Where "worst" is defined by a function of pH and sugar content pulled out of the air by a dentist.
Was it so hard to find a chemist way back in aught-9? (Go down to the nearest uni and find a chem major at least...)
The wide distribution of oversimplified, media-sanitized, pseudo-science is likely one of the affectations of Western culture that has kept it from progressing further for a solid 60 years--since television gave the ability for snake oil salesmen to multiply their effectiveness. The internet exponentially increased that ability.
EDIT: For example, there's been some interesting work discussed here: http://wholehealthsource.blogspot.ca/2009/03/reversing-tooth...
You're correct in that their classification of "worst" only applies if sugar and acidity are the most important factors. But they're clearly a large factor in tooth decay, so measuring the contents of various liquids and comparing them to each other is instructive.
Maybe, maybe not. The in-vivo environment might have many significant differences than some in-vitro setup. What's the role of saliva? of bacteria? of genetics?
Don't underestimate the power of the telephone game (http://en.wikipedia.org/wiki/Chinese_whispers). You need to go back to primary sources. It's very common for papers in software engineering, for instance, to confuse qualitative statements from earlier papers with empirical results.
However, the entire article is a hypothesis. It is a best guess by somebody with scientific knowledge, but there is no reason to believe the real formula shouldn't be 10^(5.5-ph)+10(mg/cl sugar) or something completely different.
However, my son and daughter both need a science fair project soon. I wonder if there is a way to formalize this a little.
1. Find a substance that is chemically tooth-like
2. Split it into predefined volumes (little cubes or spheres)
3. Add the beverage and saliva to the "tooth", wait a month, and measure the decrease in "tooth" volume.
4. Use regression to estimate the coefficients for sugar, pH, and whatever else.
If you normalize the pH above 5.5 since it has "little to no effect" things on the lower end of the results are better, with milk being worse than water, and chocolate milk being worse than milk.
The key point for me is why is concentration of sugar in mg/cl? Why not mg/l, or g/l, or oz/gallon? Each one would give a different answer, and all are equally justifiable. (There is also the weird idea that a decrease in pH of 1, which is 10 times the concentration of H+, would therefore be 10 times as bad.)