Hacker News Reader

I don't know the Freakonomics methodology, but I'm pretty sure the analysis you're making isn't useful.

We're talking about choices an individual homeowner can make.

Those choices are "install a pool" or "keep a gun in the house".

To understand the risks for those two choices, total drownings and total gun deaths of children in the country aren't relevant to an individual's choice.

To get a useful statistic, you need to know how many children died in their home pool in the sample group of homeowners w/ pools, and how many children died from a gun in the home in the sample group of homeowners w/ guns in the house.

Because that's the choice -- do I put myself in this sample group or not?

Think about powered trampolines -- suppose there are only five installed in the country, and on two of them a child has died this past year. That's a HORRIBLE risk, but with your method of assessment, it doesn't even show up.

If you're thinking of installing this kind of trampoline, I'd highly recommend my method of analysis over yours.

This makes little sense. Say only 7 people died from inhaling cyanide last year, does that make cyanide 100 less dangerous than guns? Nope, because 100% of the people who inhaled it died.

The real questions is, of the people who are within harms way of home swimming pools, how many suffer injuries, and of the people who are within harms way of fire arms, how many suffer injuries.

Both are not that easy to measure, especially firearms, but looking at absolute causes of death is clearly nonsense.

>In 2008 there were 376 deaths of "children" (aged 14 and below) by gun. In the same year there was 791 deaths by drowning (all sources).

You should really be using the number who were killed by a gun kept in their home for this comparison. Also, there are other ways to die in a pool aside from drowning.

Even if what you suggest made sense it would still not come close to 100x more dangerous. Even if we were REALLY generous and called it 3x more, we're still not even in the same universe.

Plus as I said, the drowning statistics above are a huge over-estimation as they don't give individual breakdowns of the context of the drowning (e.g. pools, boats, bathtub, etc).

What he said does make sense. Your numbers are incomplete. You're saying "All firearm deaths" which include drive-by's, hunting accidents, etc. If you can't break down the numbers to ONLY those children who died from a gun that was stored in their own house and children who died in a pool in their own back yard, then you aren't making valid assessments of the multiple at all.

It's simple math.

That's exactly what he's saying though - that the Freakonomics article does NOT break the numbers down sufficiently, for either Drownings nor Gun Deaths. As such the 100x figure they came up with is bogus.

And yes, the Maths is simple. But gathering the right figures to plug into the Maths is hard.

> What the Freakonomics guys did was take the above numbers and then use the total number of guns and total number of pools, and then abuse those statistics in order to try and get this 100x more dangerous figure.

I don't understand. This seems sensible to me. Why wouldn't we include this figure to calculate risk? Isn't risk a matter of incidences/exposures?

I agree, it doesn't make sense to compare absolute numbers. The risk of having a gun at home should be calculated as [deaths of children by gun]/[number of family homes with guns], and the same should be done to calculate the risk of having a pool at home.

For instance, if in 2008 376 children died by gun at home and as much as 376000 families had a gun at home, then the risk would be 0.1%. But if the number of families with a gun at home were 752, then this would mean a risk of 50%!

> I agree, it doesn't make sense to compare absolute numbers.

It makes complete sense. We take number of children (i.e. population) and we take the number of times something bad occurs to that population, and that gives us our level of risk per each individual (e.g. 1 in 100,000 or similar).

If we follow the logic of your argument then we start getting into stupid territory very quickly indeed. For example, how many bees are there in the US? And how many people die of bee stings? Well given the number of bees Vs. bee sting deaths we can calculate that you have more chance of winning the lottery two times than dying of a bee string, right? ... Well no, that makes no sense at all.

Basically gun advocates are abusing statistics. They're arguing that keeping two guns in your gun safe instead of one makes guns twice as safe! And four guns is twice as safe again! So as they increase guns the guns get safer and safer, isn't that wonderful...

> It makes complete sense. We take number of children and we take the number of times something bad occurs to that population

So what you're saying is [deaths of children by gun]/[number of children]. That's not an absolute number, 1 in 100000 is just a percentage disguised (0.01%).

My only remark on your comment is that only children living in houses that have a gun should be taken into account when calculating the risk of "having your child killed by a gun that you keep in your house".

Except we're not talking about the likelihood of a child being killed by a gun v. drowning in a pool. We're talking about the likelihood of a child being killed by their guardian's gun v. drowning in their guardian's pool.

In addition, if I'm never exposed to a gun or a pool, then my risk of dying by either is zero. That doesn't mean those things have no risk. It means that my personal risk of dying from those things is zero and I shouldn't be included in any calculation that measures the risk of owning those things.

Zero people in my town were attacked by lions last year, but I'm sure a few were attacked by dogs. Cool! Lions are way safer than dogs!

UnoriginalGuy, I'd suggest taking a step back from your argument for a minute. Every single response here has explained the flaw in your original claim. It might be worth taking a minute to understand where you could have gone wrong in your reasoning.

We may be talking about that, and that would be the figure to find, but that's not the figure the Freakonomics article used. That's the point.

Here's the quote from Freakonomics:

"In a given year, there is one drowning of a child for every 11,000 residential pools in the United States. Meanwhile, there is 1 child killed by a gun for every 1 million-plus guns"

And then the author draws conclusions based on that. It says nothing about "drowning in a pool" - only drowning. It says nothing about death by a gun at the gun owner's house. It says nothing about households with multiple guns (nor households with multiple pools, I guess).

The point UnoriginalGuy was trying to make is that this is absolutely shoddy statistics at work. Now his statistics may be equally as shoddy, but he's not selling books about this stuff.

How many people own multiple pools?

Also how many people are using each pool (and indeed using each gun)?

You're not taking into account the difference in risks between children who live in a home that has a pool and those that don't.

Perhaps a valid point that Freakonomics had bad methodology in this instance, though this is similariy bad:

> So drowning is is roughly 2x as dangerous as guns

Drowning kills 100% of people who drown, so it's certainly more dangerous than guns.

Replace drowning with "pools" and there might be something more reasonable to compare, though hard to know what should be compared.

> Drowning kills 100% of people who drown, so it's certainly more dangerous than guns.

I heard from a person that had to go through helicopter rescue training that they're, in their course of training, required to drown at one point; and then, after drowning, are brought back up and revived.

Drowning may kill a large percentage of the people that drown, but it's not a guaranteed death-sentence.

>Drowning kills 100% of people who drown, so it's certainly more dangerous than guns. //

Being shot to death kills 100% of people who are shot to death. I don't think you have a valid point.

The number of deaths is a poor measure of harm. One reason is that trauma care keeps improving. So there are now a large number of amputees and brain-injured soldiers that would have previously died.

Cars are getting more survivable, and almost certainly safer, too, but do numerical comparisons with numbers from the past remain valid in light of external influences on these outcomes?

In this case, improvements in trauma care may mask both brain injury from lack of oxygen and serious disability from gun accidents in this comparison. Or we don't know the methodology and they could be counting ER visits, which might provide a better comparison.