My own experience is that service is more average when tipping isn't an issue. That is it's always OK, or even good, but very rarely excellent or horrible.
Where tipping is the norm, there is more variability. In both directions - sometimes excellent but sometimes horrible. Ok and Good are still the more common, but there are more experiences in those extremes.
I'm not sure why this is.
I couldn't tell from this article if the studies discussed caught that (or could refute it). That is, they all talked about average tips and service, did they measure volatility.