The article titled "Trump is leading the polls, but there's plenty of time for Biden to catch up", by G. Elliott Morris on 538 (ABC news) discusses how well the current polling between Biden and Trump can be used to predict the election. Currently, Biden is losing to Trump in the presidential election polls. The article begins by describing how this does not necessarily mean Biden will lose the next presidential election. Morris points to many historical examples where polls at this time of the election cycle were off by a shockingly large amount of points when compared to the real election results at the end of the year. This could be caused just by people changing their minds, but this article explores what statistical issues could explain how reliably we can use these polls as predictions for the election.
The article considers that we have to factor in that America is more politically polarized than it used to be, making these early predictions more reliable in the sense that a Republican supporter is less likely to change their mind and vote for Biden (and vice versa). This political polarization also affects the outcome as politically polarized elections tend to be closer, so small changes in support over this time can still determine the outcome of the election.
Additionally, with political polling response rates being at an all time low there are more statistical adjustments required by pollsters to make their predictions. These adjustments can lead to what Morris calls “non-sampling” errors, which are mistakes in trying to adjust for the smaller sample sizes. This adds another layer of uncertainty to the current polls predicting the next election.
In conclusion, this article summarizes some reasons why current polls can be misleading in predicting how the presidential election will end up. While polls show that Trump is currently leading against Biden, the election is far from decided.