Eating candy daily as a child is correlated with adult violence.
This does not mean that candy causes violence. Correlation is not causation, as the article does a good job of pointing out. There are actually three possibilities, even though the article only mentions two.
Possibility 1) Candy causes violence. Perhaps being spoiled as a child leads to poor self-control as an adult.
Possibility 2) Violence causes candy. Perhaps violent tendencies cause the child to be bribed with sweets.
Possibility 3) Some other, unobserved factor is responsible for both violence and candy. Perhaps bad parenting results in both spoiling and later criminality.
The only way to be sure would be to run a randomized controlled trial, giving candy to some children but not others, making sure the children are otherwise similar, and waiting to see what happens. This would obviously be difficult. Another way would be to find a natural experiment, something that changes the amount of candy consumed without changing anything else about the child's environment. Perhaps you could find a time when the price of candy jumped because of a tariff or a sugarcane crop failure or something, and then look to see if that had any effect on crime down the road.
Economists do that kind of thing all the time, but it requires both a lucky accident to give you the data, and really high-powered statistical techniques. Simple linear regressions just don't do the job.