Banner
any day today rss X

News and Comment September 2020

Index: 2009201020112012201320142015201620172018201920202021202220232024

19 September (Part 2) - A Cabinet arse elbows his way into the realm of false statistics

I listen to too much talk radio, not nearly as much LBC as a few months ago although it is on right now, because far too much of the Monday to Friday schedule has become so unrelenting left wing. I could listen to James O’Brien for a couple of hours but not the same drivel all day long.

Yesterday morning I was listening to the gabble-fest which is Julia Hartley-Brewer’s breakfast show on TalkRadio. She was interviewing a man who claimed to be the Health Secretary but maybe I got that wrong because his knowledge of anatomy appeared to be zero. Certainly he did not know his arse from his elbow.

When asked how many Corona victims were in hospital in the allegedly worst Covid19 hotspot, Bolton, he didn’t know but it was “a lot”. The official count was two.

The test numbers are all over the place daily but being treated as if they were Gospel and I was confused by all the statistics relating to test accuracy. I have heard doctors arguing about Hancock’s figures and how he is abusing the statistics to confuse the gullible and the easily confused, i.e. me. I didn’t really understand the discrepancies under discussion.
False Positives False Positives
Then I stumbled across this graphic. Ignore the accuracy figure in the yellow box because it is nonsense, but just follow the arithmetic to establish the principle.

A thousand people of whom 5% are, or thought to be, infected makes 950 healthy and 50 with the bug. That’s the easy bit.

But there is a 10% failure rate so the 50 divides into 45 infected and five healthy. On the other side of the equation the 950 drops to 855 negative with the other 95 assumed to be positive tests.

Now swap the wrong results into the correct box, the third row down on the graphic. 140 out of a thousand are now said to have tested postive but only 45 definitely are. False positives rise to 68% of those deemed to be postive by the test.

I am not convinced the graphic is exactly right because I don’t follow every last detail but I get the general idea. Let’s move on to real figures.

On 7th September the number of tests (Government figures) was 169,930. 3,447 were positive. Matt Hancock said on TalkRadio that the false results were under 1%, a quick bit of research says it might be 0.8%.

0.8% of the 169,930 tests creates 1,359 false positives. 1,359 false out of 3,447. That’s 39% of reported cases being wrong.

I’m not sure why I bothered with the graphic now, it suddenly makes sense.

In April the Government was keen to scare us all with the death numbers but the numbers have thankfully fallen to barely double figures and as everyone knows only big numbers make an impression.

A big number is needed to keep the the paranoid sufficiently paranoid. The way to do that is to publish the number of positive tests rather than deaths as was the case during the peak. Then inflate the largely meaningless figure by about 40% and include people who are showing no symptoms at all.

I assume that Matt Hancock is not so stupid that he does not know the truth, although no one can be sure, but he appears to be knowingly engaged on an underhand deception. 0.8% false positives sounds as if it might be really impressive, but it isn’t.

 

Return to the top of this page
Bonkers is a cookie free zone. Not a single one