Friday, April 25, 2003

Immigration and Muslims

I continue to read the news with growing concern. There are stories about hate crimes committed against Muslims, anti-Muslim abuse and harassment on campuses and off, and the defacing of mosques. But the thing that bothers me the most is the requirement that all Muslim immigrants are required to register with the Justice Department in a process that includes photographing, fingerprinting, and “interviews” with officials on everything from their movements since arriving in the U.S. to their bank accounts.

I’d ask when we became a nation so terrified of itself that we required the registration of 130,000 Muslims in hope of catching a few undesirables in the process if I didn’t already know the answer.

Since 9/11, we have listened to our religious leaders railing against Islam from the pulpits, the streets, and even from the Pentagon. The raw hypocrisy of these so-called Christians is nauseating. Didn’t Christ say to love your neighbors? Didn’t Christ say to turn the other cheek? Christ was not a warrior, and he would have loved terrorists as he would have any other sinner. Religious leaders like Franklin Graham give Christianity a bad name when they reject the tenets of their own faith and the teachings of a man they believe to be the son of God.

Since 9/11, our political leaders have also assaulted Islam. Bush himself initially referred to the War on Terror as a “crusade,” the worst possible word he could have used when describing our actions anywhere in the Muslim world. The invitation of the afore-mentioned Franklin Graham to the Pentagon for Good Friday shows remarkable insensitivity to the many Muslims in the military. And now we have everybody’s favorite Attorney General, John Ashcroft, with support from the rest of the Administration and the Congress, demanding that we pull aside 130,000 Muslims and treat them like criminals just because some of them may actually hate the United States.

Maybe we should rename the Justice Department the “Injustice Department.”

Seriously, where is the justice in demanding that a group of immigrants, determined by their ethnicity and religion, submit to fingerprinting, photographing, and interrogation by the US government? If we were performing these detailed checks on all immigrants, I might not complain. But what we’re doing smacks of a level of paranoia not seen since the Japanese detentions of World War II. The sectarian conflict between Catholic and Protestant in Northern Ireland has been going on for decades, yet we don’t require Irish immigrants to register with the Justice Department. Hindu nationalists slaughter Muslims in Kashmir and other Indian states and we don’t require every Indian Hindu immigrant to register. And we don’t require every Hispanic immigrant from Central or South America to register with the Justice Department just because Colombian terrorists have been killing each other for years. So why do we require Muslims to register?

Oh, that’s right. None of the other religions have perpetrated terrorism on US soil. But wait a minute, what about Timothy McVeigh? Wasn’t he Christian (he claimed to believe in the basic tenets of Catholicism even if he didn’t practice much)? I guess we should demand that all Christian immigrants register with the Justice Department too. Well, since everyone living in the United States who is not Native American is a son or daughter of an immigrant, I guess that means we should all go to the Justice Department to get immigrant photographs, fingerprinting, and answer questions as to our employment, prior residences, and citizenship. “Yes, sir, I was born in California, but my ancestors came to the US before the Revolutionary War from England. Yes, that’s right, the shoe bomber came from England too. And yes, there’s a strong Muslim presence in England as well.” Guilt by association is a wonderful thing.

What’s even scarier is what’s coming down the pike from the Justice Department. The sequel legislation to the Patriot Act has been nicknamed the Patriot Act II, and it has some seriously terrifying potential. What makes me more than a little nervous is the idea that the Attorney General will have the power to revoke the citizenship of anyone who is shown to have ties to terrorists. If you’re unlucky enough to have rented a room to a terrorist and Ashcroft comes a-knocking, it won’t matter if you were born in Texas, are a card-carrying member of the GOP and the NRA, and are a son or daughter of the American Revolution - your United States citizenship could still be revoked. And once you’re not a citizen anymore, you can be deported or indefinitely detained without any rights like due process, legal representation, and a trial by a jury of your peers.

Luckily, I have some faith that the Senate and/or the Supreme Court will look at this piece of the legislation and say “Hell no!” But just the idea that all our President’s men support such anti-American drivel as stripping the citizenship off of naturalized and US-born citizens should make your skin crawl.

I’ve worked closely with Muslims, and I grew up with friends who are Muslims. As such, I cannot and will not allow an anti-American President to monster-mash through the Muslim population of this country. We’re better than that. Well, at least we used to be. These days, I’m starting to wonder. With our politicians and religious figures tearing into Muslims and Islam as wolves into a wounded bison calf, our national and personal hypocrisy is starting to show. If we don’t stand up to our government and say that this has to stop, we should just go drape the Statue of Liberty in a sheet and re-chisel the tablet to say “Muslims not Welcome Here.”

