemiological proof.

It wasn’t our model.

  • “Positive, tobacco could have brought on the illness — however not our cigarettes.”
  • In Ierardi v. Lorillard, the corporate argued that the plaintiff’s publicity to asbestos-laced cigarette filters (Micronite) occurred exterior the slim 4-year window after they have been used, regardless that 585 million packs have been bought throughout that point.

Cherry-pick tactic 3: Concentrate on model or product variation as a method to shift blame.

In a number of circumstances, comparable to Ierardi v. Lorillard and Lacy v. Lorillard, the protection admitted that cigarettes may cause most cancers however argued that the plaintiff:

  • Didn’t use their model on the time of publicity,
  • Or didn’t use the particular model of the product that was most harmful (e.g., Kent cigarettes with the asbestos-containing Micronite filter),
  • Or didn’t use the particular model of the product that was most harmful (e.g., Kent cigarettes with the asbestos-containing Micronite filter),
  • window years in the past, making it unlikely the plaintiff was uncovered.

This tactic shifts the narrative from

Our product brought on hurt.

to

Possibly smoking brought on hurt—however not ours.

Cherry-pick tactic 4: Emphasize each different potential threat issue — no matter plausibility — to deflect from tobacco’s position.

There have been different threat elements.

  • In lots of lawsuits, firms pointed to various causes of sickness: asbestos, diesel fumes, alcohol, genetics, weight-reduction plan, weight problems, and even spicy meals.
  • In Allgood v. RJ Reynolds, the protection blamed the plaintiff’s situation partly on his fondness for “Tex-Mex meals.”

Cherry-picking isn’t at all times apparent. It may cover in authorized defenses, advertising copy, dashboards, and even tutorial reviews. However when solely the information that serves the story will get informed, it stops being perception and begins changing into manipulation.

Bias 3: The Mirror within the Forest: How the Similar Knowledge Tells Totally different Tales

How we phrase outcomes can skew interpretation. Ought to we are saying “Unemployment drops to 4.9%” or “Thousands and thousands nonetheless jobless regardless of positive factors”? Each will be correct. The distinction lies in emotional framing.

In essence, framing is a strategic storytelling approach that may considerably affect how a narrative is acquired, understood, and remembered. By understanding the facility of framing, storytellers can craft narratives that resonate deeply with their viewers and obtain their desired objectives. I current some examples in Desk 1.

  Body A Body B Goal description
Unemployment “Unemployment hits 5-year low”
Suggests progress, restoration, and powerful management.
“Thousands and thousands nonetheless with out jobs regardless of slight drop” Highlights the persistent downside and unmet wants. A modest drop within the unemployment fee.
Vaccine Effectiveness “COVID vaccine reduces threat by 95%”
Emphasizes safety, encourages uptake.
“1 in 20 nonetheless will get contaminated even after the jab.”
Focuses on vulnerability and doubt.
A medical trial confirmed a 95% relative threat discount.
Local weather Knowledge “2023 was the most popular 12 months on file.”
Calls consideration to the worldwide disaster.
“Earth has at all times gone by pure cycles.”
Implies nothing uncommon is occurring.
Lengthy-term temperature information.
Firm Monetary Reviews “Income grows 10% in Q2.”
Celebrates short-term acquire.
“Nonetheless under pre-pandemic ranges”.
Indicators underperformance in the long term.
Quarterly earnings report.
Election Polls “Candidate A leads by 3 factors!”
Creates a way of momentum.
“Inside margin of error: race too near name.”
Emphasizes uncertainty.
A ballot with +/- 3% margin.
Well being Warnings “This drink has 25 grams of sugar.”
Sounds scientific, impartial.
“This drink incorporates over six teaspoons of sugar.”
Sounds extreme and harmful.
25 grams of sugar.
Desk 1. Alternative ways of framing the identical story. Examples generated by the creator utilizing ChatGPT.

Bias 4: “The Dragon of Design: How Magnificence Beguiles the Reality”

Visuals simplify information, however they will additionally manipulate notion. In my older article [1], I listed 14 misleading visualization ways. Here’s a abstract of them.

