Democrat Lawmakers Accuse Me Of Playing Games With Statistics

Democrat state Senators George Sinner (Fargo) and Phil Murphy (Portland) have a letter to the editor in the Dickinson Press today calling me out for my rebuttal of John Oliver’s commentary on worker deaths in North Dakota.

First, let me say that I think it’s odd when people like Sinner and Murphy criticize me by writing letters to the editor of newspaper which never published the things they’re criticizing in the first place. I guess I should take it as a compliment that they see my writing as so well known in North Dakota that the readers of these newspapers will know who and what the hell they’re talking about when they reference “a North Dakota political blogger.”

I’m flattered.

Anyway, to the content of their letter, they accuse me of playing games with numbers when I pointed out that rate of oil worker deaths in the state has actually declined in the context of increased oil activity (I compared fatalities to the number of spudded wells).

Here’s their argument:

With statistics chosen to deliver that outcome, the writer denies the main point of HBO’s John Oliver’s recent rant; that North Dakota workers are dying on the job at rates much higher than any other state in our country. The most recent numbers from the Federal Bureau of Labor Statistics in 2013 prove just how dangerous North Dakotans have it: with the average worker fatality rate across the United States at 3.4 deaths per 100,000 workers, North Dakota’s rate was 17.7 deaths per 100,000, over five times the national rate and ranking us the worst stated among all 50 states in workers dying on the job.

“Oh, but that’s the oilfield. It’s more dangerous!” We’ve all heard that explanation before, right? But when we compare just the mining and oil extraction sectors, North Dakota’s job fatality rate shoots up to 104 per 100,000, far exceeding the rate of 15.9 per 100,000 in the same job sectors across the U.S. In construction, we suffer 10 times the national average with North Dakota’s fatality rate 97.4 per 100,000, compared to 9.9 on average in the same field.

Many of us might not like Oliver’s portrayal of North Dakota’s workplace fatality rates, but on worker’s injuries and deaths his facts are accurate.

Speaking of statistics chosen to deliver outcomes, there are some serious problems with the numbers Murphy and Sinner are using. They reference the Bureau of Labor Statistics, but I honestly don’t know how they got to the numbers they’re publishing from what the BLS has actually reported.

They cite the 2013 BLS numbers as the most recent data available. The preliminary 2014 report is actually available right here, but let’s stick with the 2013 numbers since they’re finalized and that’s what Sinner/Murphy used.

From that they tell us that there were 104 deaths per 100,000 workers in the mining and oil extraction category in 2013. Which, you know, is a lot. Except when you actually look at the BLS data for 2013 what we see is that there wasn’t anywhere near 104 deaths in North Dakota in that category in 2013. There were exactly 12, in a category that includes much more than just oil activity (including, say, coal mining).

What’s more, there was nowhere near 100,000 people working in the oil/gas industry in North Dakota in 2013. Again via the BLS, here’s the North Dakota trend line for the mining and logging employment category (which includes oil extraction) over the last several years. If anything, this broad category likely overstates the number of oil and gas workers in the state, but you can see it peaks out around 30,000 before falling along with oil prices in recent months.

In 2013 specifically the highest monthly employment number for this category for the entire year was 27,800. So how in the world did Sinner and Murphy get from what the BLS actually reported – which was 12 deaths in 2013 out of about 27,800 workers – to the 104 deaths out of 100,000 workers they reference in their letter?

The math certainly doesn’t work out. Even if you extrapolate the actual numbers into a per 100,000 rate you wind up with less than 50 deaths per 100,000 workers.

But such an extrapolation wouldn’t be terribly honest in that it would leave people assuming that there were 104 oil industry deaths in North Dakota in 2013 which is about 766 percent more than the actual number.

I suspect the problem is that Sinner and Murphy aren’t really using numbers from the BLS. Rather, I think they’re using a report put out by Big Labor (the AFL-CIO, specifically) which supposes numbers much more in line with Sinner and Murphy’s. But obviously the AFL-CIO isn’t an objective source. It sounds much better, if you’re trying to undermine a statistical argument made by a political blogger, when you attribute your numbers to the BLS.

However they came up with their numbers – and I honestly have no idea where those numbers are coming from – it takes some chutzpah to accuse me of playing statistical games while they play statistical games of their own.

Also, I’d like to address this:

Historically, the Bureau of Labor Statistics tracks jobsite fatalities by comparing the number of worker deaths to the number of people actually working.

This gives results that can be accurately compared across states and occupations. They don’t compare the number of worker deaths to the number of telephone poles being set, the number of houses being built or the amount of oil produced. Those would be meaningless comparisons, as is that blogger’s comparison of worker deaths to spudded oil wells.

When I used the spudded wells comparison I wasn’t using it to diminish oil worker deaths. I was using it to put lie to a report from the Center for Investigative Reporting which claimed that North Dakota’s oil boom was a “serial killer.”

I was pointing out that the reason oil worker deaths in North Dakota increased with the oil boom was because the amount of oil activity in the state increased. And I was right. When you put oil workers deaths in the context of oil industry activity (for which I used spudded wells, the most labor-intensive part of the oil development process, as an approximation of that activity) you see that the fatality rate is actually rather flat:

Now if Murphy and Sinner have a problem with that comparison, then fine, but then let’s use a metric that takes into account the fact that we have more oil workers in the state doing more oil development activity, and by that I don’t mean ideologically driven and mathematically mysterious data from Big Labor talking points.

Don’t get me wrong. Worker safety is of the utmost importance. That said, we should recognize that North Dakota’s primary industries – energy and agriculture – are inherently dangerous. That doesn’t mean we shouldn’t look at ways to make them safer. But let’s not let partisan politics inspire us to exaggeration.