Under the CEU bottleneck model, we observed consistently minor effects of phasing errors. The difference between ARG inference methods is much larger than the difference in performance between using true haplotypes and computationally phased haplotypes.
5/7
