Telling the truth with statistics

[Andrew Rothman]

When my local paper's columnist joined the throngs of journalists who report the (just-998,000-short-of-a-)Million Moms' allegations as if they were facts, I had to respond in my blog (and in a letter to the writer and the paper's ombudsman).

The full response is on my blog.

Some highlights:

Ask about guns? Sure, but ask about the other 99% of dangers, too
Star Tribune columnist Kim Ode exhorts her readers to Ask about guns before letting their kids play at others' houses.

We're supposed to ask other parents if there are guns in their homes. Simple, right? As if asking will keep all the dragons at bay. We ask about other things -- peanuts for allergies or helmets for biking. But guns? Yikes. The question feels accusatory, political, nosy or just plain paranoid. All of which really are only excuses to dodge the issue.

We're supposed to ask, no matter how uncomfortable it feels. The Citizens for a Safer Minnesota are trying to make this as painless as possible, but they know it takes some practice. Its annual ASK campaign -- for Asking Saves Kids -- kicks off on Tuesday, the first day of summer. It's a day of facts and figures, a day to learn that nationwide, 40 percent of homes with children have guns, to learn those guns often are left unlocked or loaded.

There are statistics about fatalities, too, but thinking about the thousands who die is actually less frightening than thinking about just one child it could happen to, especially if that one is a child you kiss goodnight.

Kim, how many thousands of kids do you think are killed in gun accidents every year? You do know that even as a columnist, you are allowed to fact-check, right?

In journalism school, this is what we call a "single source" story. Sure, it's a column, but when you only hear from one person, your view can be a bit, well, one-sided.

Anyway, asking about guns is a good idea, surely. Guns are involved in over one percent of accidental deaths of children.

Did you get that? Read it again. I'll wait.

...

A picture is worth a thousand words; this picture, I think, speaks volumes:

My conclusion? If you ask about guns, you'd better be asking about...

• seat belts and car seats
• bathtubs and five-gallon buckets
• plastic bags and old refrigerators
• lighters and matches
• cleaning chemicals and vitamins, and
• bikes, helmets and riding areas

...because each of those things results in more accidental deaths than guns.

If you ask about guns, and not the other things, I'm going to assume your concern is more political than practical.

 

[Standing Wolf]

If you ask about guns, and not the other things, I'm going to assume your concern is more political than practical.

Yep. Thanks for the chart, eh?

 

[jefnvk]

this is what we call a "single source" story

Ummm...

I only see one source from you.

Still, a good response.

 

[22-rimfire]

I don't think if the writer referenced your statistics had said something like "1.32% of deaths between Age 0-17 in 2002 were firearms related" the story would not have quite the same impact. By the way, Matt, where did you get your statistics? No reference included here, but I checked your blog and it referenced the CDC. The article is clearly an emotional appeal and not rationally fact based.

 

[Infidel]

Does the "Source: CDC" in the bottom right corner of his chart mean nothing to you?

 

[goalie]

The CDC has the statistics easily obtainable on their website.

 

[22-rimfire]

Sorrry, didn't see the chart reference.

 

[Aguila Blanca]

Great chart. It makes the point very ... graphically.

Which touches on one (only one, mind you) of my pet gripes about the media these days: Misuse of graphs.

How many times have you seen a graph (usually a vertical or horizontal bar graph) comparing two or three items, and at a glance (which is what graphics are supposed to be for) it appears that Item B is twice as big/heavy/fat/expensive as Item A or Item C.

Until you actually zero in on the legends, and realize that while Item B was ranked at 96 and Items A and C were ranked at 92, the person putting the page or article together conveniently truncated the graph so that instead of showing the bars from zero to 100, they only showed the upper tenth -- from 90 to 100. So while the graphical difference between 92 and 96 is negligable, by chopping off the first 90% or the graph they make an insignificant difference appear HUGE.

Watch out for this, and by all means call them on it when you catch it. The entire purpose of a "graph" is to convey data graphically. If you chop off 90% of the date, whether it's to fit the chart on the page or if it's to deliberately mislead the audience, you are not conveying the data at all. You are graphically falsifying the data.

 

[Andrew Rothman]

I don't think if the writer referenced your statistics had said something like "1.32% of deaths between Age 0-17 in 2002 were firearms related" the story would not have quite the same impact.

Exactly. "Gun deaths in children, though tragic, are almost negligible compared with other causes" would be a really short article.

Ummm...

I only see one source from you.

I was countering their biased secondary source with an unbiased primary one.

 

[DonNikmare]

Well done Matt!

 

[mnrivrat]

Thanks Matt ! Not only do I commend you for your activity, but thank you for the pie chart that I had not seen before - I pulled it and put it in my info files.

 

[Mannlicher]

There are lies, damn lies, and statistics.

 

[swacje41]

There are lies, damn lies, and statistics.

