The conclusion here might be due to a logical fallacy which I learned in college. We called it “correlation does not imply causation.” In other words, when two things are related statistically, one cannot always know for sure the backstory of how that came to be.
Sure, people *might have a tendency to be less honest in a personal interview than online.
Or, it might be that phone interviews sweep up people who still have landlines while online interviews sweep up millennials who are wired more than the average older person . . .
See how that works? We have statistics. We don’t have an automatic explanation as for why they fall the way they do.