But a ground ball isn't the only thing that produces a run in that situation.
Dunn is a shell of his former self but, in his prime (say 2005), he's going to K 25% of the time, walk 17% of the time, hit a ground ball 35% of the time, and hit a fly ball 47% of the time. Of those fly balls, 22% are going to leave the park.
Why ask him to cut down on his swing if he's going to hit a ground ball 35% of time anyway? You're giving up any shot at the absolute best option (HR) in favor of increasing the odds of one of the worst options with no guarantee that he can avoid a strikeout.
EDIT: I realize that those add up to way more than 100%. Fangraphs calculates K and BB rates per PA. GB and FB are calculated per ball in play, which is way less helpful in this situation. Ignore everything I said until I can find a per PA calculation for those two.
EDIT#2: If you include HR in BIP, Dunn's rough per PA FB rate is 18% and his rough per PA GB rate is 14%. So he's going to K more often than anything else. The question is, is it worth it to try to increase that GB rate/decrease the K rate at the expense of the BB, FB, and HR rates. I don't know.