All Half-Truths: How the Media Misunderstands, Misuses, and Abuses Statistics

Edward Hopper, Two Comedians (1966). Image courtesy of Sotheby’s.

It is ironic that the United States looks to the awards made possible by the endowment of Joseph Pulitzer for their standard of excellence in journalism (and, indeed, beyond journalism), given that at the height of the war between Pulitzer’s New York World and William Randolph Hearst’s New York Journal in 1895–98, the two papers could be found competing over which of the two of them could most quickly drive context, research standard, accuracy of portrayal, balance and reason — all of the things that make reporting worthy and vital — out of the news cycle to make room for the purest concentration of sensationalism as a single issue of a newspaper, in America or elsewhere, could reasonably expect to contain.

It should not be contended that Pulitzer’s rough and competitive pursuit of circulation supremacy made his papers entirely hostile to great feats of journalism, nor should it be contended that similar feats have not been accomplished or recognised in Pulitzer’s name since; but it does suggest that the nature of Pulitzer’s name as a shorthand for universal excellence in that field is, as it were, a half-truth.

Fake News was one of our most flatulent and perennial of companions throughout the 2010s, a Falstaffian figure which charmed and disarmed (before subsequently re-arming) many, and blustered its way into the company of the very powerful owing to its exhilarating ability to approximate the force, warmth and ruthlessness of honesty without needing to burden itself with any of the truths or qualifications-of-reality which give true honesty its weight and expense. Fake News’ footprint on the decade just gone was much spoken about. In its brutish desire to make a show of itself, its desperation to do so by any means available— by demonstration of its bulk, its sentiment, its drunkenness, its callousness, or its oddly impressive imperviousness to its own despicableness — it could not be overlooked. It insisted upon itself. It succeeded in cribbing the spoils from the infamy it sought for itself, though some have suggested otherwise.

While episodes like The War of the Worlds [1] give one reason to suggest otherwise, Fake News is young — because our age celebrates youth as a moral good, not merely as an aesthetic or athletic preference, and disdains history, and because Fake News insists on making a show of itself, it takes up a great deal of academic attention, journalistic consideration, and institutional consternation. However, it is Fake News’ cousin, well over a century old, born in earnest in the form which we recognise (or ‘recognise’) now during that conflict between Pulitzer and Hearst and seen prospering ever since, that should demand at least as much of our attention and reformative action.

It is civil, subtle, as much the product of accident as intention. In moving so sinuously, without drawing attention to itself, it has done more damage than could be accurately surveyed. It goes so quietly, with skin so slippery that, not only has no solution to it ever been devised, but much systematic thought can hardly even apprehend it for long. It does not so much as even have a name, but the phenomenon to which I refer might be adequately referred to as half-truth news.

How Does Half-Truth News Come About?

It seems clear how half-truths in the media come about and are spread. Suspicion and prejudice (the kind which goes well beyond the common demographic associations of the word, to encompass any kind of dogma) still claim a greater share of the throne of the collective mind than do the heirs of objectivity. Half-truth news comes about and is spread because it is easier to capture an audience by flattering them, aggravating their sense of the dreadful, or pandering to their existing preconceptions, than it is by confronting them with a demand for sustained attention, information expressed statistically, and any honest attempt to reshape or expand their views. It is hard even for the very accomplished to hold an audience captive with these elements.

There’s one party late of that last parade which might have slipped by without us noticing initially — but take its temperature, and it proves to be pregnant with half-truth potential and as much in need of self-isolation as an Atletico fan at Anfield. The party in question is the statistic.

Image from Wikimedia Commons.

All half-truths do not begin with statistics, but whether or not statistics are the primary means by which half-truth news is birthed or spread is not so much the point. The point is rather that the concept of the statistic epitomises the thinness of the truth/half-truth boundary like no other. Many truths are encapsulated within a statistic. Many half-truths begin with a statistic, too: a statistic over-emphasised, a statistic misconverted, a statistic omitted; a statistic erroneous either in how it was gathered or how it was parsed. Without a keen numerical and analytical sensibility, and absent of knowledge of where the statistic came from, the majority of the reading public would find it hard, if not impossible, to tell the difference.

On top of that natural ambiguity, it’s often hard to adjudge whether that erroneous statistic was formed of a deliberate mendacity of agenda, or a handicap on behalf of the journalist in handling numerals. News has become so reliant on half-truths that journalists themselves may not even notice.

Not all causes for this reliance are sinister, either — our world is more open, more evidently complex than we could ever have afforded to refuse to believe. Given the need to conquer such complication, as many half-truths will come from a quest for truth-proper that has merely come to incompletion or frustration as may come from bad motives.

