It seems disingenuous to separate number of arrests from "actual frequency" when the data suggests that the ballot initiative actually made things worse. The lead researcher behind the study shows their bias with their conjecture:

"The study also looked at arrest rates for curfew violations, loitering, vagrancy, trespassing and disorderly conduct. Davis said arrests for those categories increased in Oregon after Measure 110 took effect but said there isn’t evidence to suggest the frequency of those offenses actually increased.

“I personally think it’s probably more … that the cops are like, ‘Well, I can’t arrest you for the drugs, but I told you to move along and you didn’t. So I’m giving you a ticket for that,’” Davis said."

They can say that despite the data, they think actual frequency of those offenses haven't increased, but equally valid to their gut feeling is the anecdotal evidence that homeless activity has greatly increased and most car lots now have 12ft. tall electric fences.