Media reports of Quinnipiac poll overstate possibility that Trump is in trouble in the Lone Star State
♦By Rich Shumate, ChickenFriedPolitics.com editor
AUSTIN (CFP) — Media organizations have been trumpeting results of a new poll that purports to show President Donald Trump trailing former Vice President Joe Biden in Texas — a somewhat shocking result, given that a Democrat hasn’t carried the Lone Star State since 1976 and hasn’t been close since 1996.
But a closer look at the numbers of the Quinnipiac Poll show that, from a purely statistical perspective, those reports may be misleading.
The poll, released June 5, found Biden leading Trump by a margin of 48-44 percent among registered voters in Texas, with a margin of error of plus or minus 3.4 percent.
What that margin of error means is that Biden’s actual support among registered voters in Texas could be as high as 51.4 percent or as low as 44.6 percent; Trump’s actual support could be as high as 47.4 percent or as low as 40.6 percent.
Because those two intervals overlap, it is statistically possible that Trump is actually ahead, as the high end of his support is above the low end of Biden’s support. So although it is probable that Biden has a lead, because the range of his support is higher, the poll does not absolutely establish a lead for Biden.
In order to establish with statistical certainty that Biden is leading Trump, his lead would need to be at least 6.8 percent, or double the margin of error.
The poll’s results for hypothetical match-ups of Trump with other Democratic 2020 contenders are even more muddled because neither candidate has a lead larger than the margin of error, which means that no one can even be said to even probably be ahead.
The poll results do, however, indicate that at this early stage of the race, the battle for Texas is closer that what might be expected and Biden appears to be performing more strongly against Trump than other 2020 competitors, including two Texans, former U.S. Rep. Beto O’Rourke and former San Antonio Mayor Julián Castro.
In 2016, Trump carried Texas by more than 800,000 votes, a margin of 9 points.
The Quinnipiac poll did turn up one result well beyond the margin of error — 60 percent of Texas don’t want to see Trump impeached, while only 34 percent do.
That’s not really how “margin of error” works. In a poll, margin of error is an estimate of the interval in which the actual result will lie with a 95% probability. So the actual outcome will be outside of margin of error about about 1 time out of 20. The biggest problem with the poll is that 48% and 44% add up to 92%, and in an actual election there will not be anywhere near as many people not voting for either of the two.
Thanks for the comment. Good point that the total number of voters will be much different in an actual election, and, indeed, there is only a one in 20 chance that the margin of error is different, either larger or smaller than 3.4 percent. However, when the plus or minus calculations are applied to the results for both Biden and Trump, there could still be situations where Trump would be ahead if all registered voters in Texas were surveyed, instead of a random sample. If Biden’s lead were more than 6.8 percent, that would not be the case.
My larger point here was that the implication that Trump is behind Biden in Texas was being blown out of proportion based on these results, which is something too many news outlets do routinely — to my considerable annoyance.
Thanks for reading ChickenFriedPolitics!