Nevertheless, in a time when the global populace have become uniquely sensitised to statistics, and many among us a little more concerned about avoiding becoming a statistic than we might have been before, it is time to have that discussion about the use of statistics in media; of how frequently the media exploits public misunderstanding of core concepts in statistics for structured gain; and of how generations of exploitation along such lines have led to a wider pandemic of statistical illiteracy.

Statistics and Sensationalism

It is easy to make use of the term ‘half-truth’ as a de facto synonym for the word ‘lie’, but to do so would be to make the exact same misuse of our categorising intellects as half-truth news itself would seek to exploit. News that is entirely false can be adhered to by a minority — a minority which may be organised, may be powerful, and may, in absolute terms, be pretty large — but it is most often adhered to in bad faith, that most corrosive of adhesive agents. The thing that is most dangerous about the half-truth is not the part that is false, but the part that is true — it is this which gives license for papers to print it, for public voices to defend and advocate for it, and for it to gain easier admittance to the mind’s court of preconception than would an outright falsehood.

The fact that statistics in particular have such blunt impact on readers of the news is what makes them such plush carriage for half-truths. There is no combination of words which can spark sensation in an audience as fast as a statistic can. Tell someone that their risk of cancer is increased by eating bacon, and they may well scoff, snigger even as they smother their copious rashers will plenty of life-giving ketchup. However, tell the same person that 5,400 people per year are diagnosed with cancer related to processed meat intake, and you may well find that they stop chewing (if only to swallow).

This makes the statistic a powerful rhetorical device which is often wielded irresponsibly. The use of statistics to give charge and drama to a news story is often accomplished through the disorienting practice of hopping between percentages and absolute numbers — as we can see the Daily Mail do in this article, which inundates the readers with considerable quantities of both.

This example is merely to demonstrate the potency of statistics as a rhetorical tool — though Daily Mail may rightfully be considered one of journalism’s collective bete noires with respect to its general cynicism, they do not misuse the data in question in the article, however gaddishly they fly between percentages and absolutes. The article a reasonably good example, in fact, of assessing statistics disinterestedly without attempting to put them in service of a more stringent narrative.

There are other examples, however, wherein the presentational differences between percentages and absolute numbers are exploited quite deliberately, in such a way as permits the exploitation itself to justify the entire story, and form a half-truth. Glimpse the following headline (from a considerably more highly reputed publication than provided our last example):

We immediately come to appreciate the centrality of statistics in this piece. Percentages feature in the byline (and that in a 1000+ word thinkpiece no less; quite unusual). Absolute numbers are strewn throughout the opening paragraph as well. However, here we find something considerably more dishonest in the interchange. If it is indeed true that the wealthiest people in the United States donate considerably less money than poorer ones, then we might do well to go beyond the statistics into the psychological factors that motivate donations. Financial largesse has a unique set of customs attached to it in America; the States is a tipping culture, after all.

As Mano Singham shows in his analysis of the same Atlantic article, when making a gesture of charity of any kind, we usually make it spontaneously, off instinct’s impetus, without considering what % of our income we are parting with in the process. Goodness is often irrational this way. Without first obtaining indication of what proportion of the wealthy make charitable-donations-as-income-%, it is redundant to report the discrepancy as such. That $20 one-off donation, spontaneously pledged in a fit of generosity, to the American Red Cross, World Vision, Doctors Without Borders, or even a local church group will represent different percentages of overall income to the orthopaedic surgeon as to the construction worker. Both will feel in that instance that they have acted generously.

Perhaps there is a case to be made that America’s wealthiest are not as charitable as they might be — the fact that only two of America’s wealthiest states by median income could be found on a list of its ten most charitable states suggests as much, though it alone does not suggest why that may be the case[2]. One would be none the wiser either way to read the Atlantic’s article, which presses its statistical half-truth in the service of a conventional “vs. the 1%” narrative tripe.

Of course, the wildly varying amounts different charities pass on from their overall donations makes for another statistically exciting parlour game. Personal philanthropy itself is something which benefactors can be advised to investigate for ethical and value dimensions in the same way they would any other use of their money.

Absolute COVID

The use of absolute numbers was and remains a favourite handle for the representation of COVID statistics as well. One smart-chart that we observed being used by the New York Times in the early stages of the pandemic remains a fine case study for the perception-shaping effects of the presentation of values on a chart.

