A number of conservative sites addressed the M. Steven Fish study when it was quoted by Vox. But they didn't really look at its statistical basis. Here's the general argument that Fish makes.
According to UC-Berkeley Professor M. Steven Fish, that judging by murder rates, people in Muslim-majority countries actually tend to be significantly less violent.
Predominantly, Muslim countries average 2.4 murders per annum per 100,000 people, compared to 7.5 in non-Muslim countries. The percentage of the society that is made up of Muslims is an extraordinarily good predictor of a country’s murder rate... More Muslims, less homicide.
Now as critics have pointed out, the argument is that Islam is an ideology that justifies collective violence, not that people who happen to live in Muslim countries are more violent. Fish's argument is the equivalent of defending Communism by pointing to a low murder rate while ignoring the mass murders by the government.
M. Steven Fish's list of "peaceful" Muslim countries with low murder rates includes Saddam Hussein's Iraq, Syria, Pakistan, Iran, Turkey and Indonesia. All of these countries have engaged in genocide.
The ridiculous argument that M. Steven Fish makes is that Sudan, which committed genocide, but has an official murder rate of 0.3 is far less violent than the United States at 5.6 or Canada at 1.9.
Fish is making the argument, "more Muslims, less homicide" by comparing the ridiculously fake murder rates from totalitarian Muslim states that have murdered hundreds of thousands of people... to the real murder rates in non-genocidal Canada.
This is the nonsense that the various pieces in Vox, the Daily Beast and Slate claiming that Muslim countries are less violent than non-Muslim countries is based on.
That is how to abuse statistics. Statistics can be used to make just about any ridiculous argument. And that's what M. Steven Fish does. People with a background in statistics may have more fun with this, but let's take a look at the data Fish uses.
He compares 19 Christian and 19 Muslim countries. The list of Christian countries includes South Africa and Colombia, some of the most violent places in the world, along with Mexico, Russia, Uganda, Peru, Kenya, Argentina, and a number of European countries. So you already know part of how this cake is baked. M. Steven Fish isn't comparing non-Muslim to Muslim countries, he's comparing a global list of countries that happen to be Christian despite having little to nothing else in common. He doesn't compare Muslim countries to Buddhist countries because that would wreck his thesis.
On the Muslim side are a bunch of Muslim countries that are known for their peacefulness. Indonesia, Pakistan, Egypt, Turkey, Iran, Sudan, Saudi Arabia, Iraq, Yemen, Syria and a number of others.
And these countries have very nice and low murder rates. Especially since Fish makes sure to start from a period when Western murder rates were higher, while murder rates in places like Turkey and Egypt were lower. (Fun with Statistics 101, always make sure to select data that favors your political agenda.)
A mere 0.3 for Sudan. 0.4 for Egypt. 2.9 for Iran. 1.1 for Indonesia. 0.9 for Saudi Arabia. 0.7 for Mali. Even the worst countries, Iraq's 5.9 is barely higher than America, Pakistan's 6.9 is still lower than Ukraine.
The numbers are in, as the various explainers say based on M. Steven Fish's book, science says so. It's statistics. And statistics never lie. Except when they do.
Some countries fairly accurately document every known murder. Others barely even have functioning governments. Or governments that are not about to share real statistics that make them look bad.
Even in the United States, urban police forces have been known to fudge numbers. What's Chicago's real murder rate? Good question. It depends on classification and clearance rates. Murders can be reclassified as death by unknown causes. And voila, a lower murder rate.
Japan's low murder rate is often cited by gun control advocates. But in reality, it's kept artificially low because police choose not to classify many murders as such for cultural reasons. Instead Japan has one of the world's highest "suicide" rates. Researchers consider Japan's murder rates to be mostly worthless. And oh yes, it somehow has a conviction rate higher than most totalitarian states.
And that's Chicago and Japan. What are the statistics from Saddam's Iraq (Fish starts from 1994) or Assad's Syria or Sudan, whose leader is wanted for genocide, worth?
