The Washington Post claimed that non-voters handed Trump his victory in 2016. Here is a link: Washington Post
On the whole, the article makes the case that non-Voters had significant majorities of groups likely to vote for Clinton – had they voted. Some preliminary non data oriented remarks:
- Doesn’t that make Clinton’s loss all the more a failure on her part, if this is true? She was not able to motivate those who need the least persuasion to vote for her. Perhaps Trump was effective at casting sufficient doubt on Clinton as to prevent those most likely to vote for her from doing so.
- The establishment doesn’t like to admit that not voting may be a choice. Not voting isn’t necessarily laziness or failure to make a choice. People may make the affirmative decision not to vote. (Some suggest adding “None of the above” to each column on the ballot is the surest way to increase turn-out.)
To the extent this is true, “Non Voter” is a misnomer and, to that extent, many who would never vote for Trump also decided not to vote for Clinton. How some of these people would have voted is a false hypothetical because it is factual that they voted by not voting.
Of course, these remarks can be applied to Trump mutatis mutandis.
Delving Into the Pew Data
Here is the summary exhibit:
In order to perform some analysis, it is helpful to present these percent break-downs in table format:
The Pew Research Center also provides a download to additional information here: Detailed Tables Link
We now examine each of age, race, education and income by using some of this additional information. We especially focus on breakdowns of Trump versus Clinton voters.
We use some of the additional information made available from the Pew Research Center to make this table:
If we want to know if non-voters would have voted for Trump or Clinton had they voted, then we might ask whether the distribution of non-voters across age groups more closely resembles the distribution of Clinton voters or Trump voters. Again, this analysis cannot possibly predict how anyone who didn’t vote would have voted had they voted. At best, it might make certain conclusions a bit more (or less) plausible.
Ocular Inspection of Pie Charts
Here are the pie charts for each kind of voter:
It is clear that the distribution of non-voters is very different from the other three. The distribution is unlike the distribution of Clinton voters as well as being unlike that of Trump voters. The Clinton distribution more closely resembles that of all voters than does Trump’s.
Computing the Difference Between Distributions
We compute the absolute difference (positive difference) of each group of voters from non-voters and make a column chart:
We see that the distribution of Trump voters differs a bit more from non-voters than the distribution of Clinton voters differs from non-voters. Does that mean non-voters would be more likely to vote for Clinton or does it mean that Trump is more remote from those who don’t vote?
The follow-up post (part 2) will analyze the race, education, and income data made available by Pew Research Center.