UK newspaper, The Sun, is no stranger to controversy. Indeed, as part of News Corporation, it is currently mired in the phone-hacking scandal that engulfed its now killed-off sister paper, The News of the World. (The sheer speed of developments prevents me from summarizing that scandal here, but regular updates are being posted to the relevant Wikipedia page.) The Sun has also been the subject of a decades-long boycott by the people of Liverpool following a series of fabricated stories regarding the 1989 Hillsborough stadium disaster. And one of its most famous front-page headlines, from way back in 1986, relates to another specious story, which claimed that the British comedian Freddie Starr ate his girlfriend’s pet hamster after she refused to make him a sandwich. Yes. It’s one of those types of newspapers.

As such, rarely does The Sun dabble in science news. However, this week it carried a story relating to a scientific paper recently accepted for publication in the prestigious academic journal Proceedings of the Royal Society B: Biological Sciences. Now, as an academic journal, PRSB is not to be sneezed at (its impact factor places it in the top 10 biology journals in the world). Moreover, as the paper is as yet only in press, we surely must give credit to The Sun’s journalists for keeping abreast (sorry) of the relevant scholarly abstract databases. Then again, maybe a press release was involved.

The Sun was only too glad to cover the findings of this particular research. After all, the study appears to corroborate some of the claims of physiognomy (namely, that you can judge a person’s moral character simply by looking at them), which is fairly consistent with The Sun’s general approach when reporting diversity issues. In this particular case, the research findings suggest that liars have different shaped heads compared to other people, about as Sun-friendly a research outcome as is possible to imagine. However, the way The Sun explained the findings was not without irony…

Firstly, the research. The PRSB paper — entitled “Bad to the bone: Facial structure predicts unethical behaviour” — describes experiments conducted at the University of Wisconsin-Milwaukee, which were designed to test the belief that facial width-to-height ratio (WHR) is related to moral behaviour. Based on a number of previous (albeit ambiguous) findings, the researchers predicted an inverse association between WHR and morality — in other words, the greater a person’s WHR, the greater the likelihood they would engage in deception and/or cheating. They conducted two experiments:

• They assessed deception by asking participants to engage in a role playing exercise (by email) in which they pretended to negotiate a property sale. Deception was then inferred from ratings of transcripts of what they said during the negotiation.

• They assessed cheating by asking participants to privately roll a dice (actually, a virtual dice-rolling website was used) and to record the number that was produced. Participants were told that this number would represent the number of tickets they would receive for a subsequent lottery. The researchers then simply checked the statistical association between participants’ WHRs and the numbers reported. They predicted that people with higher WHRs would enter higher numbers (thereby entitling them to more lottery tickets). As dice rolls should really be random, no such statistical association should emerge; if it did, it would indicate cheating.

Suffice to say, the data from both experiments confirmed the researchers’ predictions for men (but not for women). Men with higher WHRs were found to be more likely to engage in both deception and cheating.

Now, like all studies, the validity of the conclusions can be questioned in a number of ways. For example, the methods used to confirm the occurrence of deception (i.e., judgements of transcripts by independent raters) might lack precision. However, as the researchers ensured that raters were unaware of the individual WHRs of the various participants they were rating, this imprecision would not explain the emergence of a statistically significant association between WHR and deception.

Of course, statistical significance can occur by chance alone. In the first experiment, only 13 men from a sample of 50 engaged in deception at all. To establish whether these had higher WHRs than the others amounts to a statistical comparison between one group of 13 cases and another of 37 cases. By conventional standards, such group-sizes are very small indeed (in statistical terms, they produce 33% “power“, far less than 80% which is the conventional threshold of reliability). On the other hand, while there remains a risk of error, the very fact that a statistical effect was found could be taken to suggest that the analysis was indeed sufficiently powered.

Needless to say, The Sun stopped somewhat short of exploring these methodological minutiae. Instead, they were happy to declare the outcome with tabloid-style confidence:

SOMETIMES you come across a bloke who is just downright nasty – and it really is written all over his face.

Men with wide faces are more likely to lie to get their own way and cheat to make themselves some extra cash, according to a report in the respected Royal Society journal. This certainly rings true in the cases of moon-faced US gangster Al Capone, East End crime lords Reggie and Ronnie Kray and mega-mugged North Korea dictator Kim Jong-il…

…So next time that bloke with a huge head insists he’s got your best interests at heart, remember to give him a wide berth.

To illustrate, they provide us with a nice picture of Reggie and Ronnie Kray, presumably to help explain the concept of “blokes with huge heads“:

But there’s just one problem. Their heads are the wrong shape. By any standards, the Krays’ heads are more tall-and-narrow than short-and-wide. Remember, as far as the research is concerned, it is not the “hugeness” of a bloke’s head that counts, it is the ratio between the width and height of his face. The bigger this ratio is, the wider the berth you should give him.

In the PRSB study, the average WHR for men was 1.74. In other words, the typical man’s face was 1.74 times wider than it was tall. And, remember, this is just the average — only a quarter or so of men (generally those with the highest WHRs) were cheaters or deceivers. In contrast, the Krays appear to have WHRs that fall well below the average of 1.74, placing them on the opposite end of the morality spectrum. In fact, judged purely on a WHR basis, the Krays are ordinary decent fellas who wouldn’t hurt a fly.

The method of measuring WHR is pretty specific, but an explanation is presented in this paper. Briefly, width is indicated by the distance between the left and right zygion (also known as the bizygomatic width), while height is indicated by the distance between the upper lip and brow.

Now, had The Sun really wanted faces to illustrate the idea that some people have higher WHRs than average, they might have left the Krays alone and instead used some of the following:

Now, I’m not saying that any of these guys are liars.

In fact, I am completely sure that none of them are!

Ahem. But I do think such pictures can help us to draw at least two pertinent conclusions.

Firstly, even the authors point out that these particular research findings are at best suggestive. And what they suggest is merely a trend. In other words, even if the findings were independently replicated, there would no basis to draw conclusions such as “all liars have wide faces” or “all wide-faced people are liars“. As the association is merely a trend, there will always be lots of exceptions. Information about WHR is completely useless for diagnostic purposes in individual cases.

Secondly, when it comes to reporting the outcomes of studies like this, newspapers can be extremely selective in the examples that they use. Instead of drawing new conclusions from new research findings — in other words, instead of actually learning something — newspapers will tend to use the new findings to bolster their pre-existing views, and then choose their examples accordingly.

Which goes to show that, if you fly too close to The Sun, your pants might catch fire.