Tackling misinformation with research and insight

OUR INSIGHTS

In the “post-truth” era, distinguishing between “fact” and “fiction” has never been more challenging. Nowadays, the validity of any given assertion or stat is merely dependent on the number of clicks it receives (or how authentic the meme appears to be). Even with the emergence of fact-checking algorithms, misinformation and disinformation continue to augment at an alarming rate with nearly nine in ten (86%) online users believing they’ve been exposed to fakes news (Ipsos, 2019). Undoing the damage and preventing the spread of “fake news” is no easy task, it’s everywhere we look and, as is often the case, the lie outlasts the liar. However, researchers, data users, analysts and everyday consumers of information can play an active role in reviving confidence. Here are three things we can learn about information and research integrity:

1. Always show your workings out; we need verifiable evidence of our credibility

Consider the stat in the first paragraph (taken from a global Ipsos survey November 2019): “86% of online users believe they’ve been exposed to fake news”. How can we be sure that this stat is credible? Well firstly, like many reputable organizations of their kind, Ipsos adhere to certain trust principles and market research standards which provide a certain level of reassurance and confidence in the information they share. Secondly, but most importantly, they evidence and contextualise their research approach; they provide sample sizes, information on how and when they captured the data as well as who they captured data from. However, not all sources of information provide evidence of their claims and nor do they need to.

Social media influencers, for example, are a testament to this point. They are not held accountable by “standards”, “principles” or “evidence” – there’s just an unspoken expectation amongst their followers that they will influence with integrity. Their success is predicated on the number of followers they have, their credibility is driven by their ‘like me’ or ‘aspire to be’ image and this is what makes them a powerful marketing tool. They are a modern attestation of the Hypodermic Syringe Theory: many people believe in what they say and sell without question. If they speak, people listen.

"In the research and data world, credibility is not a given – it’s earned"

In the research and data world, credibility is not a given – it’s earned. Our ability to influence others lies in our verifiable evidence. Knowing and showing proof of where information comes from is just as important, if not more so, than the results themselves because, ultimately, saying something doesn’t make it so. This is what separates us from “falsehoods”; without visible evidence of things (such as sampling representativeness, accuracy or lack of bias) our stats are just numbers. If we don’t evince where our information comes from, and make it accessible to others, how can we possibly defend our points of view? In short, we can’t.

2. Data is as malleable as playdough; we need to be careful in how we mold it

There are a lot of misnomers about what Data “is”. It often denotes a sense of ‘truth’ and by capturing, extracting and analyzing data we can unearth the answers to questions in a way that is undeniable, unbiased and impervious to conjecture. Fact-check: Data is not a “truth”, data just “is”- it has the potential to be as open to interpretation as any novel, song or painting. People take different meanings from it. While data can be used to formulate evidence to enhance our understanding and guide us towards answers, it can just as easily be misused and bent out of shape (whether by mistake or deliberately). Essentially, data and analyses are only as objective as the people who use them.

In the recent US election, for example, many social media users shared posts with what they believed to be mathematical proof of fraud: “Benford’s Law”. Benford’s Law is essentially a mathematical rule (primarily used in the finance world) to detect fraudulent activity, mistakes or anomalies. In short, the indicator of potential fraud is when the distribution of numbers deviates away from Benford’s Law. In many naturally occurring datasets, from rainfall amounts to town populations, the numbers adhere to the rule. However, in the case of the 2020 US election, Biden’s vote tallies did not follow Benford’s Law whereas Trump’s did. Does this mean fraud had taken place? The short answer is no; there are more complex factors and intervening geographical/demographical variables at play here. The “success rate” in determining voter and election fraud using this approach is practically equivalent to the toss of a coin, “rendering it problematic at best” and “wholly misleading at worst” (Deckert et al., 2011). In fact, the work of Deckert et al. (2011) shows that when applying these indicators to election data from counties in Ohio, it would appear that “significant fraud” has occurred in every presidential election within certain Ohioan counties since 1992.

“Data is not a “truth”, data just “is”- it has the potential to be as open to interpretation as any novel, song or painting”

Benford’s Law shows us that analysis techniques can be used to manipulate data to tell an alternative story (whether in error or intentionally); the saying “torture the data, and it will confess to anything” holds true. There will always be a temptation to skew data to fit our pre-ordained conclusions but the need to share what’s right (as oppose to what’s easy or desirable) is central to informational integrity. Now more than ever, during a time where trust in societal institutions is at an all-time low, researchers must seek methodological objectivity at all costs. The best way to avoid error or deliberate mishandling is through collaboration between users. Getting more analysts and researchers to review and work with the data not only brings more thinking to the table, but also enables them to spot issues more quickly. This leads us to our third and final learning.

3. Don’t let a good story get in the way of the data

Analytical storytelling has become a much sought-after research skill. It’s not enough to simply analyze and crunch data; we need to be able to tell the story behind it. It’s hard to argue with this. If we can’t harness and share the insights in a way that is universally understood by the audience, then what purpose does data serve? Very little. Stories are, and always have been, a form of communication which serves to simplify complexity, create excitement and leave a lasting impression. However, the real skill of analytical storytelling lies within finding the balance between narrating like Morgan Freeman while ensuring that the data plays the protagonist (the Andy Dufresne if you will).

There are generally two main types of analytical storytellers; first, you have those who are information orientated and fixated on the detail behind the data. They know it like the back of their hand and can make sense of the complexity but struggle to convey it to others. As a consequence, audiences become disinterested and most of the key facts fail to compel. In the world of data and quantitative research, these are the more common presenters.

“It’s important to stay true to what’s at the crux of the story: the data insights”

Then you have the second type; those who are excellent communicators with a high narrator IQ. These are people who can leave their audience hanging on their every word through simple articulation but are skin-deep when it comes to the details. The story is remembered but the findings get lost, the messages don’t land as they should and, ultimately, the data ends up playing second fiddle to the rhetoric. These are people with deep narrative intuition; when they speak, they speak in narrative structure and simple problem-solution terms. They speak to people’s hearts, not their minds (Olson, 2015). This is their craft; one that they have perfected over the course of their career.

When it comes to presenting data, there are learnings to be leveraged from both types of storytellers. While there is almost certainly a need for storytellers to master the art of communication, it’s important to stay true to what’s at the crux of the story: the data insights. Without it, we risk potential bias and the oversimplification of something that is very nuanced. That said, to become more well-rounded, data professionals need to improve their narration skills: creative storytelling and accessible language are key to stakeholder engagement. Without it our messages aren’t received.

Concluding thoughts…

In a “post-truth” era it's difficult to separate fact from fiction. We can’t eradicate all misinformation, but we can, and must, learn from its impact and ensure we start rebuilding informational integrity once more. This article, I hope, serves to support how we, as researchers and data professionals (but also consumers of information) can aid the recovery process and overcome misinformation.


Jonny Revell
Research and Insights


External References:

Deckert et al, 2011, California Institute of Technology - Political Analysis

Olson, 2015, Houston, we have a narrative: why science needs a story, http://www.press.uchicago.edu/ucp/books/book/chicago/H/bo21174162.html

Reuters, 2020, Fact check: Deviation from Benford’s Law does not prove election fraud | Reuters

Ipsos, 2019, Fake News: A Global Epidemic Vast Majority (86%) of Online Global Citizens Have Been Exposed to it | Ipsos

MORE ON OUR THOUGHTS

Other insights

Promoting Sustainable Behaviours

Read more

Listening with our Eyes and Ears (Remotely)

Read more