Free Capitalist Network - Community Archive
Mises Community Archive
An online community for fans of Austrian economics and libertarianism, featuring forums, user blogs, and more.

More Accurate: Online Polls or Gallop, Etc?

rated by 0 users
This post has 2 Replies | 1 Follower

Top 50 Contributor
Posts 2,028
Points 51,580
limitgov Posted: Thu, Oct 27 2011 12:11 PM

Ron Paul always wins online polls.  Everytime.  My question to you is, which is more accurate, online polls with much higher numbers of people involved or something like a gallop poll with much lower numbers of people involved?

I've noticed none of the big media does online GOP polls anymore.  To me, it just doesn't make sense.  Ron Paul easily dominates any online poll, but only places 3-4th in gallop, etc polls. 

Its very easy to design an online poll that takes in your IP address, so you couldn't vote twice.  Yeah, you could unplug your cable modem, plug it back in and wait for it to reconnect, but then, it still might use the same IP.  So, IP address online polls can do a very good job from keeping people voting twice.

  • | Post Points: 35
Top 10 Contributor
Posts 6,953
Points 118,135

It doesn't have anything to do with double voting.  (And no, you can't just unplug your modem and get a new IP.  Try it.)  The point is accuracy.  The argument is online polls are flooded with Paul supporters, and don't give an accurate representation of the population. 

And there is certainly validity to this.  The Daily Paul, Ron Paul Forums, and other major grassroots sites regularly compile and share links of every online poll they can find, so that supporters may be able to participate in every one.  Some naysayers may try to claim it's a few hundred people voting again and again forever, but yes, that's not possible.  But at the same time that doesn't mean the poll is any more of an accurate representation of public sentiment.  You have to remember, even though they always quote statistics by saying: "___ percent of Americans", the truth is it's: "___ percent of Americans that participated in this poll".  And as with any poll, not only do you have to look at sample size, but the sample pool is just as important.  An online poll generally consists of people who:

  1. Have access to a computer and Internet
  2. Visit that particular website
  3. Notice the poll on the page
  4. Care enough take the initiative to click a vote (which also generally means they have a particular candidate in most online polls don't put "don't know" as an option)

That's a pretty select group of people...even more so than the "Americans who have a landline phone and the number listed" (which is also becoming less and less representative of the population).  And those online polls are dominated by Ron Paul votes because supportors not only care enough to vote, but actually actively seek out the polls just for the purpose of voting.

Of course, this in itself does tell us something about Paul's support.  In fact, as noted by Justin T. P. Quinn in this excellent piece — I mean's excellent.  Read it. — "[Ron Paul's supportors] are the paragon of what the ideal political activist would be.  They are arguably the only true grassroots activists in existence right now. They don't wait for direction or leadership from Ron Paul himself.  Rather, inspired by Paul's ideals, they act on their own. Nothing else in history comes close to what they've accomplished."

And consistantly making Paul a landslide winner in virtually every straw vote and online poll is just one example of those actions they take.

However, as far as gauging overall public opinion, it doesn't do much.  But of course, that's due to the unscientific nature of online polling to begin with.  Paul supportors or not, those polls wouldn't be an accurate measure.  (Not that Gallup-type mainstream polls are much better in terms of real-world accuracy, but at least they're a more random sample.)


  • | Post Points: 5
Top 10 Contributor
Posts 4,987
Points 89,490
Wheylous replied on Thu, Oct 27 2011 4:27 PM

Yep. Online polls suffer from 1) convenience sample bias 2) selection bias. They are good for gauging the power of a candidate in terms of immediate supporter control (which sounds a bit scary), but they do not reflect the general public.

Note that even the official polls have problems because of the time of day calling (people who work nights might prefer a certain candidate), trying to poll "likely voters," the type of people who have listed phones, the type of people who bother responding to polls in general, etc.

I was considering doing a national poll by myself just to see how the data compares, but I would need a lot of free time and I would need to consult with my statistics teacher first.

  • | Post Points: 5
Page 1 of 1 (3 items) | RSS