TEDx Fact-Checking Guide

What is fact-checking, and why does it matter?
Fact-checking means going through a script’s empirical claims — basically, anything that can be verified in the real world — and making sure that they are as accurate as possible. There are real consequences to the talks you produce, since people will often change their behavior based on a talk that they watch. So, you want to be sure you are not accidentally sharing misinformation or disinformation on your TEDx stage. It protects the reputations of you, the speaker, the event, and TED at-large.

Getting started: What do you check?

Check any detail or claim that relies on external information. This can take the shape of something like a quote, anything with a number (dates, statistics, estimates, etc.), references to research, the findings of a study or a survey, and pretty much all general claims. Basically, if it can be checked, it should be checked.

The exception to this is a statement that falls within the realm of a speaker’s own opinion (e.g. “I think this will happen”) since this is their own personal speculation. However, a speaker’s personal opinion doesn’t preclude it from needing to be fact-checked. Maintain caution when the speaker uses their own opinion since the evidence supporting the assertion or argument must still be evaluated.

If there is a portion of a speaker’s talk that refers to a personal experience or anecdote, you may want to ask for sourcing on that as well, to ensure accuracy. Some of the things that can be helpful to verify accuracy of anecdotes include:

● Full names, ages, and geographic locations of other individuals referred to in the talk (even if their full names do not appear in your talk)

● Contact info for individuals who can attest to that portion of the talk

● Documentation from that period of time such as emails, medical records or other records

A fact-checklist: How do you know if there are problems with a speaker’s talk?

As a quick guide, here are some questions that will help you figure out if there are issues with a speaker’s
talk. If you answer YES to any of these questions, there’s a good chance that the talk contains questionable
and/or flaggable content.

Are there generalizations (e.g. “most”), absolutes (e.g. “all”), and vague references (e.g. “research
shows...”)? These wordings can suggest that everyone agrees when they don’t, and they can have a
big impact on how people see a topic. Generalizations, absolutes, and vague references can be
easily fixed with words like “many” or phrases like “scientists at Columbia found XYZ,” either of which
can greatly improve a talk’s accuracy.

● Does the talk highlight a lot of numbers from multiple sources? Every reference varies in how it
collects and analyzes data, so drawing comparisons between references can be challenging, even for
experienced researchers.

Is the talk promising a revolutionary solution or business product for a tangible problem which exists
in society? It can be easy to accidentally overextend the implications of conclusions of your results
to-date, especially if those results aren’t yet public. Talks which feature “magical” solutions or
products that claim to solve, rather than simply fight against, existing problems are a red flag. This
issue may require small phrasing changes to help protect the speaker and the audience/viewers from
miscommunication of the idea.

● Is the speaker suggesting new ways to treat health problems? Science- and health-related scripts
should be held to high scrutiny. If any talk highlights research with health implications, you need to
make sure that the research is up-to-date, peer-reviewed, and consistent with the rigorous guidelines
outlined by the scientific method.

● Is a speaker taking credit for all of the work behind an idea (without acknowledging collaborators)?
Published research or projects often rely on the contributions of others, and so the speaker should
make an effort to acknowledge other collaborators or partners wherever possible.

● Did the speaker forget to acknowledge any counter-arguments to or limitations of the idea? A strong
talk idea takes into account at least some of its own weaknesses. If counter-arguments or limitations
aren’t mentioned, the speaker may not be providing an honest representation of the idea. Remember,
the talk is the best opportunity to address questions about the idea. And, if the speaker doesn’t
address the questions, others may do it in the comment section for them.

● Is it unclear what stage the research is in? A talk can highlight the potential of research or work
underway. However, it shouldn’t imply or make unclear that the work is already in its final stage if it’s
still early in testing. The speaker should responsibly distinguish between where exactly the idea is and
where they hope it will go.

Sources to avoid: What makes a reference unreliable?

