As discussed in the Wall Street Journal story, weighting is designed to adjust poll results to give more weight to people whose demographic characteristics are under-represented in the sample and less weight to people whose demographic characteristics are over-represented. Think about how certain demographic groups might end up being under or over-represented (that is, there are too many or too few people in the sample from that group, compared to the population as a whole). Weighting can therefore potentially help fix which types of bias? (Check all that apply.)

selection bias

non-response bias

response bias