Statistics, Propaganda and Social Media

The first casualty of war is the truth.

Not just the first, but throughout and after.

I saw yet another comment on social media about the death toll in Gaza at the hands of the IDF being unreliable.  I explained how they are checked and considered reliable.  The person replied saying

Despite the statistical anomalies.

How many people reading or posting on social media are good at research, media analysis and statistical analysis? I was trained in the latter a long time ago and qualified in social science research more recently. I like to think I have a fair idea how to check this stuff and the academic marks I got tended to agree.

So, let’s look for sources for statistical anomalies in the Gaza death tolls reporting. They are almost all Jewish media or Israeli media. So there is bias in the reporting if only those sources are doing so.

Checking those articles, they are referring either to one source, or to another media article also using that one source or other media sites using the Jewish or Israeli sites as their source. So this is not widespread conclusion of differing groups, but of one individual.

The source is Abraham Wyner who produced a paper which was published as “How the Gaza Ministry of Health Fakes Casualty Numbers”. It is unusual for an academic title to be editorialised like that. It implies he was starting with a conclusion and looking for the evidence to support it, which is a red flag. So, who agrees with him?

The Gazan death figures, the underlying data on names and ID numbers and the method used to count them, have been analysed by media organisations, academic research departments, mathematicians and social scientists all of whom have been satisfied they are accurate.

He is one lone exception. Could he be right and everyone else wrong?

In his blog a Bit of DNA, mathematician Lior Pachter wrote ‘A note on “How the Gaza Ministry of Health Fakes Casualty Numbers”’. In this he explains how Wyner carefully chose a specific 15 day sample, used graphs arranged as cumulative rather than an x-y plot to mislead, and took advantage of how the statistics come in batches from verification of ID numbers to provide the false answer he sought. If you go through the comments on that post by other specialists, you’ll get explanations of how Wyner was able to produce his misleading article.  It is more to do with how the data is passed from hospitals to be checked and then on for reporting and it being done in batches, and then utilising that fact to identify a cherry-picked sample to suggest all the data is wrong.

(During the Covid-19 pandemic it was noted how few people died at the weekend but lots on Mondays, suggesting the numbers were false.  That was because admin people like me, in the hospitals providing the statistics to National health England, don’t tend to work at weekends.  So the stats for Monday included Saturday and Sunday.  A similar thing happens with how the identification records of the Gazan victims are validated in batches.)

So, the one person challenging the figures has been debunked.

tl:dr: The source of the statistical anomaly suggestion has been debunked. That one poor source It is used, however, by biased sources as counter-propaganda to claim the death toll figures are false, when they are, actually, very reliable.  This poor journalism is then picked up and repeated across the Internet as if it were truth.  And so people propagate the view that killing in war is acceptable because they don’t have to think about the victims because the numbers might be dodgy.

It is amazing the harm can be caused by one bad academic + rubbish journalism + biased media + poorly educated people + social media.

Leave a Reply

Your email address will not be published. Required fields are marked *