Lies, Damned Lies and Statistics*

While the source of this quotation is questionable how you present data to management can really change the message without distorting the underlying data.  A peer reviewed study of COVID-19 deaths of people over 50 in NSW, Australia, by PLOS One[1] shows:

Anti-Vax: More vaccinated people died than unvaccinated.  A semi-true statement, but omits the fact that more than 95% of the population had at least one vaccination. 

Anti Vax: There is no real need to get vaccinated, the death rate is very low 0.38%.  True, but this average brings in the three vaccinated groups to offset the one unvaccinated group.

Pro-Vax: You are nearly 10 times more likely to die of COVID-19 if you are not vaccinated, compared to if you are fully vaccinated. Also true!

What is the real situation?  Looking at the chart at the top of this post shows a dramatic difference between vaccinated and unvaccinated people. The unvaccinated had a 1.03% chance of dying in the study year! This is much higher than the 0.093% probability of the fully vaccinated.  And by way of comparison, both are much higher than the probability of dying in a vehicle accident in Australia, which is 0.00445% per year.   

The population of NSW is over 8 million, from a public health perspective, the reduction in the demand for beds between vaccinated and unvaccinated is massive – literally 1000s of hospital beds were not required, particularly considering the numbers of people who get sick and live.  Vaccination saved the health system from collapse.

From a personal perspective the decision is more nuanced. The risk of COVID is relatively low and has reduced significantly since this study, but this low risk can be reduced by a factor of 10 by being vaccinated; offset by a one-in-several-million chance of an adverse reaction to the vaccine.  We did the numbers and decided to be fully vaccinated. Others decided to accept the relatively low risk of COVID.

But what does this tell you about project performance data?  Getting accurate data is one thing how you present this information is another.  Far too many controls people stop at the first point – data, and do not think through what message they need to communicate to management, the COVID data supports all three messages above, but the one that really matters is in the graph and the last point. Effectively communicating controls information is a skill in itself, see : Reporting & Communicating Controls Information

*Lies, Damned Lies and Statistics

Neither of the above options seem to be correct according to the University of York: https://www.york.ac.uk/depts/maths/histstat/lies.htm

[1] The full study is at: https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0299844#pone-0299844-g004