Opinion editor’s note: Strib Voices publishes a mix of guest commentaries online and in print each day. To contribute, click here.
•••
The country, and perhaps much of the world, will watch the U.S. presidential election returns on Tuesday with bated breath, though it appears unlikely that a winner will be determined that night, or even by the next Tuesday. So while we wait, here are four issues to ponder.
The polls
Ever since the first Gallup survey of voter intentions in 1936, polls have captivated the attention of voters and the media as a leading indicator of electoral outcomes. However, the predictive accuracy of polls has recently come into question for three reasons. First, the issue of nonresponse bias is a significant concern since those responding to polls may not be representative of the population at large. Second, random digit dialing (RDD) in the era of mobile phones frequently connects (if they connect at all) with a voter who has moved to another state but has retained the area code associated with her state of origin. Third, pollsters who recognize the need to correct for polling errors have taken to weighting their polls with the respondents past voting behavior, thus systematically under-weighting first time voters, typically young people.
Black voters
There is a mathematical error that people make when dealing with percentages called base value neglect. People often think a discount of 40% followed by a discount of 25% yields a total discount of 65%. It actually is a discount of 55%. By the same token, an 8% shift in Black voter support from Joe Biden to Donald Trump ignores the base value, the “out of how many” element. Comparatively, Kamala Harris’ gain of 4% among non-college educated whites appears woefully small. (Her support among Black voters is roughly 65%, reflecting a substantial percentage drop relative to Biden). But, in absolute terms — the number of votes such shifts might translate into — the arithmetic yields a different conclusion. There are roughly 133 million registered voters in the U.S., of which 14%, or about 18.5 million, are Black. Eight percent of that base is a little shy of 1.5 million votes. Meanwhile, non-college educated whites account for about 50% of the electorate, or 66.5 million voters. Four percent of that base is more than 2.5 million votes. So looking at absolute numbers rather than percentages that do not account for the base value will generally yield a different conclusion, something media pundits rarely consider.
Latinos