Yesterday news broke about a poll, commissioned by the Boston Herald, which showed an overall lead for the Biden-Harris ticket of 14 points. Which, as a report for the overall poll of 1003 registered and 928 likely voters, in the field from September 30th through October 4th, is a highly encouraging number.
But, wait! There was more. The poll spanned a period before and after the announcement, late Thursday night, that the President had contracted COVID-19. There was a huge contrast between the horse-race numbers pre- and post-announcement. Before announcement, Biden had a 5 point lead over Trump, that jumped to a 21 point gap in the days following the announcement.
A set of cautionary notes:
- The pre-COVID sub-set of likely voters is smaller at 363 (MOE 5+ %)
- The post-CovID sub-set larger at 565 (MOE 4+%)
- The pre-COVID sub-set of registered voters shows a 2-point Democratic advantage
- The post-COVID sub-set expands that advantage to +17 points
- The pre-COVID and post-COVID geographic compositions are also totally different
- Splits are defined as East, Industrial/Midwest, West/Midwest, South & Pacific
- Pre-COVID the splits were 23, 26, 13, 24 and 14 points
- Post-COVID the splits were 21, 18, 18, 29, and 14 points
- Splits are rounded by me — numbers will not add up to 100
- Pre and Post Gender numbers also shifted, markedly, before nearly equal, post announcement cohort was +9% Female.
- White respondents shrank post-COVID announcement from 79% to 58% (66% overall)
- Black or African-American response rates exploded from 10% to 23% post-announcement (18% overall)
So, while the overall poll data may be good, that post-COVID data may be biased by the higher number of female respondents, as well as by other demographic & geographic quirks. Likewise, the smaller Biden lead in the pre-COVID sub-set is likely explained by the higher number of men responding than did post-COVID announcement.
Source for this data is at the link.
Other documents and poll discussion at Franklin Pierce.