For the fun of it, I'm going to post below final polling data relating to the Super Tuesday primaries today. If you want, you can look back later, when actual results are known, and see how accurate they were. My idea is that people underestimate the accuracy of polls. For example, everyone says that in 2016 "the polls got it wrong," which is not exactly right. The Real Clear Politics final average of all national polling data showed Clinton ahead by 3.3%. In the event, she won the national vote by 2.1%. Pretty close. No one claims polls are more than a "snapshot in time," so, if as seems likely there was movement toward Trump in the last days of the campaign, the RCP average of polls was really very accurate.
Of course, some state polls were off, like in Wisconsin, Michigan, and Pennsylvania. Most of these were older, so it's hard to say exactly how inaccurate they were: maybe lots of people changed their minds. Not to say there aren't bad polls. The science behind polling acknowledges that. If the poll has a margin of error of 3%, that might mean there is a 96% chance that the measurement is true within 3%. What of the other 4%? All bets cancelled. It's not an exact science, but a pretty good one.
Two more things. There might be a psychological aspect to the notion that polling is "just a guess" and "often wrong." There's a legend in Minnesota concerning how there's always a blizzard during the high-school basketball tournament in March. People can think this because the times there is a blizzard prove the theory is right and the times there isn't a blizzard don't mean anything—everyone forgets "those times." Similarly with polls. No one remembers when they're proven to be pretty good, which is usually.
Second thing is that this particular Super Tuesday might not be the best one to test my theory. Events of the past couple days have roiled the race. Everyone agrees that "the situation is fluid," and polls, remember, are only the latest snapshot. But, as I say, for the fun of it.
The data for California, Texas, and North Carolina are from the Real Clear Politics average of different polls. For the other states, I'm using the "final super Tuesday polling" from Data for Progress.
CALIFORNIA
Sanders 35
Biden 23
Warren 16
Bloomberg 14
TEXAS
Sanders 29.5
Biden 28
Bloomberg 18
Warren 14.5
NORTH CAROLINA
Biden 36.7
Sanders 23.3
Bloomberg 14.3
Warren 10.7
VIRGINIA
Biden 39
Sanders 24
Bloomberg 18
Warren 17
MASSACHUSETTS
Warren 28
Biden 26
Sanders 26
Bloomberg 15
COLORADO
Sanders 32
Warren 21
Biden 18
Bloomberg 16
TENNESSEE
Biden 34
Sanders 27
Bloomberg 20
Warren 15
MINNESOTA
Sanders 32
Biden 27
Warren 21
Bloomberg 16
ALABAMA
Biden 47
Sanders 22
Bloomberg 18
Warren 12
ARKANSAS
Biden 36
Sanders 23
Bloomberg 22
Warren 15
OKLAHOMA
Biden 35
Sanders 28
Bloomberg 19
Warren 16
MAINE
Sanders 34
Biden 25
Warren 20
Bloomberg 18
UTAH
Sanders 29
Biden 23
Warren 19
Bloomberg 17
VERMONT
Sanders 57
Biden 16
Warren 16
Bloomberg 8
Fun factoid: Just saw an MSNBC reporter talking from a polling place in Harris County, Texas, where Houston is. He said there will be more ballots cast today in Harris County than there were in Iowa and New Hampshire combined. I just checked, and this seems plausible: the population of Harris County (4.7 million) exceeds the combined population of Iowa (3.15 million) and New Hampshire (1.34 million) by about 210,000.
Comments