  1. Utilizing the improper chart sort: Selecting charts that confuse relatively than make clear — like 3D pie charts or inappropriate comparisons — makes it tougher to see the story the information tells.
  2. Including distracting parts: Stuffing visuals with logos, decorations, darkish gridlines, or litter hides the necessary insights behind noise and visible overload.
  3. Overusing colours: Utilizing too many colours can distract from the main target. With no clear coloration hierarchy, nothing stands out, and the viewer is overwhelmed.
  4. Random information ordering: Scrambling classes or time collection information obscures patterns and prevents clear comparisons.
  5. Manipulating axis scales: Truncating the y-axis exaggerates variations. Extending it minimizes significant variation. Each distort notion.
  6. Creating pattern illusions: Utilizing inconsistent time frames, selective information factors, or poorly spaced axes to make non-trends look vital.
  7. Cherry-picking information: Solely displaying the elements of the information that assist your level, ignoring the total story or contradicting proof.
  8. Omitting visible cues: Eradicating labels, legends, gridlines, or axis scales to make information laborious to interpret, or laborious to problem.
  9. Overloading charts: Packing an excessive amount of information into one chart will be distracting and complicated, particularly when essential information is buried in visible chaos.
  10. Exhibiting solely cumulative values: Utilizing cumulative plots to suggest easy progress whereas hiding volatility or declines in particular person intervals.
  11. Utilizing 3D results: 3D charts skew notion and make comparisons tougher, typically resulting in deceptive details about measurement or proportion.
  12. Making use of gradients and shading: Fancy textures or gradients shift focus and add visible weight to areas that may not deserve it.
  13. Deceptive or imprecise titles: A impartial or technical title can downplay the urgency of findings. A dramatic one can exaggerate a minor change.
  14. Utilizing junk charts: Visually overdesigned, advanced, or overly creative charts which might be laborious to interpret and straightforward to misinterpret.

Bias 5: “The Story-Spinning Machine: However Who Holds the Thread?”

Trendy instruments like Energy BI Copilot or Tableau Pulse are more and more producing summaries and “insights” in your behalf. To not point out crafting summaries, narratives, or complete shows ready by LLMs like ChatGPT or Gemini.

However right here’s the catch:
These instruments are educated on patterns, not ethics.

AI can’t inform when it’s making a deceptive story. In case your immediate or dataset is biased, the output will doubtless be biased as effectively, and at a machine scale.

This raises a essential query: Are we utilizing AI to democratize perception, or to mass-produce narrative spin?

Picture 5: Photograph by Aerps.com on Unsplash

A current BBC investigation discovered that main AI chatbots ceaselessly distort or misrepresent present occasions, even when utilizing BBC articles as their supply. Over half of the examined responses contained vital points, together with outdated details, fabricated or altered quotes, and confusion between opinion and reporting. Examples ranged from incorrectly stating that Rishi Sunak was nonetheless the UK prime minister to omitting key authorized context in high-profile legal circumstances. BBC executives warned that these inaccuracies threaten public belief in information and urged AI firms to collaborate with publishers to enhance transparency and accountability.[3]

Feeling overwhelmed? You’ve solely seen the start. Knowledge storytelling can fall prey to quite a few cognitive biases, every subtly distorting the narrative.

Take affirmation bias, the place the storyteller highlights solely information that helps their assumptions—proclaiming, “Our marketing campaign was successful!”—whereas ignoring contradictory proof. Then there’s end result bias, which credit success to sound technique: “We launched the product and it thrived, so our method was good,”—even when luck performed a serious position.

Survivorship bias focuses solely on the winners—startups that scaled or campaigns that went viral—whereas neglecting the numerous that failed utilizing the identical strategies. Narrative bias oversimplifies complexity, shaping messy realities into tidy conclusions, comparable to “Vaping is at all times safer,” with out ample context.

Anchoring bias causes folks to fixate on the primary quantity introduced—like a 20% forecast—distorting how subsequent data is interpreted. Omission bias arises when necessary information is ignored, for example, solely highlighting top-performing areas whereas ignoring underperforming ones.

Projection bias assumes that others interpret information the identical approach the analyst does: “This dashboard speaks for itself,”—but it might not, particularly for stakeholders unfamiliar with the context. Scale bias misleads with disproportionate framing—“A 300% enhance!” sounds spectacular till you be taught it went from only one to a few customers.

Lastly, causality bias attracts unfounded conclusions from correlations: “Customers stayed longer after we added popups—they have to love them!”—with out testing whether or not popups have been the precise trigger.

Easy methods to “Unbias” Knowledge Storytelling

Each information story is a selection. In a world the place consideration spans are quick and AI writes quicker than people, these decisions are extra highly effective — and harmful — than ever.

As information scientists, analysts, and storytellers, we should method narrative decisions with the identical degree of rigor and thoughtfulness that we apply to statistical fashions. Crafting a narrative from information isn’t just about readability or engagement—it’s about accountability. Each selection we make in framing, emphasis, and interpretation shapes how others understand the reality. And on the finish of the day, essentially the most harmful tales aren’t the false ones—they’re those that really feel like details.

On this a part of the article, I’ll share a number of sensible methods that can assist you strengthen your information storytelling. These concepts will concentrate on methods to be each compelling and credible—methods to craft narratives that have interaction your viewers with out oversimplifying or deceptive them. As a result of when finished effectively, information storytelling doesn’t simply talk perception—it builds belief.

Technique 1: The Smart Wizard’s Rule: Ask, Don’t Enchant

On this planet of information and evaluation, essentially the most insightful storytellers don’t announce their conclusions with dramatic aptitude—they lead with considerate questions. As an alternative of presenting daring declarations, they invite reflection by asking, “What do you see?” This method encourages others to find insights on their very own, fostering understanding relatively than passive acceptance.

Contemplate a graph displaying a decline in take a look at scores. A surface-level interpretation would possibly instantly declare, “Our faculties are failing,” sparking concern or blame. However a extra cautious, analytical response can be, “What elements may clarify this variation? May it’s a brand new testing format, adjustments in scholar demographics, or one thing else?” Equally, when gross sales rise following the launch of a brand new characteristic, it’s tempting to attribute the rise solely to the characteristic. But a extra rigorous method would ask, “What different variables modified throughout this era?”

By main with questions, we create area for interpretation, dialogue, and deeper considering. This methodology guards towards false certainty and encourages a extra collaborative, considerate exploration of information. A robust narrative ought to information the viewers, relatively than forcing them towards a predetermined conclusion.

Technique 2: The Mirror of Many Truths: Provide Counter-Narratives

Good information storytelling doesn’t cease at a single interpretation. Complicated datasets typically permit for a number of legitimate views, and it’s the storyteller’s accountability to acknowledge them. Presenting a counter-narrative—“right here’s one other approach to have a look at this”—invitations essential considering and builds credibility.

For instance, a chart could present that coronary heart illness charges are declining general. That looks like successful. However a more in-depth look could reveal that the advance is concentrated in higher-income areas, whereas charges in rural or underserved communities stay excessive. Presenting each views—progress and disparity—supplies a extra complete and sincere image of the difficulty.

By providing counter-narratives, we guard towards oversimplification and assist our viewers perceive the nuance behind the numbers.

Picture 6. Including the earnings class dimension permits for higher perception discovery. Chart generated in ChatGPT, pretend information.

Technique 3: The Curse of Crooked Charts: Keep away from Misleading Visuals

Visuals are highly effective, however that energy should be used responsibly. Deceptive charts can distort notion by refined tips, comparable to truncated axes that exaggerate variations, unlabeled items that obscure the dimensions, or ornamental litter that distracts from the message. To keep away from these pitfalls, at all times clearly label axes, begin scales from zero when applicable, and select chart sorts that greatest match the information, not simply their aesthetic enchantment. Deception doesn’t at all times come from malice—typically it’s simply careless design. However both approach, it erodes belief. A clear, sincere visible is much extra persuasive than a flashy one which hides the main points.

Picture 7. Two variations of the identical visible. One is telling the story, the opposite…?. Picture by the creator.

Take, for instance, the 2 charts proven in Picture 7. The one on the left is cluttered and laborious to interpret. Its title is imprecise, the extreme use of coloration is distracting, and pointless parts—like heavy borders, gridlines, and shading—solely add to the confusion. There are not any visible cues to information the viewer, leaving the viewers to guess what the creator is attempting to say.

In distinction, the chart on the proper is much simpler. It strips away the noise, utilizing simply three colours: gray for context, blue to focus on key data, and a clear white background. Most significantly, the title conveys the primary message, permitting the viewers to understand the purpose at a look.

Technique 4: Communicate Actually of Shadows: The Knowledge of Embracing Uncertainty

Uncertainty is an inherent a part of working with information, and acknowledging it doesn’t weaken your story—it strengthens your credibility. Transparency round uncertainty is a trademark of accountable information communication. While you talk parts like confidence intervals, margins of error, or the assumptions behind a mannequin, you’re not simply being technically correct—you’re demonstrating honesty and humility. It exhibits that you just respect your viewers’s capability to interact with complexity, relatively than oversimplifying to take care of a clear narrative.

Uncertainty can come up from numerous sources, together with restricted pattern sizes, noisy or incomplete information, altering situations, or the assumptions inherent in predictive fashions. As an alternative of ignoring or smoothing over these limitations, good storytellers convey them to the forefront—visually and verbally. Doing so encourages essential considering and opens the door for dialogue. It additionally protects your work from misinterpretation, misuse, or overconfidence in outcomes. In brief, by being open about what the information can’t inform us, we give extra weight to what it could actually. Under, I current a number of examples of how you may embody data on uncertainty in your information story.