From the perspective of the cynical, hard-bitten newspaperman/woman, and even more so from the viewpoint of the editor of a strapped bureau, these graphs would insist on their own inclusion in any article which deals with a coronavirus narrative. All of the basic components are there — a line mounting the Y-axis (fetchingly shaded in red, in case your sense of urgency about the matter wasn’t already at a sufficient pitch) and absolute numbers which move into four digits, all of which look considerably more impressive when reprised in-article as emboldened text.

This does not, however, mean that these statistics are being permitted to do their job — or at least, not the job for which they are most qualified. This particular set of visualisations focuses on the idea of “flattening the curve”, a popular buzzword in the early stages of the coronavirus pandemic. However, in order for these statistics to:

  • Give an adequate impression of the degree to which the nations in question were or were not ‘flattening the curve’ and reducing their citizenry’s rate of infection
  • Provide accurate suggestions as to how fast the disease was spreading in each location

…they would need to have been presented not as absolute numbers, but in terms of increase-in-confirmed-cases-as-%-of-population. Telling the reader that Italy and Spain had more cases at the time of reporting than Iran or Germany is not particularly useful. Telling them that Spain had recorded 0.026 cases per capita (i.e around 1-person-in-50), while the more populous Iran was recording only 0.011 cases per capita (i.e. around 1-person-in-100) may seem less straightforward to grasp, but is more representative. It should be borne in mind here, also, that these statistics were published at the very beginning of the pandemic’s global phase, when the public was first beginning to build up their sense of the virus’ nature and the associated transmission risks.

Telling readers that Spain had a recorded-cases rate that was 28% higher than Iran’s, despite having a population that is only 65% of the size, would also have ostensibly been more useful. While not a conclusion in and of itself, this comparison at least provides some direction for further consideration, and a clearer sense of the nation’s respective abilities to control their outbreaks. These statistics are not, as I’m sure the discerning eye will have noticed, as interesting when we contextualise them in this way — but this reaction in itself is a sign of the degree to which we have been conditioned to expect statistics to perform, not to inform.

Of course, there is a further argument to be made that all such statistics in this New York Times article, however displayed, are redundant for all purposes aside from increasing the theatrical spectacle anyway. All nations concerned have differing social customs relative to physical contact; their citizens all travel different mean distances per day; they allocate different absolute funding to their healthcare systems, and according to different priorities; and, moreover, they report and publish statistics according to different methods. While a visualisation of statistics of such various natures would be supremely useful for a government trying to handle the outbreak, and perhaps even edifying for a reading public, such a visualisation would simply be too complex to pack the rhetorical wallop that a single red line scuttling like a scorpion up the Y-axis does.

Charles Joseph Minard’s “Carte Figurative”, depicting the details of Napoleon’s retreat from Russia, has been described as by Edward Tufte as ‘probably the best statistical graphic ever drawn’. It is hard to imagine a visualisation of this order being produced in many conventional newspapers in our day. You can have it interpreted for you here.

So the reader is summoned to feeling, without having been brought in earnest to any greater degree of informedness — but there is a more abstract risk about this propagation of dirty statistics than a pure and simple misinformation of the public. It encourages a broader social disposition that cannot understand probability and is more resistant to perform the requisite analytical calculations required for proper context.

Such toxins so liberally handled come to wither the bearer too — the media itself shows limited ability to appraise the most appropriate statistic in a given moment. It’s not uncommon to come across a graph with a rescaled or even chopped axis, to overstate increase or decline. As regards vaccines to coronavirus, vanishingly few outlets so far have assessed developments in terms of absolute risk, despite it being a generally more informative statistic that is privileged by most serious medical journals as a much more credible measure of a solution’s effectiveness than relative risk (and even this is rarely used by high-circulation news outlets).

The Benefits of Misunderstanding

The resulting states of collective misunderstanding are of supreme value to news outlets, for they can then exercise the principle of infinite jest on their readers: plying them with one set of concerning information to incentivise them to return the next day in hope of being delivered some more hopeful numbers and better news. Of course, such days are vanishingly rare.

We see in these affairs a key difference in approach to statistics according to the genre of media outlet surveyed. What differentiates a publication like the Financial Times from those papers who are more ‘liberal’, if not in view then in their means to state their view, is that the Financial Times, like a bookmaker, are threatened by the immediate invalidation of their value proposition in such an instance as their command of statistics should falter. If they fudge the digits in bad faith, a considerable loss of trust will ensue, because their readership uses the FT’s forecasts to manage their assets. Their readership knows how the statistics work, and the readership are willing to hold the paper to account with their most powerful weapon — their buying power.

For those papers who have no such concern, whose respective audiences do not have the same competencies, statistics are like a luxury resource. Many of those papers are not much more concerned whether or not a statistic is used ‘right’ than Theodore Roosevelt or a maharaja would have been as to whether or not their method of hunting game was ‘right’. It is to be considered ‘right’ if it serves their ends agreeably.