Sadly, that is one of those statements that is very misleading and causes a lot of misunderstandings. Statistics have no relationship to lies. Statistics are simply mathematical representations of probability or of gathered data. And much like a computer, GIGO--garbage in, garbage out. If more people knew how to accurately look at statistical findings and knew what those findings meant they would probably be better off. Properly used statistics are one of the ways we identify the lies.

 

[geekWithA.45]

Properly used statistics are one of the ways we identify the lies.

A properly constructed and correctly analysed double blind study that isolates one and only one variable sheds light on the closest thing science will ever come to Truth. Anything and Everything else is Voodoo.

Prof X, formerly of the Pentagon, who taught me the ways of research methodology and statistical analysis of human behavior.

 

[peacefuljeffrey]

I don't think if the writer referenced your statistics had said something like "1.32% of deaths between Age 0-17 in 2002 were firearms related" the story would not have quite the same impact. By the way, Matt, where did you get your statistics? No reference included here, but I checked your blog and it referenced the CDC. The article is clearly an emotional appeal and not rationally fact based.

Well, it is an invalid presupposition that the article was ever -- from its conception to its execution -- worthy of having an impact.

I think that zero people were killed by rail-guns last year. That doesn't mean that we need to have someone write an article about the dangers of rail-guns and then drum up some "impactful"-sounding article about a non-issue.

If telling the truth about this gun-accident issue doesn't have much of an impact, then the issue isn't DESTINED to have an impact. If a columnist wants to write a story with impact, that columnist should find a story that naturally HAS one, without b.s. embellishments and hyperbole.

-Jeffrey

 

[rhubarb]

When it comes to the debate about whether or not guns should be legal, which guns should be legal, who should have guns etc., I have become bored. I'm no stastistician, but I bet that 95% of people are not gonna change their mind based on a gun grabber's or a gun nut's statistic supporting a position. The statistics are irrelevant to me.

If an individual commits a crime with a gun, put him in jail for the crime, not for having a gun. If parents don't take care of their children and a child is hurt or killed with a gun, prosecute the parent for neglect, not for having a gun. There are laws against any mayhem or negligence that people do with guns that are broad enough that any crime done with a gun is prosecutable without infringing on our rights.

The rule of law from the beginning of our country listed the right to keep and bear arms as a right. For me, it is about our natural rights as enumerated and guaranteed by the Founding Fathers, not about whether more crimes were committed by guns than pickaxes this year. When the citizens, legislators, and courts of the United States of America realize that ever-changing statistics are not justification for picking apart the Constitution and Bill of Rights, then we will be on the right track.

 

[dk-corriveau]

Matt,

You are right on with your point. More of us need to call out the media when they portray guns as the most evil and threatening man made creation of all time.

That being said I have one question regarding your data. We are looking at 1.32% of what number?

I ask because I believe that we, we being the Pro Second Amendment crowd, must be explicitly clear about the data and not fall into the same tactics that the antis use. In other words, we need to say that Guns account for 1.32% or X,XXX deaths (or whatever that number is) in 2002 and poison for accounted for 2.94% or X,XXX deaths (or whatever this number is). As a community that assumes a higher responsibility by exercising our Rights as citizens, we should, in my opinion, assume a higher responsibility with regard to making our arguments.

Thank you again for calling out your local reporter!

 

[taliv]

I was countering their biased secondary source with an unbiased primary one.

uhh, if you're referring to the CDC, think again. they're as biased as they come.


also you've succumbed to two very misleading evil-gun-grabber hidden assumptions.

a) you've equated "accidental deaths" with a chart on "injury deaths" that may include for example, drive-by shootings or something as opposed to true ADs.
b) note that the chart includes data all the way up to 17 yrs old. when you "THINK OF THE CHILDREN" are you thinking of 17 yr olds? sure we may think of them as kids... but would you talk to a 17 yr old neighbor's parents "before letting their kids play at others' houses."

the nut is, if you cut that down to like, 0-13 yr olds, and exclude gangs and other criminal elements, the percentages is a helluva lot lower than 1.3%

 

[Brat7748]

Keep in mind that the CDC stats include, I think, suicide and gang bangers killing each other. True accidental rate is even lower

 

[mnrivrat]

rubarb, a very valid point ! What statisics are being use in arguments for is to counter emotional and biased "feelings" . I suspect you are correct that they do little to change how people feel about the subject and the fact that statisics are being generated with so much bias seems to bear that out.

Frankly if you showed me a statistic that was valid and indicated that 9 out of every 10 deaths of children in the US were due to being shot, I would not give up my right to protect myself. I would simply say that something needs to be done to prevent this from happining ,and trying to remove all firearms from the population is NOT it. Probition of firearms will never work to reduce crime , and will have insignificant effect on non-criminal deaths.

dk - corriveau ,

I think the heading on the pie chart is pretty clear - the actual numbers won't mean any more than the % of total population IMO . I'd say we have not so much of a higher responsibility , but a responsibility to present non-biased data in order to establish fact and not fiction. Find studies that are first and foremost statisticaly sound and accurate based on good science and not pre-determined BS .

