Little Green Footballs

Friday, April 04, 2008

Abusing statistics, one at a time

In another perpetually recurring incident, Charles has managed to twist a survey around to promote his own anti-Arab agenda: (To avoid redirects: http://littlegreenfootballs.com/weblog/?entry=29494)

Arab Survey: 55% Say Offensive Words Justify Violence

A survey of Arabs conducted by YouGov reveals a strange mix of deep irrationality, denial, and self-contradicting views on the subject of .
A YOUGOV poll commissioned by the Doha Debates has concluded that nearly one-third of all Arabs believe that Saudi Arabia is at greater risk from religious extremism than any other country in the world.

The poll adds weight to the vote at a session of the Doha Debates held on March 3 in Doha, where the motion “This house believes that Muslims are failing to combat extremism”, was carried by more than 70% of the audience.

In the YouGov survey, nearly half of all Arabs in the Gulf, Levant and North Africa said they have met someone who holds extreme religious views.

Asked under what conditions violence is permissible, more than 60% cited Western interference in a Muslim country, while 55% said offensive words or behaviour was a trigger.

More than half the respondents also believed that poor religious leadership is to blame for today’s extremism – although seven out of 10 said the size of the problem had been exaggerated.
===

Of course it's self-contradictory, assuming the brilliant and non-existent analysis of CJ was correct. Take a close look at the phrase in question:
Asked under what conditions violence is permissible, more than 60% cited Western interference in a Muslim country, while 55% said offensive words or behaviour was a trigger.
Any reasonable person will agree that saying "Greed triggers theft" is not the same as saying "Greed justifies theft". Does this matter to the lizard king? Apparently not very much. Just as little as the torrent of anti-Arab and anti-Muslim comments that ensued on the thread shortly afterwards..

9 comments:

Deuce Geary said...

No, greed doesn't justify theft . . . except to the thief.

Anonymous said...

Semantics, semantics, who cares so long as it can be used to make me feel better about hating people?

Anonymous said...

Sphinx, you're pushing way too hard on that word "trigger." You need to keep in mind that journalists commonly will vary words simply to avoid repetition without giving too much thought to the full spectrum of nuances.

The evidence is in the clause which introduces that sentence. The question was when violence was "permissible." "Trigger" is meant here as an ellipsis for "a trigger that brings on permissible / justifiable violence."

You don't just have to take my word for it. Look at how this journalist phrased her report on this very same poll:

The preferred "action" against religious extremism is highlighting the cause in the media. However 6 in 10 believe violence is legitimate if a Western country interfered in a Muslim country. A further 55 per cent would deem it appropriate in the instance of offensive words or behaviour against Islam.

Source: http://www.gulfnews.com/News/Gulf/qatar/10202241.html

Respondents here indicated that BOTH circumstances were legitimate causes for violence. In other words, while there are many Arabs who recognize the danger of religious extremism, nearly half of those polled still see what they regard as "blasphemy" as sufficient cause for violent retaliation.

Rick

The Sphinx said...

Ok, well that's a point.

But if you look at the numbers of polled people, it says in the article CJ has quoted that the survey was given to 940 people. In the one you gave me it mentioned 960, and said something about an online panel of 102'000 respondents.

The first two figures make out roughly 0.000029%, the second one about 0.031% of the 325 million people living in the Arab countries in North Africa and the Middle East. Far below even 1% of the people. Needless to say I'm still a bit wary about the expressiveness of this result.

Chuckles doesn't give a damn of course, and happily concludes that it applies for "all Arabs". I think if you conducted the poll on the same amount of Arab expatriates in America for example, the figures would look entirely different.

stvip said...

Augh,
Mr. Sphinxter (yes, I realize insisting on using this moniker when addressing you is childish on a CJ magnitude, but I'll live), what is it that you study abroad?

"Far below even 1% of the people"?

It's about mean, variance - standard deviations. The concept of statistical significance is of paramount importance in any academic field, I suggest you familiarize yourself with it.

Polls like this can be misleading for a different reason, though. The results are very much dependent upon the particular phrasing of the question, and thus there's plenty of leeway for the researchers to get whatever results they want. If you took the trouble to make a blog post about this, you might as well have done it properly and read the research itself to see if there's actually a valid critique of how it was presented in the media.

In order words: once again, LGF sucks. LGFWatch more so.
(though at least for allowing open discussion and criticism, LGFW gets full marks, LGF zero)

The Sphinx said...

" 'Far below even 1% of the people'?

It's about mean, variance - standard deviations. The concept of statistical significance is of paramount importance in any academic field, I suggest you familiarize yourself with it."


Of course I am familiar with all this, but conducting a poll on such a meager percentage on the population, and then concluding that ALL Arabs (as mentioned twice in the excerpt here) yield the same result is plain inaccurate at best, dishonest at worst. Of course it not possible to ask all Arabs about their opinion, but it's the way and terminology how the results are presented which makes me very wary. Needless to say, CJ has found his four-leaf clover and made use of it for his agenda, namely as anonymous #1 said quite nicely: Making himself feel better about hating others.