There are A LOT of sources and websites available for you and speakers to review, but not all sources are
created equally. To help with finding the right ones for a talk, here are some common warning signs that you
can watch out for:

● Is the author unknown? Wikipedia can be helpful, but mainly as a portal to help you track down the
original source of the information, not as a primary source in and of itself. Tertiary sources like
encyclopedias can give background on a topic or point you towards other sources, but these
sources shouldn’t be the references you cite because you cannot know who wrote the entry. This
also applies to articles published in news outlets, unless they attribute where they got their
information and you're able to follow it to the source and verify its accuracy; or institutions that are not
transparent about their funding (for example, a marketing wing of a for-profit company that releases
research about the benefits of the product they produce).

● Do you know where the data really comes from? It’s always good to know where your source got its
information. One common problem is rooted in “data holes”: information cited across multiple
sources with no clear indication as to where it originated from. To help avoid falling into this hole, it's
good for you to be skeptical of statistics that seem to be used everywhere, but never lead back to the
place that generated them. A reputable source will share the methodology of how it arrived at the
info, and could share things such as how many people were surveyed, where they were located, the
time period over which the information was gathered, and whether there were any factors that could
undermine the findings.

● Is there an identifiable bias? Beware of agendas from your sources and the institutions with which
they are associated. A general rule of thumb to follow is that if it feels like a source is trying to sell you
something, it probably is. Try to look for citations that acknowledge corrections and
conflicts-of-interest, that make a reasonable effort to highlight other perspectives, and that state the
sources behind their content’s funding.

Stick to the original source: What makes a citation reliable?

Strong sources can be pretty difficult to find. Indeed, there are a lot of variables that you need to consider
when evaluating claims and research, far more than we have room to cover here. However, a good sample
of the most important variable to consider in what makes a citation reliable include the following:

● Is the source primary or secondary? The best references are typically primary, meaning they are the
original source that produced the information the speaker is talking about (e.g. first-hand accounts of
events or original research). However, some secondary sources, such as an explainer article by an
expert in the field, can also be reliable.

● How strong is the author’s credibility? Does the source come from a credible author or institution?
Are they well-regarded by independent, recognized figures from that source’s field? Strong authors
don’t need to be scholars or have a PhD -- they just need to have a meaningful relationship to what
they are talking about.

● How old is the source? Age isn’t always a bad thing, but make sure references are actually
up-to-date, especially when it comes to statistics and published research. And, the age of the
research should make sense in the context of the talk. For example, it doesn’t make sense for a
speaker to talk about a so-called new technology, but then only have provided older sources.

Topics to avoid: Why is it so difficult to fact-check certain ideas?

We’ve found that some topics seem to be particularly problematic because they are difficult or impossible to
substantiate. We recommend approaching these with skepticism and more often than not, avoiding them

● “Healing,” including naturopathy, alternative medicine, self-healing, energy fields, and energy
balance. This can consist of condoning alternative methods that are unproven, disproven, and/or
potentially dangerous, or exaggerating their results.

● Paranormal, including mediumship, spiritual channeling, extrasensory perception, second sight,
ghosts and spirits, and out-of-body experiences. No observable phenomena fall outside the realm of
scientific scrutiny and asserting that they do is an abandonment of evidence-based thinking.

● Quantum physics, including, but not limited to, personal theories on “energy,” “vibrations,” “waves,”
“consciousness,” etc. can sometimes be used without providing relevant context or explanations
and instead rely on vagueness and mystery to establish a guise of scientific credibility.

● Denial of anthropogenic climate change: Denying the human causes of climate change (global
warming) or misrepresenting valid climate models and predictions is irresponsible and wrongfully
undermines the robust existing body of data and knowledge on the subject.

Topics to approach with caution: How do you share ideas about health and medicine responsibly?

