Second Place is the Second Loser

In a recent poll by GlobeScan, Inc., public opinion on several of the more influential countries in global politics was surveyed. The results?

Well, as far as negative opinion goes, the world doesn’t seem to hate us as much as it hates Iran. Iran bumped us out of first place, where we were in last year’s poll.

The polling agency goes into surprising detail about who specifically doesn’t like our policies. The breakdown is this: African countries love us like the rich uncle that never colonized them. Middle eastern countries for the most part don’t like us, with the exception of Saudi Arabia, which can’t make up its mind either way, and Afghanistan, who loves us in a way that should be illegal.

I’m guessing that when the polling agency made its composite numbers, they were weighted by the nation’s percent of the global population. I guess this since the agency conducting the poll meets international standards, whatever that means.

Their polling methodology is listed in a pretty obvious location, where you can see who was asked what questions, where, how, as well as what the weather was like. Most of the countries surveyed had sample sizes of around 1000 represented. The notable outliers are South Africa with 3497 people surveyed, Afghanistan with 2098 people surveyed, and Iraq with 2200 people surveyed. To someone who knows more about statistics, my question is this: does this constitute a sampling error and if so, would such an error provide a significant bump in either direction, positive or negative?

No comments yet to Second Place is the Second Loser

  • Actionstance

    Getting a larger sample size usually won’t help you here, by the way. Bad selection criteria on a small sample is still bad selection criteria on a large sample which limits the population about whom you’re trying to make an inference. You THINK you’re talking about all Afghans, but you’re really only talking about those that (1) own phones or have access to phones, (2) are home when the phone rings; and (3) care enough to participate in a sample.
    (2) If you only called American homes during business hours of 8-5, who are you really sampling then? Same question for Afghans.
    (3) Why are folks answering your survey? I typically think of the average respondent as someone who really hates you or really likes you or is really bored. Those are the only reasons that *I* would answer a survey. But if you think those are truly the only reasons, then you must assume that you represent the ideal sample – which you/I don’t.

    There are lots of details to cover in a well-designed survey and more still to cover in a well-planned implementation of a survey. I trust Gallup’s methodologies and math behind their work – but probably only because I’ve bought their rhetoric.

Leave a Reply




You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>