When news was parsed at a local level, statistics invited a greater degree of cynicism. Former journalist and legendary creator of HBO’s The Wire, David Simon, had this to say on the matters of statistics during a talk at U.C. Berkeley.

“I learned as a reporter to start despising statistics, and to regard anything that was cited to me in advance of an argument as dubious, just because someone was pulling it out and using it. I got to be that cynical.”

Before the digitisation of both news and public record keeping, statistical ownership was more exclusively institutional. Somewhat miraculously for the journalist, the internet has provided a kind of data-layer (which fulfills both discovery and query layer functions) by which raw institutional information could be properly parsed before it could be redacted or tampered with. The price of that miracle is, of course, that after the digitisation of the news, the regional element sharply went extinct and journalists were afforded many more means — both store-fronted ( and back-alleyed (Wikileaks) — to parse statistics in raw form than they’d ever had before.

Simon goes on in his address to profess that

“Journalism [prior to the internet] aspired to the toughest job of them all, and that’s the ‘Why’. Who, What, When, Where, Why, How. Of those, only the Why matters, the rest are all bullshit.”

Naturally, statistics are fundamental in articulating the ‘What’ of news — and if we extract the pearl of Simon’s point from both his rhetoric (which is honest) and the broader point his rhetoric serves, we can come to understand that the ‘What’ of news is of course very important, sometimes in ways that only statistics can properly capture. However, the power of the presence of statistics, of the presence of any objective entity in a news narrative, often comes to obscure the need of providing the ‘Why’ by presenting the observation made empirically through parsing of statistics as though it were a brute fact.

Articles on the British labour force’s skill deficit, and the according need to fill the resultant labour shortage with cheaper European labourers, often provide typical examples of this style of incomplete analysis. The fulsomeness with which this sample article’s main points are substantiated with statistics obscures the fact that many of the statistics involved are treated as ends in themselves, not bases for further investigation. For a generation, the supposed unwillingness of British people to do certain forms of work, the need to find European labour to take their place, and the mysterious lack of governmental ability to provide a suitable solution for the skill deficit at the root of the issue have all been taken as simple laws of nature, much like the fact of British naval supremacy was in the late 19th century.

Without the use of statistics to drill into the ‘Why’ of these matters, there is no insight, merely a repetition of the point. Without the use of statistics to drill into the ‘Why’, no social pressure can be exerted on the appropriate agents to encourage the formulation of a solution to these issues. The problem simply stands.

Then, there are examples of how statistics in journalism may abet the Why with great success. A recent article in the New Yorker by Keeanga-Yamahtta Taylor is one such example. As a piece of journalism that supports a thesis via the inclusion of statistics, it is fairly exemplary. It then builds on its use of statistics, into the Why, through hyperlinking. Through this means, we are told of the abjection of a culture of sterilisation-as-birth-control practiced in various American states during the 20th century, and then invited further to understand the philosophies of eugenics and racial prejudice which have caused much of this already objectionable culture of practice to be centred on African-American women.

Taylor’s article sits in a new stylistic lineage of truly digitised journalism, one that Wonk Bridge also occupies in its work, which argues for the hyperlink as being the great journalistic advance of the Early Digital period. Writers who use it judiciously can look forward to an almost indefinite amount of outsourced real-estate when it comes to providing supporting material for their thesis from outside of their word-count [3].

The Conditioning of Expectations

The misuse of statistics engenders situations wherein the public good comes to some harm. This may come in discrete forms. Take for instance this example of the misrepresentation of the concept of exponential decay. This misrepresentation encourages the public to presume that the recession of coronavirus was happening faster than it actually was.

There is not always a pandemic to worry about —and even without defined backboards to measure against, there’s still reason to believe that exploitation of a culture of statistical illiteracy leads to the dangerous proliferation of half-truth news. The greater the degree of elasticity in the ‘performance’ of dirty statistics, the harder it is for the appropriate traction to be gained around honest ones.

The press should ideally be the party to hold improper handling of statistics in institutional circles to account. When the opportunity comes packaged in exciting opportunities for narrative (e.g. as in cases of creative accounting among an elite), the press frequently does execute on this responsibility. Occasionally, as we saw above with 2013’s example from the Atlantic, they over-execute on it. The fact remains that the diligence required to uncover strategic misuse of statistics is demanding and expensive, and few news outlets, most of whom are obliged by ever-waning bottom lines to prioritise low investment editorial genres, have the means or patience to do so. As a direct result of this, and of news media’s complicity in statistical intransigence, social pressure generally sides with the manipulators. It is they who are able to inflate or disguise numbers, and then put them to work, for the sake of ulterior interests, whereas those who would use statistics are left the way honest parties so often are in these kinds of situations: as victims of their own integrity, sat on numbers that, because they are true, serve no interest but the public’s.