With that said , I think we have already determined that it makes little difference what the numbers say - they realy have very little to do with guns and at best speak more toward whether are society needs to address crime in a different way. Death by firearm is NOT the fault of the ability to have guns . The deaths are caused by carelessness, intentional suicide, criminal acts, or justifiable self defense.

If numbers are important it is noteable only that there are way more deaths associated with other activities in our society, and that those who concentrate on deaths where firearms were involved are not so concerned about the welfare of the children . If they were, they would expend their energy in the area's where it would do the most good. Not on the bottom rung of the ladder so to speak.

 

[Berek]

My favorite stat was in an ad I found in the 13th Annual Edition of Hunter's Handbook magazine. On page 13, it says "Lookout! Low Bird! Last year, 30 people per 100,000 were injured while playing badminton. Compare that to hunting. Hunting related firearm injuries are at an all-time low at 6 per 100,000 participants."

One thing that the anti's do not take into consideration is that, should I sprain my ankle on the way to my tree stand, it becomes hunting related as I was in the process of hunting. Meanwhile, the injury had nothing to do with the firearm I was carrying nor anyone else in the woods at the time.

Sub RANT(Anti AS String, Me AS Byte)
Dim Anti
Clueless = Anti * Anti
Firearms != HuntingAccidents
End Sub

Let's get real. Not all hunting related injuries are caused by firearms.

Berek

 

[R.H. Lee]

I know you mean well, and I suppose it is necessary to counter lies with the truth, BUT- why is it necessary to justify our RKBA to anyone? It’s as though it’s open to negotiation if only they can show that private firearms ownership is somehow ‘a public safety concern’. It’s tiresome constantly beating them back. But I suppose it’s better than just shooting them.

 

[Andrew Rothman]

I have one question regarding your data. We are looking at 1.32% of what number?

Accidental injury deaths -- the sort that asking about guns is supposed to prevent. Read the blog entry for the whole story.

uhh, if you're referring to the CDC, think again. they're as biased as they come.

Really? Tell me how!

also you've succumbed to two very misleading evil-gun-grabber hidden assumptions.

a) you've equated "accidental deaths" with a chart on "injury deaths" that may include for example, drive-by shootings or something as opposed to true ADs.

Nope. I made the chart. It includes only accidental deaths (the label could have been clearer, I grant). In that group, accidental gun injury deaths were 115 of the 9000-or-so accidental deaths, or about 1.32%.

b) note that the chart includes data all the way up to 17 yrs old. when you "THINK OF THE CHILDREN" are you thinking of 17 yr olds? sure we may think of them as kids... but would you talk to a 17 yr old neighbor's parents "before letting their kids play at others' houses."

Sure. Sixteen- and 17-year-olds together account for more than half. But even giving them the benefit of the doubt by including the legal definition of children, the number is miniscule. To include only up to 13 or 14 would invite further criticism.

Keep in mind that the CDC stats include, I think, suicide and gang bangers killing each other. True accidental rate is even lower

Again, no. The CDC lets you slice and dice the data almost any way you'd like, and a simple spreadsheet program like Excel lets you derive the rest.

While I have no doubt that some of the "accidents" were actually suicides, I'm not going to fight that fight without data.

I know you mean well, and I suppose it is necessary to counter lies with the truth, BUT- why is it necessary to justify our RKBA to anyone? It’s as though it’s open to negotiation if only they can show that private firearms ownership is somehow ‘a public safety concern’.

Well, I often make the point that freedom can be dangerous. I just love pointing out that, when it comes to "askng about guns," our freedom is pretty danged safe.

Thanks for reading, everyone.

 

[BigG]

Good job, Matt! It's funny though, I've often had the suspicion that "truth" and "news" are mutually exclusive topics...

 

[dk-corriveau]

mnrivrat,

I agree that the % tells the story, which is that guns are a small portion of the injury deaths for 0-17 year olds in this country. All I am saying is that to complete an argument the total number should be included along with the percentage in my opinion. And this is true for this issue or any other argument where one is trying to use data to make a point.

For example, if someone who worked for me came to me and said: “Boss I increased our profit margin from 10% to 15% this year, I would like a promotion.” My response would be: “Great, but how much profit, in DOLLARS not % margin, did we make this year.”

Likewise, if someone said that the percentage of deaths due to guns has decreased over the past five years from say 2.5% to 1.32% (made up percentages), I would want to know if that is because the total number of deaths has increased and the number due to firearms has held constant or has there been a real decrease in the number of firearm deaths.

Percentages are enlightening and important for comparison, but often numbers are needed as well to tell the whole story. In this case, I would like to find out what that number is, so that when an anti says: “What about the tens of thousands of children that die from guns every year?” I can respond with the actual number and the relative percentage for the other causes.