Wednesday, April 02, 2003

Lies, Damn Lies, and Polls

There’s an old maxim by Benjamin Disraeli that I like – “There are three kinds of lies - lies, damn lies, and statistics.” This has always been the case, as a skilled statistician can phrase his questions to get any answer he desires. Therefore, poll results should always be taken with a reasonably large grain of salt. But recently, I’ve been wondering just how the statisticians have managed to get a 70%+ approval rating of Gulf War 2.0.

In essence, I’ve met some people who support the war. But the vast majority of people I know support only the soldiers, not the President’s decision to go to war. I interpret this opinion as lack of support for the war. Of the people I’ve asked, there isn’t anywhere near 70% approval. I’d say that approval of the war was closer to 20% among my friends and family, but they are most definitely a self-selected group – most of them are reasonably liberal like myself.

Now, I don’t pretend to know how to design a good, scientific survey. My expertise is engineering, not polls and statistics, but I do know a bit about designing a good experiment, and a poll is a type of experiment. So I’d sure like to know the methodology behind the polls. For example, are the identical questions being asked from one poll to the next? Are the same people being asked the questions repeatedly? And the big one, how are the pollsters guaranteeing that they aren’t self-selecting their respondents?

In order to guarantee repeatable, comparable results, you must design any experiment such that the number of variables being tested is small and that each variable is controllable. In the case of a poll, if you don’t ask the same question from one poll to the next, you can’t compare the results. This is the case whether you’re comparing different parts of the country or the overall change over time. Asking “Do you support the war?” is not even remotely comparable to “Do you support the troops?” or “Do you support the President?” In addition, you might have people answer the questions they thought you were asking rather than the actual question being asked. Answering that you support the war because you support the soldiers fighting it is very different from actually supporting the war itself.

Also, if you change the variable of who is answering the question, then any subsequent polls may not have results that can be compared to the previous poll. Initially, you’re polling a group of usually randomly selected individuals in order to get a large but still incomplete cross-section of the citizenry. In the next instance, if you choose a different group of individuals, you will get a slightly different cross-section simply because your initial cross-section could not be complete. If you’re dealing with atoms in a solution, the difference may not be significant, but each person polled is an individual with unique experiences, and their different experiences make them inherently a different cross-section of society. Therefore, there’s a pretty good chance that the poll will cover a statistically significantly different cross-section of the citizenry with each new randomly selected group of people. So, if you’re trying to determine the change in a response to a question over time, you’d best ask the same group of people each time.

Finally, I’m very concerned about self-selection in polls. Most polls are taken via phone, and there’s a whole lot of people who simply won’t answer polls over the telephone. This fact alone means that the poll is selecting out an entire segment of the population, namely people who won’t answer questions over the phone. In addition, if you poll a segment of the population randomly, it’s difficult to guarantee a valid distribution. There is a very real possibility that a random phone poll will poll all very conservative areas of the country, and thereby produce very skewed results. After all, polling people about support for the war in Boulder or San Francisco is very likely to produce radically different results than a poll taken in San Diego, Houston, and Orlando.

Experiments can be designed to test multiple variables simultaneously. However, experiments designed in such a way cannot be random. Similarly, I suspect a poll can be designed to test multiple variables simultaneously, such as changes in support for the war from region to region, from Green to Democrat to Republican to Libertarian, and how support changes over time. In order to get such a poll to work correctly, the poll must be large and each individual within the polled group must meet specific criteria from person to person. I don’t see how this kind of a poll could possibly be performed on a randomly selected group of individuals – the criteria alone requires that the poll respondents not be chosen at random.

I don’t know exactly how polls are designed, but I can’t imagine that a group of 2000-3000 randomly selected people could possibly be used to determine the approval rating of a President or his policies, as the pollsters want us to believe. I’m cynical enough to wonder if the President and the military are fudging the numbers with the help of the media. But I’m also optimistic enough to think that either the pollsters just don’t understand that their polls aren’t valid, or that there might be a fundamental problem with my own logic. But so long as the treatment of a poll as a scientific experiment is valid, I don’t see how we can ever look at small polls as representing the whole of national opinion. We just don’t get enough information to separate the lies and the damn lies from the statistics.