Who, after all, is impressed — or terrified — by a single-digit percentage, or a bunch of numbers behind a decimal point?

Moreover, half-truth news and the manipulation of numbers disgrace the noble statistic in and of itself. A large portion of David Simon’s artistic corpus — The Wire in particular, still a show with which no other televisual fiction can do business — is devoted to the exploration of the idea contained within a sentiment he expresses in that same keynote, that…

“In my city, Baltimore, every single effort to quantify progress was an effort by somebody to advance themselves.”

To an extent, perhaps the media is complicit in this state of affairs simply because they are as much in professional ‘need’ of faulty statistics as certain institutional actors are. The media exaggerates statistics to move copy; institutional interests will often marginalise statistics to reduce accountability. It is a dialectic in which both parties can ostensibly get a portion of what they want by playing to the other’s interests.

At Wonk Bridge we remain engaged in investigation as to whether or not there has been a demonstrable increase in overall use of numerical data in English-language news since the 1940s —it is an extensive process full of variables and disciplines that are hard to properly qualify, and of course we remain for now an outlet of humble-enough resource. What is easier to observe is that the number of channels by which we can receive information, including statistics, has increased radically over that period. We have been encouraged to absorb more numerical data, without a proper framework for digesting it. At least as far as these purposes are concerned, there is an inadequate focus on probability in mathematics syllabuses in Britain and elsewhere in Western Europe. Probability is a virtually non-existent entity in the United States’ approach to teaching mathematics. We have, like the bovines of perspective we are encouraged to be, become grass-eaters, albeit without plenty enough in gut for the cud.


It has been partially done for effect, but perhaps you already will have noticed — for an article ostensibly devoted to it, the name ‘half-truth news’ has cropped up relatively few times in the past several thousand words. This is artificial testament to its elusiveness, the way in which it is always several steps removed from the scene of action, usually masquerading in the guise of a discrete phenomenon with its own name. Through this means, half-truth resides in COVID reporting. Half-truth resides in the reporting of crime and violence in our cities. It resides in reporting of the political affairs of foreign lands. It resides in much of what conditions us to feel the way we do about other persons, parties and social interests in our own. It resides in particular luxuriance wherever there are political points to be scored, and no sense of the sanctity of the instruments of truth to bar their usage for such purposes.

After all, truth and fact are not the enemies of narrative, precious, lucrative narrative — they are its most formidable friends, when used correctly. Their allegiance is to be enlisted wherever possible. Nothing abets the corruption of fact like fact itself — that is, so long as fact is not doing its own bidding.

Perhaps it is not so ironic after all that the prize bears Pulitzer’s name — it is in the vain pursuit of his name-as-accolade that so much half-truth news is committed to print. A statistic is cheap soldiery for the cause, after all, and even when Falstaff complained about cheap soldiery, it was all for show. He knew the King’s press was responsible.

[1] …the work of a man who was himself Falstaffian (Mihály Zichy’s Falstaff, to be precise) in his corpulence, self-celebration, and willingness to touch his ear and tell outrageous lies in the service of charm. It is fortunate for us, given Orson Welles’ profoundly adaptable intelligence, energy, and propensity for the manipulation of truth, that he had no systematic conception of self as a politician or any other governor of people (unless those people were Mercury Theatre employees), and too thorough a sublimation of self to frequently obfuscated artistic venture, to have made as malevolent a manifestation of his Falstaffian tendencies as others might have given the same gifts.

[2] The way in which different charitable organisations structure their avenues for donation would be key in any thorough statistical analysis of this subject. The number of American states considered pronouncedly religious on the list of those that are also most charitable is almost certainly correlative — especially given that religious organisations received $123 billion (2016) in donations during an average ‘10s year. If one is religious, it is safe to presume one will be in a place and situation — a church pew — where one will be entreated to be charitable more often than those whom are not, although the American wealthy are only slightly less likely to be believers in God than their counterparts further down the economic ladder.

[3] Whether Taylor’s thesis itself is constructive is another matter. It is not always the case in publishing that an author chooses their own headlines. However, if we take the title and thesis here as one and the same (given they are likely to be taken as such by readers), it is regrettable that such a technically excellent standard of journalism is pressed in the service of a stance that, by incident or calculation, is likely to increase division and act against public health interests at such a critical time.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store