How COVID-19 statistics data is adjusted “from the pencil”

What about the “number of unassigned cases”

Authorities decided on October 15 not to include outbreaks in their daily reports of new infections. The authorities have also decided that they will not take into account the residents of a locality, but only those who live there.

Both decisions result in a decrease in the incidence “in the pen” in the localities and a manipulation of the actual spread of SARS-CoV-2.

GCS reports, at each daily assessment, a variable called “number of unassigned cases” at the bottom of the list of new infections present in each county.

“We now have a pool of 5,000 cases that, when assigned, will result in decreases in daily values ​​as they have not been reconciled this week. They will probably be reconciled by accident next week, when we expect on the forecast curve of At the beginning of the month they reach the assigned values ​​between 11,000 and 12,000 cases per day, but the official values ​​will still be around 10,000, “Octavian wrote on Facebook Jurma.

“Since the beginning of the epidemic, there have been small differences between the county total and the national total. These differences were reconciled by assigning these cases to a category labeled” – “or a fictitious county” county 43. “It is important to understand that It was about cases that remained permanently unassigned, so their emergence had a negligible impact on the epidemic curve.

However, as of November 3, there is a large number of temporarily unassigned cases adding to the number of permanently unassigned cases. In the official report, the title “-” is replaced by “new unassigned cases [temporar] by counties “and we received an explanation that this artifact is due to” changes in the electronic platform. “

It is easier to understand if you follow the curve of new cases in officially reported (red curve) in relation to the sum of the number of new cases by county (blue bars).

Because these values ​​were reported only after reconciliation up to November 3, the two curves overlap, but there is a lag of a few days between the time of measurement and the time of the report we discussed earlier.

After November 3, the number of new cases reported nationwide is consistently higher than the cumulative number of new cases by counties.

At the end of last week we have a surplus of almost 5,000 new cases.

Normally, the correction is made by updating the database with the correct value, but for this we must correct this week the values ​​already reported last week. You realize what a scandal it would be, even though he would be a normal prefect of a computer scientist.

The only alternative to politically correctly reconcile the two sets of values ​​is to subtract the cases you assign to counties today from the total number of new cases for the day or, obviously, to do so so far and not report values ​​that require further correction.

Thus, on Thursday 8,536 cases were registered, but 10,142 were reported, adding 1,701 cases that could be assigned to the counties, and on Friday 9,489 were reported instead of 10,163 cases by reducing 674 cases that were assigned today to the counties in the “buffer “of unassigned cases. This is how the number of unassigned (red bars) and reassigned (blue bars) cases is displayed, in other words added or subtracted from the daily total in the last 3 weeks, Octavian explained. Jurma.

The researcher says that this is a system that produces statistical fog and corrupts daily data, but maintains the correct grand total.

“It is as if a banker takes and adds money every day from the deposits of the 42 clients, knowing that in a possible control he can demonstrate that the missing money is in the bank’s safe and will return it when he has time to count them. . The advantage of this system is that you work with the customer’s money, but it is never lost, so you cannot be accused of theft, ”explains Octavian Jurma.

The authorities’ explanation of the number of unassigned cases is as follows: “We specify that their number is determined by the changes made in the electronic platform through which the results of tests for the new coronavirus are reported and centralized. Thus, according to the new methodology, the test centers directly enter the results of the tests carried out, following that the epidemiological investigation will be carried out from the DSP level and the positive cases will be assigned to the region / locality to which the tests belong. infected people ”.

Benford’s Law and data reported by GCS

The Graphs.ro statistics use another, more complicated method to show that the data reported daily is not necessarily the real one.

Benford’s law says that the numbers in a database have, from a statistical point of view, certain characteristics. Physicist Frank Albert Benford observed that in a database, the starting number in these numbers is 1 in about 30% of the time, number 2 is about 17.6% of the time, and so on, up to number 9 (5% of the time).

Based on this law, statisticians calculated deviations from the rule, which means that the data chain is, from this point of view, corrupt.

The psychiatrist Gabriel Diaconu explained what these deviations mean.

“My colleagues from Geospatial.org, Dragoş Vana from Graphs.ro, respectively, made this exploration possible and I thank them from the bottom of my heart for their efforts. There is, for every county in Romania, a Benford curve that shows how” true “It was the numbers. You will see EVERYWHERE in Romania there are significant deviations from the Benford curve in terms of the number of newly announced cases, the number of reported tests, but also the number of deaths. Significant deviations = someone lied, someone falsified data. I assume this assessment. She is not an accusation. But in a matter of such public importance, and with so many ramifications for the lives of so many Romanians, we can neither mock, nor hide, nor can we beautify people “, Dr. Gabriel Diaconu wrote on Facebook.

Source: Graphs.ro