I have blogged on this point at least 10 times, but it is worth doing it again.
This time I am going to use "obesity". The USA is one of the most obese countries in the world. Data shows that about 36% of USA citizens are obese. If we were to normalize data related to Covid-19 it would have a very significant impact on what the media is reporting about the impact of Covid-19 around the world. I think that anyone who understands normalization would agree. As I see it there is another problem with just using the percentage as the basis for the normalization. In many cases percentages is an adequate measure, but not when it comes to obesity. Let me explain my position.
It is my position that obesity is not linear. Now, what does than mean? If a country had only one percent of its people who were obese, you can expect that this small percentage of people were just over the line or just barely obese. Let's think of is this way. As the percentage of people who are obese, the degree of obesity also increases. That means that a country that has 5% of its population who are obese a country that has 10% of its population who are obese is not have 'twice" as bad of a problem. What we should do is to integrate the relationship. That means we should calculate the "area" under the curve. It is not actually a curve, it is a straight line, but higher on one end than the other. If the science were to calculate the slope of the line, we would then have a much better understanding of the impact of obesity.
I will try to explain this concept to those who do not remember the simple math that they should have learned in high school. Let's say the slope of the line is 5%. If a country is 1% obese, it counts as just 1%. If it has 5% who are obese, the person who is in that last 5% is actually 1.22 times as obese as the person who is in the first 1%. Now, if we normalize the obese data for a country that is 3% obese with a country that is 36% obese, the country that has 36% obesity is much more than 12 times more obese and here is why.
One way to calculate the impact of obesity is to use the compound interest tables. Here is what I get when I use 5% the uniform series compound amount factor for the country with 5% obesity the factor is 5.526 and for the country with 36% obese it is 90.320. If we were to have normalized the data just on the percentages, the country with 36% obesity would be 7,2 times as bad as the other country. If, on the other hand, we had use the model that I have proposed the worst country would now be 16.34 times as bad. That is a very big difference and if not calculated as I have done would have lead to the wrong action by the medical community.
Next time you consider normalizing data so that it can be honestly compared, think about the above; I just how that most of you actually understand this blog!
