COVID-19 SCIENCE
Misleading mathematics or did Victoria, Australia really need lockdown stage 4?
By Dr Juergen Ude | February 12th 2021
Victoria, Australia has had the highest cases in Australia and one of the longest lockdowns in the world. As of 12th February 2021, Victorians, at the end of their tethers, were again asked to lockdown over two cases. To obtain support, Victorians were reassured that the same experts who pulled Victorians out of the second wave will now ensure Victorians do not have to go through a third wave, requiring a much longer lockdown. Sadly for Victorians, reassurances have never held much sway in the past.
Figure 1 shows the second wave cases for Victoria, as at end of September 2020.
Figure 1: Second wave cases for Victoria, as at end of September 2020.
The last red line is when the tough stage 4 lockdown started. Cases were already coming down meaning there was no data evidence that supported the need for Victorians to have been subjected to the Stage 4 lockdown.
The Victorian State government used a Moving average of 14 to prove that Lockdown 4 worked because unlike Figure 1, which clearly showed cases were already coming down, the Moving Average shown in Figure 2 implies that cases were coming down 7 days after lockdown, which would be expected if lockdown 4 worked.
Figure 2: Deceptive mathematics. A moving average falsely misled Victorians believing lock down 4 worked.
Interestingly we were told in September 2020 that lockdown 4 would mean we can get back to Covid-normal faster. It is now February 2021, almost half a year later, and another lockdown.
Moving averages are deceptive because of the lag. The lag for a moving average of 14 is 7. Therefore, the moving average plot came down 7 days after lockdown, as shown in Figure 2. This is deceptive because it implies lock down4 worked because we would expect a downward trend 7 days after lockdown if the lockdown worked.
To demonstrate the mathematical lag, refer to Figure 3. Figure 3 shows two straight lines. At point 30 the line comes down.
Figure 3: Simple straight-line data increasing until point 30 after which the data comes down.
Refer to Figure 4. A moving average of 14 was applied to this data from point 14 onwards. The moving average is the average of the last 14 points to point 14. At point 15 the moving average is the average of the last 14 points to point 15, and so on
Figure 4: A moving average of 14 applied to the data in Figure 4.
Figure 4 shows a change 7 days later after the actual change, as expected.
There are better mathematical tools for smoothing than the moving averages without lag. The simple moving average is misleading for this application.