In 2014, researchers at Washington State University raised eyebrows with a study that said killing problematic wolves can actually increase livestock attacks by disrupting the social structure of packs.

However, a new study from the University of Washington is offering completely different conclusions.

According to the Associated Press, three UW researchers who aren’t wildlife biologists but were intrigued by the earlier study analyzed the same data with a different statistical approach.

The researchers found that killing wolves that prey on livestock can lead to a short-term increase in attacks, particularly on sheep. But the year after the wolves were killed, UW researchers said that livestock attacks actually decreased, a completely different conclusion from the WSU study published in 2014.

The WSU study found that for every wolf killed in Idaho, Montana and Wyoming over the previous 25 years, there was a 5 percent increase in the sheep and cattle killed the next year. Total livestock depredations only started going down after the overall wolf numbers were reduced by more than 25 percent. In other words, lots of wolves had to be killed before livestock depredations would decrease.

According to the study’s lead author, Rob Wielgus, the reason is that killing the alpha male or female frees the other wolves in the pack to start breeding. More breeding pairs bring more pups to feed, and these breeding pairs are more likely to kill livestock than individual wolves.

Earlier research has seen similar patterns with livestock attacks from bears and cougars. When dominant males are killed, their territory is taken over by younger predators who can be more reckless.

“It’s like killing the schoolteacher, the animals that keep everyone else in line,” Wielgus said to the Associated Press in 2014. “You’ve got no brakes anymore.”

Since its publication, the WSU study has been heavily touted by pro-wolf groups. However, UW researchers said that when they tried to replicate the original study’s results using what they say is a more accurate statistical model, they found that culling wolves actually did decrease livestock attacks.

“One wolf killed this year would lead to decrease in the expected number of cattle killed next year by 1.9 percent,” the study’s results read. “In other words, three wolves culled this year would be associated with one less cattle depredated next year, ceteris paribus.”

According to the Associated Press, WSU researchers who worked on the 2014 study are criticizing UW’s conclusions, saying the work is based off flawed statistical modeling and that the authors of the UW study are not wildlife biologists and have never studied wolves.

The UW study was published last week in the scientific journal, PLOS ONE, which also published WSU’s research in 2014.