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.)