Boy, population densities can be so misleading. The typical approach is to just take the number of people and divide it by a given area. This then gives you something like X number of "people per square kilometer." The problem with this approach is that there are countless factors that can skew your result.
Hong Kong, for instance, is really dense. But as a city, it also has a lot of green space, mountains, and other undeveloped areas. Only about a quarter of Hong Kong's land is developed. So when you divide total people by its administrative boundary area, it is going to appear less dense than it really is.
One alternative approach is to use a method known as population-weighted density. The way this works is that you take the average densities of smaller more granular subareas and then weight them by the population of each subarea. It is a little more complicated to calculate, but the overall intent is to try and capture a density figure that more accurately reflects what the average person experiences on the ground.
And this is exactly the method that Jonathan Nolan decided to use in his new website CityDensity.com. What his site allows you to do is compare population-weighted densities across various cities, and then see how it tapers off as you move outward from their city centers.
Once again, it is hard to beat Paris' supremely dense mid-rise built form:
Well, that is, until you check out Hong Kong:
Charts: CityDensity.com
