Who Lies More: A Comparison

It recently came to my attention that a graph I created and shared on an online forum has become rather popular. Dan Savage, journalists from CNN, New York Times, Wall Street Journal, and The Verge have all shared the graph on Twitter, Facebook, and other websites. See below:


Many people are also asking questions, such as when it was created, why it includes certain names but excludes others, whether or not statements have been cherry-picked, and what I wanted people to take from it. These are all valid questions and, with the launch of this blog, I can now answer them.

When was this graph created?

I created this graph in March 2016, which makes it a couple months old. When I update it, the results will probably not look much different given the large sample size of statements for each listed politician. Hillary Clinton and Donald Trump had 171 and 111 rated statements, respectively, when this chart was made.

How did I choose which candidates to show?

Basically, I used PolitiFact’s search function to look at every candidate’s scorecard and compiled the data into a spreadsheet. I then sorted the data based on the percentage of statements that are mostly false or worse and generated the chart to compare. I only recorded candidates who have more than 50 graded statements so that the results carry more weight. There simply isn’t enough data on politicians like Martin O’Malley, who has only 18 scored statements, to make any meaningful inference on their truthfulness. At the same time, I included candidates as far back as 2008 (the earliest election cycle in which PolitiFact existed) to widen the list.

Did I cherry-pick statements?

No. I tallied the rulings for all graded statements for each listed candidate. Perhaps the next question is: does PolitiFact itself cherry-pick statements? The short answer here is also no, but it requires more consideration. Obviously, not every comment can be fact-checked. Determining which statements to grade, and then what ruling to assign, is not entirely without subjectivity. However, PolitiFact is transparent in their process, which its editors explain here. They apply consistent logic, thoroughly break down each ruling, and leave room for changes in case further information requires a fresh look.

The bottom line

What is interesting about this graph — and the reason I shared it in the first place — is that it illustrates that a) every politician lies, b) some politicians lie far more than others, and, most importantly, c) one party faction in particular lies more than the others. It is the far-right Republicans who appear to be the worst offenders when it comes to telling falsehoods, in contrast to the more frequently truthful Democrats and moderate Republicans.

I should note that this graph does not compare the seriousness of the lies. Hillary Clinton fares well in this comparison, but that does not imply the lies she has told are not without consequence. If you still question the validity of these results, your issue is with PolitiFact — I am just visualizing their data.