Our primary goal is to provide responsible information to the general public. We rely on TEDx organizers to
prevent the distribution of any misleading information that can harm people suffering from ill health or
disease. Talks should empower people by sharing facts and accurate information, and never be framed as
health or medical advice. Whether the speaker is a qualified health professional or not, using an anecdote
from their personal or professional experiences is not a satisfactory data point in and of itself. While the
following topics are not off-limits, be cautious and ask for peer-reviewed and published research papers that
would support their claims. We discourage speakers from sharing prescriptions guaranteeing universal
disease cures or preventions, especially in the following fields:

● Nutrition: Promoting or prescribing a single diet without consulting a medical professional can be
dangerous— especially when framed around chronic disease. Nutrition science is notoriously
complex and often misconstrued into a “one-size-fits-all” approach. It’s crucial that speakers be
honest about the limitations of nutrition science research and avoid oversimplification of this topic.

● Cancer treatments: Claims about cancer treatment and prevention must be significantly supported
by credible research. Be skeptical of speakers who make claims with too much certainty for the
current state of research, as well as speakers who describe alternative treatments as curative or
preventative methods.

● Experimental therapies: Look out for claims that oversell the current capabilities of new and
experimental therapies, especially in the context of independent clinics. Prescriptive advice presents
a safety issue for viewers when the list of potential side effects is misrepresented or intentionally left
out. The problem comes when speakers discuss combining established areas of research (eg. stem
cells, hormone replacement) with new, experimental techniques that are not so well-established,
giving the audience the potential misunderstanding that the experimental research is as established
as the other research.

● Epigenetics: It is important to accurately represent the nuance and limitations of this field of research
and not rely on epigenetics to support unfounded claims. Do not oversimplify the complex
interaction between genetic and environmental factors on gene expression, exaggerate the current
understanding of epigenetics, or oversell the health effects of individual lifestyle choices, especially in
the context of chronic illnesses and incurable diseases.

Mental health and suicide: Talks should not oversimplify or undermine the complexity and severity of
mental health conditions like anxiety, depression, and PTSD. Talks should not share prescriptive
advice and non-evidence-based methods on how to “cure” or treat mental illness. Personal mental
health experiences should never be used as anecdotal evidence to support prescriptive claims.
These topics are of special concern when talks make claims about overcoming mental illness
through lifestyle choices that target and stigmatize youth.

Tips for working with speakers

Working with speakers to improve the quality of their talks is crucial, but can certainly be a delicate situation.
Here are some tips to help you and your speaker uphold the content guidelines:

Focus on the HOW, not the WOW. We want talks to be interesting, but they must be credible first and
foremost. Determining credibility and steering clear of bad information can be difficult to achieve, especially
when the speaker and their ideas are compelling. We strongly encourage TEDx organizers to curate
speakers who can deliver talks that are more focused on HOW they’ve formulated their idea, rather than
trying to WOW an audience with unfounded claims.

Fact-checking is a non-negotiable. From the beginning, make it clear that for the speaker to be a part of
your event, they must be willing to modify their talk script, if necessary, to meet your fact-checking

Ask speakers to send you their sources. Since speakers write their own talks, they should be able to
provide you with valid sources to verify every claim. (If they don’t/won’t/can’t share sources with you, there is
a good chance that the talk is not evidence-based.) Specifically, examples of valid sources include:

● Direct links to research literature
● Full texts of published articles in periodicals
● PDFs/scans of relevant pages in any books you cite.

Resistance to fact-checking can signal problems later down the road. Be thoughtful in your selection
of speakers. In addition to the many ways to go about speaker selection, evaluate speakers, at least in part,
on their ability to collaborate and handle constructive criticism.

Never assume a speaker who is an expert (or a famous person) doesn’t need to be fact-checked.
Experts are error-prone humans too! Even if a speaker tells you that they’ve already fact-checked their talk, it
is your responsibility to fact-check every single empirical claim.

After fact-checking the speaker’s script, be specific in providing feedback. For claims that could not
be corroborated by sufficient or legitimate evidence, explain what you need from them for the claim to be up
to muster. Do they need to rephrase a claim or remove a generalization? Do they need to provide a different
source? Tell the speaker exactly what needs to be done and keep the process moving along.