Paul Gunson has done a statistical analysis of what are called house effects for different NZ polling companies. A house effect is basically the difference between the average results for that pollsters and all the other pollsters. He concluded:PollLab estimates TPU–Curia to have the smallest overall deviation from the estimated underlying trend of any active polling organisation analysed:An X user called The Overhang also has done in the past a similar analysis, which is here.