How NOT to conduct a poll: Korean anti-peace activists in Iraq

Although I am currently an English instructor, my education was in political science. I've studied statistics and polling methodology. I've helped conduct polls several times. I'm not an expert but I know enough to spot a crappy poll in half-a-heartbeat.

The "poll" I read about in this article of the Korea Times is pretty much a textbook example of how not to conduct a poll. First the results:

According to the survey, 113 of the 131 Iraqis questioned, or 86.3 percent, said they didn’t want South Korean combat troops to be sent.

Of those against the dispatch, 81 people, or 62.5 percent of the total, said non-combat troops such as military engineers and medics would not be helpful to Iraq.

Here we see problem number one; sample size. In polling, bigger is better. I don't have my polling 101 text near me but I would guess that you would need a sample size of about 700-800 for a nation the size and diversity of Iraq. Let's see what else we have:

Lim Kyong-ran and Bae Sang-hyun, two anti-war activists who conducted the poll, interviewed Iraqis whilst in the nation, claiming the results carry significant weight.

"The survey isn’t perfect as it was conducted on only 131 people, but we think we correctly represented the atmosphere by conducting direct one-on-one interviews," a representative from Corea Peace Solidarity said in a news conference.

Now we have the ugly specter of non-statistical error. We have two guys who probably have no training in taking polls and who are clearly biased conducting personal interviews. These guys when to Iraq looking for people who were against Korea sending troops to Iraq and that's exactly what they found. Is there any other problem? Of course there is:

The survey was conducted on 114 men and 17 women in Baghdad, Mosul, Kirkuk, Basra, Fallujah and Hadija.

Here we have the old convenience survey. I can already see what happened here. Basically Lim and Bae went to Iraq, hooked up with some anti-American buddies, had those buddies introduce them to other anti-American buddies and called it a poll. That isn't even a sample of the population, much less a random sample.

So let's add this up. We have:
-too small a sample size
-untrained interviewers
-biased interviewers
-convenience survey
-not a random sample
In short, this "poll" is pure crap.

I would just laugh this garbage off except for two things:

1. Some people, especially some Korean "netizens" take this kind of garbage seriously. Some of our less informed brothers and sisters might not notice the quote marks in the headline of the article (‘Iraqis Oppose ROK Troop Dispatch’) and assume that it is a statement of fact rather than the opinion of two guys who already made up their minds before they went to Iraq. Heck, give me a couple of weeks and I can give you a survey of 131 people in which 90% of them think Roh Moo-hyun should be run over by a tank.

(BTW, using that headline without an attribution is sloppy journalism. Yeah, Seo Soo-min, I'm calling you on it.)

2. This is the kind of mess that makes many people think that all polls are useless. A properly conducted poll can be a useful and accurate tool to study public opinion. Bozos like these are a bane upon serious social scientists and they are doomed to rot in the Hell-of-Being-Asked-the-Same-Question-Over-and-Over-Again.

OK, I feel better now.