How to Lie with Statistics. Why Journalists and Pundits Can’t Be Trusted with Numbers.

by Pitt Griffin on July 30, 2018 · 0 comments

in Politics, Trump presidency

While Donald Trump’s approval rating hovers around the low 40% for ‘all voters’ — the lowest ever for a president at this point in his first term – his approval rating among ‘self-declared Republicans’, at 88%, is the highest for any Republican President except for Bush Jr. right after 9/11.

Does this mean that Trump is supported by a larger percentage of right-wingers than any previous Republican President? The answer is ‘not necessarily’.

Let me explain. While ‘all voters’ tends to be a stable category, ‘self-declared Republicans’ is a changeable group. It is entirely possible that 2016 Cruz, Kasich or Rubio Republicans, etc. were so disgusted by the eventual nominee they no longer identify as Republican.

Indeed, the number of people calling themselves Republican did decline by 5% points in the first year of the Trump administration. So that 88% is a fraction of a shrinking number. Put another way, you could also say that the high number reflects the fact that Trump supporters tend to support Trump while conservatives who don’t, have shed the Republican label.

But journalists and pundits are not good with numbers so they continue to extol Trump’s ‘unprecedented’ approval.

Further muddying the waters is a WSJ/NBC poll that showed Trump’s approval rating had inched up despite a dismal performance with Putin in Helsinki. The Wall Street Journal ran a piece with the headline “Donald Trump’s Approval Rating Inches Higher, Buoyed by Republican Support”. Other media outlets reported the story in the same manner — which reinforced the notion that Trump was ‘teflon’.

But one poll isn’t evidence of anything. A clearer picture is given by an aggregate of polls. A well-respected version is Nate Silver’s ‘FiveThirtyEight’ — which shows that Trump’s approval is flat at best (see illustration).

Anecdotes are often presented as evidence. They usually take the form of “the sister of a friend of mine” followed by a horrific story of illness. Or for politicians it starts with, “a man/women came up to me” followed by a tragic tale. It is at best proof of nothing and at worse this nonsense leads to widespread stupidity like the anti-vaxxer movement — which has put many children’s health at risk based on hysteria, but no evidence whatsoever — because fear clouds critical thinking.

Statistics should always be assumed to be wrong until evidence strongly supports their conclusion.


Bonus track: Republicans justify their tax giveaways to the rich by pointing to JFK’s tax cuts and the ensuing booming economy. Let’s look at the facts. It is true that Kennedy cut the top tax rate from 91% to 77%. — and the 60s did see a growing economy. So tax cuts are good, right?
Hardly, no proof has been offered. You could equally well argue that a top rate of 77% is ideal and if we want to stimulate the economy we should raise today’s top rate to that level. Any thoughtful person realizes that there is far more to be considered than just tax rates out of context.
But greater consideration is seemingly beyond the average journalist’s ability and the average pundit’s desire. And besides, the average viewer shows little appetite for it.

Previous post:

Next post: