Groan, statistics.
There are two ways to do it. One is fatalities per hour of the activity. The other is fatalities per year, ignoring any differences in popularity and length of activity. Since insurance billing is by the month, and such statistics usually come from insurance actuaries, such articles usually use fatalities per month or year. I agree that it would make more sense to rate the danger level by the hours of activity or exposure, but obtaining the necessary numbers is difficult.
I don't have a clue why getting killed by lightning while playing soccer is more likely than golf. Perhaps because there are more players (targets) on the field in soccer than in golf. More likely the deity in charge of soccer is pissed off because of all the bizarre rule changes: and has decided that some divine intervention is required.