We conclude that there is no evidence in this analysis of a non-response bias.4 With acceptable response rates, the differing opinions in the results and the similarity of the early vs. late responses, it is unlikely that non-response bias influenced our results significantly