The fact that they got such divergent results doesn't prove they're not credible. Let's say the real sentiment in August was even and it's now Sheehy +4. They got samples that were somewhat off from that and didn't massage them.