"Polls like this can be misleading for a different reason, though. The results are very much dependent upon the particular phrasing of the question, and thus there's plenty of leeway for the researchers to get whatever results they want. If you took the trouble to make a blog post about this, you might as well have done it properly and read the research itself to see if there's actually a valid critique of how it was presented in the media."

Agreed. I'm really interested in what questions were asked, and how the results were evaluated based on whichever answers were given. Here lies great potential for manipulation, regardless of what the poll is about.

"In order words: once again, LGF sucks. LGFWatch more so.
(though at least for allowing open discussion and criticism, LGFW gets full marks, LGF zero)"


We do what we can.

stvip said...

"Of course I am familiar with all this, but conducting a poll on such a meager percentage on the population, and then concluding that ALL Arabs (as mentioned twice in the excerpt here) yield the same result is plain inaccurate at best, dishonest at worst."

I don't get it. You say you understand statistical analysis, then repeat this nonsense.

By analogy, consider medical trials: you take, say, 300 people, and from the results of some medical intervention (a new tested drug, for example) on them infer things which apply to the entirety of humanity (which amounts to a much lower percentage than that in the original post). That's standard procedure, and is the entire purpose of polls. To ensure reliability of such inference, there are statistical/probabilistic tools to give a gauge on possible error. There's absolutely nothing unusual about this.

Now, far be it that I assert one should accept anything with p<0.05 as truth. On the contrary, there are a great many flaws in a study which could invalidate its results despite achieving statistical significance. Especially in Social "Science" studies like these.
For example: proper randomization. Who was polled? Were random people on the street presented with question? If so, people with no strong political opinion could ignore the pollster, while the extremists are much more likely to seize the opportunity to express their views. etc. I don't accept results from such studies until I have reviewed the methodology.

All this is besides the point that I find the rage somewhat quizzical. Charles may be a muslimophobe bigot, but it isn't bigotry to point out the very real problems plaguing the Arab world. For the most part, Arab countries are primitive, second-world places, and most definitely not for lack of resources. I'm not an anti-Arab person, nothing would delight me more than for the Arab world to reform and develop itself. But there's no point in denying these problems (which really are often conveniently spuriously blamed on the West, Jews and Israel by Arab country citizens; neither is pointing this fact out bigotry).

The Sphinx said...

Look, I don't know either how they did this, which is one reason why I don't really trust it very much, and the other reason is my impression I'm getting out of

But the impression I got from how this was done: You have 10'000 people. You ask 4 of them whether they like sushi, and when two say they don't, you say: Roughly half of the 10'000 people don't like sushi, with a confidence interval of blah.
Extrapolating a sample of people questioned to approximate the real figure is inevitable, you don't need to tell me. But seriously 940 or 960 people is extremely little to say enough about the whole Arab population. If they said: "xyz% of the people surveyed", it would be a different story than saying "xyz% of all Arabs".

"But there's no point in denying these problems (which really are often conveniently spuriously blamed on the West, Jews and Israel by Arab country citizens; neither is pointing this fact out bigotry)."

You need to live in an Arab society long enough to see that this simply isn't true.

Just yesterday people were planning strikes and protests as an action against the rising prices on the most basic necessities, sinking wages, corruption and basically all the miserable conditions in Egypt. The government - in its good sport - scattered police all over the place and arrested people en mass.

Nobody thought of bringing Israel or the West into the game.

Egyptians have a consensus of criticism of our educational system and how teachers are badly paid, state schools much too overcrowded, and a lack of good resources.

Nobody blames anyone but our own corrupt government.

The blame for everything ranging from a relatively poor infrastructure of the country, lack of real organization and a growing gap between the rich and the poor is put right where it belongs.
And social problems.. Well, they're blamed straightforwardly on society, but not on any ethereal enemy that has nothing to do with it.

Virtually every time I socialize with people from the Arab world, the longer the conversation gets, the more certain it is that it'll evolve into lamenting over how bad some things are and how they could and should be better. I don't remember the last time America or Israel was part of the discussion, in terms of being a cause for those problems.

To hear what Egyptians (or Arabs in general) really complain about, you just have to hop into a few taxis in Cairo and chat a bit with the drivers. It'll open your eyes to a world you'll never see online.

Ok, I'm rambling too much. To cut it short: It's not bigotry to point out real problems, but it's a violation of the most fundamental principles of journalism when your spin on the matter shows an incomplete, inaccurate, or even false picture of reality, e.g. by leaving out important facts for your own convenience. And until now, the Arab world from Charles and his minions' eyes has just about nothing in common with the world I have lived in.

Keith said...

CJ is, no doubt, an asshat, yet the statement is clear: 55% of respondents apparently feel that offensive statements are justification for violence. It's not so long ago in this country where a similar result might have been found.