A recently published poll suggests that a majority of Americans now support warrantless online surveillance. While the headlines seem to tell a tale of turning public opinion on the subject of privacy, a dig deep into the numbers shows a much more unclear picture.
If you are a supporter of personal privacy online, you might have read the story about how a majority of American’s now support warrantless online surveillance. The story came from the Associated Press. The thought of American’s turning their backs on basic civil liberties is certainly a gut-wreching thought for those fighting dragnet surveillance in the US especially in light of the recent passage of CISA.
The article itself proclaims the following:
A majority of Americans say they support warrantless government surveillance of the Internet communications of U.S. citizens, according to a new poll by The Associated Press and the NORC Center for Public Affairs Research.
It’s at least somewhat important for the government to sacrifice freedoms to ensure safety, most say in the survey.
The blockbuster findings were picked up by Slashdot and, no doubt, further spread elsewhere.
From the perspective of an advocate of personal privacy, this, on the surface, sounds like a pretty grim picture. It’s one thing when the government tramples on civil liberties, but it is quite another when a “majority” of people are perfectly fine with it.
Such thoughts may lead to questions of what went wrong in the messaging about the importance of basic civil rights. How did we get to the point where it seems that the government is winning the support of the people even after the NSA controversy, the revelations revealed by Edward Snowden, the infamous splitter room at AT&T, and many more similar scandals that have been brought to the light of day over the last several years?
Upon further reading of the article, however, one thing to note is that the actual question being asked is nowhere to be found in the Associated Press article. That actually is a major problem given that push polling is a known problem in statistics. As a general rule, it is hugely important to know how the questions are being framed. The reason is that statistics can be manipulated in many different ways. One way is to manipulate the data coming in so more people appear to be supportive of a particular idea.
Lucky for us, it is possible to the source of the finding. In this case, we can go to the studies findings and analyze the information directly.
The question about online surveillance, which inspired the blockbuster headline, is this: “As a way of responding to terrorist threats, do you favor, oppose, or neither favor nor oppose government analysis of internet activities and communications, including those involving U.S. citizens, without a warrant, to watch for suspicious activity that might be connected to terrorism?”
As you can see, there is a whole range of problems just on the question alone. For one, it is extremely wordy. when you have an extremely wordy question in a survey, you run the risk of confusing people. This, alone, can actually taint the results of the poll you are conducting.
The second problem is that the question is actually quite vague. This leads to confusion as to exactly what is being asked. Is this question asking if you are OK with the government monitoring public communications (i.e. your Twitter messages)? Is this question asking if you support the government snooping on private communications? Is this question talking about targeting people who are known extremists? Is this question talking about dragnet surveillance of everyone whether or not there is reason to suspect them of anything? while it is possible to carefully decipher the question and figure this out, the added factor here is that you are asking potentially random individuals. Not everyone is going to spend 15 minutes trying to figure out how to even figure out what is being asked, let alone answer such a question effectively.
The third problem is the conclusion as presented by the Associated Press. The claim says that a majority of US citizens “support warrantless government surveillance of the Internet communications of U.S. citizens”
When you compare this claim to the question being asked, there is a whole world of difference between the question and conclusion. The claim simply assumes that all surveillance is fine for American citizens. The question vaguely asked about surveillance, but doesn’t do a very good job of elaborating precisely what kind of surveillance we are talking about. If the claim is saying that all forms of surveillance are fine for American citizens, then one can argue that the poll never really made such a finding. If the goal of the poll was to figure out if whether all forms of surveillance was fine for American’s, then this problem should have been broken down. A great idea would have been to ask a series of simple questions like this:
1. Do you support warrantless surveillance of suspected individuals’ public communications on the Internet?
2. Do you support warrantless surveillance of all peoples public communications on the Internet?
3. Do you support warrantless surveillance of suspected individuals’ private communications on the Internet?
4. Do you support warrantless surveillance of all people’s private communications on the Internet?
Even if you find that these questions could be better refined, this is still a much clearer way of handling the question than what was found in the poll in question – and would have likely produced better results.
One more problem of note is that the claim is a majority of US citizens support the surveillance. The problem is that, by majority, the poll meant 56% are supportive. As a result, this is merely the technical definition of a majority. In reality, this is a very slim majority. If you read the fine print, there is a note about sampling error: “The overall margin of sampling error is +/- 3.9 percentage points”. Because of this margin for error, the “majority” statistic is actually much tighter. For all we know, the true number under these circumstances could very well be 52% because that can be within the margin for error.
Because of all these reasons, the poll would have to be tossed out because it simply isn’t reliable. Between the vague wording to the slim majority, this poll fails to add anything to the surveillance debate. If someone were to use these findings to make a point, the poll should be treated as debunked.
Drew Wilson on Twitter: @icecube85 and Google+.
