Ask Snopes: Are Political Gullibles More Conservative or Liberal?

An extended-family in-law of mine frequently forwards political emails that are textbook examples of crackpot urban legends and outright lies.  And he appears to be immune to all attempts, subtle and not, to get him to stop.  He's been repeatedly redirected to snopes.com, the go-to website for researching the "truthiness" of urban legends and chain emails, where the blatant falsities of his emails are easily found.  He's been politely asked to stop.  Nothing has slowed the emails.  He has even admitted to hitting the "forward" and "send" buttons with very little thought about the veracity of the claims within his emails - and he doesn't seem to care.  He seems to be a little more than a spam relay server.

Perhaps you, too, have a family member or friend like this?   We have other examples within our own family, and the family and friends I talk to all have similar stories.  The strange thread of commonality between them all is that these crackpot emails seem to predominantly come from the conservative end of the spectrum. I say "seem" because it's possible that my simple observation is biased or flawed. 

One potential flaw in my assumption might be the sample of emails I am exposed to.  Perhaps my sample is somehow self-selecting so that my family and friends attract a disproportionate amount of conservative emails.  However, I think this is very unlikely since my family is a mixture of the full spectrum of political views, and so I am just as likely to be exposed to left-leaning emails as to right-leaning ones.

Another more likely flaw is called confirmation bias, which is the "tendency for people to prefer information that confirms their preconceptions or hypotheses, independently of whether they are true."  In other words, maybe I'm more likely to notice and remember examples that support my theory, and I subconsciously and conveniently ignore (or explain away) examples that contradict my theory.  To counter this tendency, I looked for a way to independently and dispassionately collect and analyze some data.

Data Collection: In order to test my theory, and to avoid confirmation bias as much as possible, I decided to use data already collected and sorted by Snopes.  The Snopes database is sorted into categories, which includes one for politics.  For my analysis, I selected the political subsections for "George W. Bush" and "Barack Obama".  Snopes researches and flags each claim as true, false, or a mixture of both.   It also categorizes some as undetermined or unclassifiable, which I have ignored for this analysis for hopefully obvious reasons.  For each list, I counted the number of true, false, and mixed claims made about each presidential candidate.



Raw Data Collect March 26, 2010

	Claims about Obama	Claims about Bush
True	8	20
False	54	17
Mixed	19	4

Data Analysis: This analysis starts with one assumption, but I think a reasonable one.  If you review the email titles at Snopes, you will see that the majority of claims about either candidate are negative in nature and therefore, presumably made by the opposition.  Instead of analyzing each one, I am assuming that all emails about each candidate are made by his opposition.  Otherwise, analyzing each claim for content would be both (1) very time consuming and (2) prone to further confirmation bias, which I am trying to avoid.  If someone has the time for further analysis and a method to avoid this bias, I welcome updates and corrections.  For now, I believe the imperfections in this assumption to be small enough not to change the overall conclusions.




True vs False Claims

Claims about Obama	Claims about Bush

Looking only at the true versus false claims, the above pie charts show that claims made about Bush (presumably by liberals) are true slightly better than half of the time, while claims made about Obama (presumably by conservatives) are 1/4th as likely to be true.

True, False, and Mixed Claims

Claims about Obama	Claims about Bush

Adding "mixed" emails to the analysis (emails that contain both true and false elements) doesn't change much.  The proportions of true-to-false remains about the same for each.  However, if we further assume that mixed-truth emails are predominantly false, in that their falsities are material to the overall truthfulness of the email message, then you could add them to the "false" column and conclude that the truthfulness of of both is decreased, although more so for conservatives.  With this assumption, the conservative claims are 1/5th as truthful as the liberal claims. But this assumption would require further content analysis of each claim to back up.

Total Volume of Claims

Moving away from percentages, the above stacked bar graph shows the raw counts of these emails.  In addition to the quality differences, we can see that the sheer quantity of emails regarding Obama is approximately twice as high as those regarding Bush. However, this difference becomes even more striking when the fact is considered that Bush held office (and so was the primary target of such emails) for 8 years, while Obama has only been in office for approximately 14 months.

Rate of Claims per Month

When viewed from an "emails per month in office" perspective, the volume difference is even more striking.  Emails regarding Obama are flowing at nearly 15 times the rate as those regarding Bush.

Conclusions: It is important to remember that a lot of people, both liberal and conservative, do not forward political emails. The senders of these emails are a self-selected (not randomly sampled) subset of their respective political groups, and therefore it would be dangerous to draw conclusions about the larger groups.  What the above data does suggest is that, of those that do forward such emails, the conservatives send a significantly higher volume of such emails (15 times as much), and a their emails are significantly more likely to be false (4 to 5 times as likely).