Wednesday, August 02, 2006


The Joke that is Online Polls

UK Indymedia is gloating over a poll on MSNBC:
86% Say Bush Should Be Impeached!!!!!!!!!!!!!!!!!!!

Today's MSNBC poll says it all. With over a quarter of a million votes, a full 86% say GWB deserves to go on trial. It's probably going to go to 95% after Saturday's 2 hour CSPAN presentation (8pm) of the 911 Truth panel, hosted by Alex Jones. All I can say is "Holy Sh*t!"
And all I can say is: Bullsh*t.

It astonishes me that online news services continue to carry out these polls. Online polls are nearly always subject to massive fraud. Their outcomes are worthless and meaningless at gauging public opinion. When those hosting the polls broadcast on the results, for example on associated television programs or by publishing the results online, they are misleading their audience who have previously relied on more accurate polling systems and naively apply the same (incorrect) assumption to online polls.

No matter how the MSNBC poll might have been conducted (e.g. phone, focus group, online), it was completely loaded from the start. To the question: "Do you believe President Bush's actions justify impeachment?" Come the options:
*Yes, between the secret spying, the deceptions leading to war and more, there is plenty to justify putting him on trial.
*No, like any president, he has made a few missteps, but nothing approaching "high crimes and misdemeanors."
*No, the man has done absolutely nothing wrong. Impeachment would just be a political lynching.
*I don't know.
Who wrote those answers? I don't care what you think of Bush, those questions are poorly written.

Online polls are riddled with flaws - The fact Indymedia links to one should be proof enough.

The main problem, is ballot stuffing. Supporters of both sides of an argument often mobilize their email lists to vote early and vote often.

Secondly, by tampering with browser cookies a user can vote multiple times. I believe there is software and various browser hacks that allow an even more automated version of this.

Finally, unlike traditional polling, the respondents are not randomly selected, they have chosen to vote. This virtually guarantees voters have a vested interest. The same applies for cable television or SMS polls which charge the voter a few cents to vote. Whilst it could be argued this limits people voting multiple time, the reality is that only people passionate enough (or stupid enough) to spend the money are represented in the final results. It's great for choosing America's Idol, it's lousy for measuring public opinion.

Even if a website, in good faith takes further steps to limit fraud, they can adversely affect the sample size of their poll. For example, by mandating user registration, to allow only one vote per registration, many people won't bother registering and voting. By limiting the number of votes from a single IP address, they could be preventing many other people who browse behind the same proxy server from voting. For example, one person could vote from AOL and subsequent AOL users would be blocked. Again, the sample would be flawed.

In other words, the poll is either subject to fraud, or covers an inadequate group. Either way, the results are highly unreliable.

This problem is well known. There is a reason why most news sites which have polls attach a disclaimer to the effect "this poll is non-scientific and should not be relied upon as an accurate indicator".

The sites should say "poll for novelty use only" as this is all they are really good for. The MSNBC poll has a disclaimer, which you can only read after voting. The background in their disclaimer is very interesting:
One week in the middle of the Clinton-Lewinsky scandal, more than 200,000 people took part in an MSNBC Live Vote that asked whether President Clinton should leave office. Seventy-three percent said yes. That same week, an NBC News-Wall Street Journal poll found that only 34 percent of about 2,000 people who were surveyed thought so.
In other words, MSNBC know their polls are largely unreliable, yet there they are, being relied upon via the front page.

I believe some websites arrange such polls for no other reason than to deliberately draw additional traffic to their site.

Many of these pages earn money for the number of 'impressions' of advertisements shown on their page. Chances are, they don't care about the results of the poll, just that as many people as possible view the page.

Want to know the fastest, probably easiest way to get new visitors to your website? Host a poll like the above or even better, one which asks "who do you blame in the Arab/Israeli conflict?".

Then sit back and watch the visitors (and ad revenue) flow.

Unfortunately for theses sites, the spike in traffic will usually disappear as quickly as it arrived. Questions surrounding the quality of the website will however remain.

Even better, make the options loaded like this

a) the Jews
b) the Muslims
c) both are evil.

That will be sure to generate heaps of traffic.
Relax, take a breath, and read what you're criticizing. It's not portraying itself as a valid poll. It's an online voluntary survey, so it's a non-random sample and, with that, obviously not an accurate public opinion survey.

Moreover, the fact is that Bush's public support has been extraordinarily low. (A more valid cross-section of valid surveys:

I despise Indymedia because so many collectives have turned into willing homes and apologists for terrorist violence and for anti-semitic hate.

But I also despise Bush, because he's been an inept president and has made dreadful foreign policy decisions that are gravely harmful to US security and (in my opinion) core US ethics and values as well.

I think you can build some real attention and awareness of the racist and militant debacle that Indymedia has turned into. If you'll excuse the unsolicited suggestion, why not avoid taking a bite out of your potential coalition and allies by assuming any love of the right-wing or, particularly, Bush.
Thanks anonymous.

In fairness, the Bush example is besides the point. My problem is online polls in general. I chose the example cited not because of its content (howsoever loaded) but the way in which it was taken seriously by Indymedia.

Whilst you correctly note that this one example may not be portraying itself as a valid poll, by way of disclaimers etc, my point still stands; Most people don't read polls this way. Even more so, if it is used as a source for later reporting - as has often happened.

The fact you read blogs (and perhaps Indymedia) probably suggests you have a healthier skepticism of the MSM than most of the public. This latter group are being misled by irresponsible website polls. They were my target and Bush has nothing to do with it. I'm not professing any love for any group. If you have inferred that, I was evidently unclear.

I'm not aware of a single world leader (outside of dictatorships) who could poll 100% support. The Bush poll was dissatisfying as it didn't provide softer answer options, as each option was so verbose.
Ha Ha - "reposted" and 1 commenteer now equals Indymedia uk !!!

You don't really get OPEN PUBLISHING do you?
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours? .