  1. Replace on confidence intervals
    As an alternative of: “Income will develop by 15% subsequent quarter.”
    Use: “We undertaking a 15% progress, with a 95% confidence interval of 12%–18%.”
  2. Depart a margin of error.
    As an alternative of: “Buyer satisfaction is at 82%.”
    Use: “Buyer satisfaction is 82%, ±3% margin of error.”
  3. Lacking information indicators
    Use visible cues, comparable to light bars, dashed strains, or shaded areas, on charts to point gaps.
    Add footnotes: “Knowledge for Q2 is incomplete as a result of reporting delays.”
  4. Mannequin assumptions
    Instance: “This forecast assumes no vital change in person conduct or market situations.”
  5. A number of eventualities
    Current best-case, worst-case, and most-likely eventualities to mirror a variety of potential outcomes.
  6. Probabilistic language
    As an alternative of: “It will occur.”
    Use: “There’s a 70% likelihood this end result happens underneath present situations.”
  7. Knowledge high quality notes
    Spotlight points like small pattern sizes or self-reported information:
    “Outcomes are based mostly on a survey of 100 respondents and should not mirror the broader inhabitants.”
  8. Error bars on charts
    Visually present uncertainty by together with error bars or shaded confidence bands in graphs.
  9. Transparency in limitations
    Instance: “This evaluation doesn’t account for seasonal variation or exterior financial elements.”
  10. Qualitative clarification
    Use captions or callouts in shows or dashboards:
    “Knowledge traits are indicative, however additional validation is required.”

You would possibly marvel, “However received’t highlighting these uncertainties weaken my story or make me appear uncertain of the outcomes?” Quite the opposite, acknowledging uncertainty doesn’t sign a insecurity; it exhibits depth, professionalism, and integrity. It conveys to your viewers that you just perceive the complexity of the information and aren’t attempting to oversell a simplistic conclusion. Sharing what you do know, alongside what you don’t, creates a extra balanced and credible narrative. Individuals are way more prone to belief your insights after they see that you just’re being sincere in regards to the limitations. It’s not about dampening your story—it’s about grounding it in actuality.

Technique 5: Reveal the Roots of the Story: Let Reality Journey with Its Sources

Each story wants roots, and on this planet of information storytelling, these roots are your sources. A fantastic chart or putting quantity means little in case your viewers can’t see the place it got here from. Was it a randomized survey? Administrative information? Social media scraping? Identical to a traveler trusts a information who is aware of the trail, readers usually tend to belief your insights after they can hint them again to their origins. Transparency about information sources, assortment strategies, assumptions, and even limitations just isn’t an indication of weak point—it’s a mark of integrity. Once we reveal the roots of the story, we give our story depth, credibility, and resilience. Knowledgeable choices can solely develop in well-tended soil.

Picture 8: Picture generated by the creator in ChatGPT.

Closing remarks

Knowledge-driven storytelling is each an artwork and a accountability. It offers us the facility to make data significant—but additionally the facility to mislead, even unintentionally. On this article, we’ve explored a forest of biases, design traps, and narrative temptations that may subtly form notion and deform the reality. Whether or not you’re an information scientist, communicator, or decision-maker, your tales carry weight—not only for what they present, however for a way they’re informed.

So allow us to inform tales that illuminate, not obscure. Allow us to lead with questions, not conclusions. Allow us to reveal uncertainty, not cover behind false readability. And above all, allow us to anchor our insights in clear sources and humble interpretation. The objective isn’t perfection—it’s integrity. As a result of in a world crammed with noise and narrative spin, essentially the most highly effective story you may inform is one which’s each clear and sincere.

In the long run, storytelling just isn’t about controlling the message—it’s about incomes belief. And belief, as soon as misplaced, just isn’t simply received again. So select your tales rigorously. Form them with care. And bear in mind: the reality could not at all times be flashy, but it surely at all times finds its method to the sunshine.

And yet another factor: if you happen to’ve ever noticed (or unintentionally created) a biased information story, share your expertise within the feedback. The extra we floor these narratives, the higher all of us get at telling information truths, not simply information tales.

References

[1] How to not Cheat with Knowledge Visualizations, Michal Szudejko, In the direction of Knowledge Science

[2] Tobacco producers’ defence towards plaintiffs’ claims of most cancers causation: throwing mud on the wall and hoping a few of it can stick, A number of Authors, Nationwide Library of Medication

[3] AI chatbots distort and mislead when requested about present affairs, BBC finds, Matthew Weaver

Disclaimer

This put up was initially written utilizing Microsoft Phrase, and the spelling and grammar have been checked with Grammarly. I reviewed and adjusted any modifications to make sure that my supposed message was precisely mirrored. All different makes use of of AI (for example picture and pattern information technology) have been disclosed immediately within the textual content.

Related Articles

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest Articles

PHP Code Snippets Powered By : XYZScripts.com