I think that using survey data for this type of statistic is going to be much, MUCH less accurate than other forms of data gathering. Different social pressures, human memory, and biases will skew the results heavily.

One could easily request data from multiple snack\sweets manufacturers, look at sales numbers per capita across a variety of brands, and then come to much better conclusions. This would require contacts and influence with many big companies though.

Sure using Google surveys will be cheap and easy so that even a college student could knock out a report for their health class in a week, but I wouldn't put any stock into this data.

I'm from Utah, and I have no idea how we ranked bottom on the list. Utah probably consumes more sweets than average due to cultural pressure to avoid other vices, such as alcohol and cigarettes. Maybe Utah parents don't feed as much to their kids(unknown), but Utah adults eat a ton of sugar.