For a murder rate per 100,000 measure to be worth anything, the country needs comprehensive policing. There has to be a credible reason to believe that the average murder will be investigated and listed.
How accurate is Pakistan's murder rate considering that the government doesn't control sizable sections of the country? What's the murder rate in Waziristan? Or in many rural areas, particularly tribal areas, where crime isn't settled by law enforcement, but by local elders.
And of course, it's not murder if it doesn't count as murder. There are an estimated 1,000 honor killings in Pakistan of women and girls every year, most of which go unreported and are not considered murders. Because in some Muslim countries, women aren't people.
To the extent that you can get any useful murder rates in Pakistan, it's going to be from the more developed areas, particularly from major cities. If you look at Peshawar, for example, the numbers are quite high.
The homicide rate of 22.99 per 100,000 population per year is one of the highest reported in the world with only South Africa, Columbia and Estonia reporting higher rates
But if you use in the limited reported murder rates from more developed areas and then run them against Pakistan's total population, you get an artificially low per 100,000 murder rate.
What's Sudan's murder rate? According to Fish, the country which engaged in genocide, has one of the lowest murder rates in the world. If you believe M. Steven Fish, Sudan, which has more State Department advisories than hell, is safer than Japan and New Zealand.
Except, aside from the genocide, a large part of Sudan is not under the control of the central government. Or really any government. The murder rate in South Sudan is estimated at one of the highest in the world, but really who knows. The current WHO estimate for the entire country is 28.6. That's a whole lot more than 0.3. Fish can no doubt defend his statistics on technical grounds, as can the author of the study that says the moon is made of cheese, but that doesn't change the fact that they have nothing to do with reality.
When you look at Saudi Arabia, the Interpol numbers are 1.0, close to Fish's 0.9, but the WHO estimate is three times that. And it's still ridiculously low.
M. Steven Fish's premise has nothing to do with his thesis. His statistics are bad. His entire structure is based on bad faith. Using per 100,000 rates to compare the United States and Canada to Pakistan and Egypt, where the government has lost control of entire territories and police credibility is low even in major cities is worthless.
Fish is treating the statistical validity of ridiculously low crime statistics from Sudan, Saudi Arabia and Egypt as being equally legitimate to those of the United States and Canada. This is junk science.
A more accurate way to compare Christian and Muslim violence rates, if M. Steven Fish is bent on that task, would be to compare the rates of Muslims and Christians living side by side in the same country. That means comparing Muslim to Christian violence rates in the UK or in Egypt. The results would no doubt be quite interesting, which is why Fish and Vox wouldn't touch them with a ten foot pole.
We can start by looking at the quite disturbing numbers for Muslim crime in the UK.
The Home Office figure of a Muslim prison population of 11% in 2008 is the obvious place for this discussion to start.2Comparing Muslims to non-Muslims, and taking the Muslim population of the country to be approximately 4% (2.4 million out of 62 million), we calculate a disproportionality of three for the Muslim population, which is to say that three times more Muslims are in prison than we would expect given the number of Muslims in the country.
More disturbingly, it appears that this disproportionality may grow significantly if we look at high-security prisons. Four high-security prisons seem to have even larger Muslim populations than one would expect from the 11% figure. HMP Whitemoor has a Muslim population of 34%, HMP Long Lartin of 24%, HMP Full Sutton of 15%, and HMP Belmarsh of 22%. However, HMP Frankland has a Muslim population of only 3%, and there are other high-security prisons in the UK (HMP Strangeways, HMP Woodhill, HMP Wakefield, etc.) whose Muslim populations we have not been able to ascertain. Taking a weighted average of these figures to reflect different population sizes at each prison, we arrive at a figure of 18% for mid-2008
As at 31 March 2014, the latest point in time for which data is available for public use, the male prison population in England and Wales for all offenders serving immediate custodial sentence for rape was 5,682. Of this, there were 676 offenders who self-declared their religion as Muslim (12% of the total).
The data is in. More Muslims, more crime.