THE FALLACY OF NUMBERS
Editor's Note: Zoe E. ('16) investigates the nature of statistics.
The modern world is one dependent on the apparently indisputable, rational, and empirical realm of numbers. Numbers inform our beliefs, our decisions, and our sense of self – accordingly, it is important that we understand them.
Donald Trump knows well the value of numbers. Polls have reported that as high as 39% of Republican voters plan to vote for Trump in the primary, marking Trump’s domination above the other Republican candidates. And Trump is not quiet about his support: he sports his success in the polls in rallies, interviews, and especially when criticized.
However, Trump’s appearingly overwhelming majority might not be as massive as he thinks. There are numerous factors that may lead to an overstatement of Trump’s support, which frame a critical disconnect between the indubitable and almost enigmatic nature of math and serves to keep a lay-person misinformed and distanced.
Firstly, when statisticians take random samples to find out the voting statistics of an election, an extremely common way of doing so is by calling random numbers. With this approach, there are most likely going to be people who do not respond to their phones. Often, those conducting the research will argue that the group that does answer is in and of itself a random sample, however it is entirely possible that people who do not answer their phone represent a particular demographic, and as a result, that the sample is not entirely random. As a result, excluding the margin of non-response may raise the support level of a candidate.
Secondly, of those who pick up the phone, there are many who respond that they are undecided about who they will vote for in the primary. Similarly, this response is excluded from the data, which could increase the amount of support of a candidate.
Thirdly, those conducting the polls have no effective way of verifying that the people on the phone with which they are speaking are registered as Republican, or even registered to vote at all! In a recent article for the New York Times, Nate Cohn claimed that, in fact, Trump’s “very best voters are self-identified Republicans who nonetheless are registered as Democrats.”
This is not to say that Donald Trump is not the front runner for the Republican party, but that his support is not nearly as expansive as he imagines.
Donald Trump knows well the value of numbers. Polls have reported that as high as 39% of Republican voters plan to vote for Trump in the primary, marking Trump’s domination above the other Republican candidates. And Trump is not quiet about his support: he sports his success in the polls in rallies, interviews, and especially when criticized.
However, Trump’s appearingly overwhelming majority might not be as massive as he thinks. There are numerous factors that may lead to an overstatement of Trump’s support, which frame a critical disconnect between the indubitable and almost enigmatic nature of math and serves to keep a lay-person misinformed and distanced.
Firstly, when statisticians take random samples to find out the voting statistics of an election, an extremely common way of doing so is by calling random numbers. With this approach, there are most likely going to be people who do not respond to their phones. Often, those conducting the research will argue that the group that does answer is in and of itself a random sample, however it is entirely possible that people who do not answer their phone represent a particular demographic, and as a result, that the sample is not entirely random. As a result, excluding the margin of non-response may raise the support level of a candidate.
Secondly, of those who pick up the phone, there are many who respond that they are undecided about who they will vote for in the primary. Similarly, this response is excluded from the data, which could increase the amount of support of a candidate.
Thirdly, those conducting the polls have no effective way of verifying that the people on the phone with which they are speaking are registered as Republican, or even registered to vote at all! In a recent article for the New York Times, Nate Cohn claimed that, in fact, Trump’s “very best voters are self-identified Republicans who nonetheless are registered as Democrats.”
This is not to say that Donald Trump is not the front runner for the Republican party, but that his support is not nearly as expansive as he imagines.