By Douglas Valerio
As Aaron Levenshtein noted, “Statistics are like a bikini. What they reveal is interesting. But what they hide is vital.” As a mechanical engineer by degree, I have taken my fair share of math courses. Statistics are easy to calculate, harder to research, and frequently difficult to interpret. With guns as prevalent in our society as they are, there is plenty of data on many aspects relating to them, some more reliable than others. While I may not be a statistician or a college professor in the field, the basic concepts seem to escape the normal person to a point that is almost painful to witness. We can break this into 3 sections, correlation, causation, and how correlation does not equal causation . . .
Correlation is defined as a mutual relationship or connection between two or more things. All that’s required for something to correlate is that they relate to each other. One must move in relation to the other, whether it’s up, down, left, or right. Take, for instance, the fact that ice cream sales increase in hotter months. This makes sense, seeing as the hotter weather makes people want a cold treat like ice cream. Per businessinsider.com, ice cream sales top out in August and drop in December.
In 2015, the homicide rate in Chicago per the Chicago Tribune, peaked in August and September. Taking a look at both sets of data, I can say that ice cream sales correlate to the homicide rate in Chicago. Both ice cream sales and the homicide rate in Chicago go up during the hotter months and down in the cold months. But that doesn’t mean that eating ice cream causes homicides. Remember, for correlation, I don’t have to show any other relationship between the two other than their being mutual. Logically, this relationship makes sense to the average person.
Causation, however, is something else entirely. It’s the action of causing something, or the relationship between cause and effect. For causation to exist between two things, I must prove that one thing caused the other. Pulling a trigger mechanically causes a bullet to be fired from the barrel of a gun. I can prove this. So can you.
In statistics, this can be a more difficult task. How does one prove, without doubt, that one thing in society causes another with simple statistics taken separately of each other? Looking back at my previous example, does the fact that both ice cream sales and homicides in Chicago rise in the hot months and drop in the cold months mean that ice cream sales cause homicides in Chicago? Does it mean that homicides in Chicago cause ice cream sales to drop or rise? No, it does not. I have not statically proven that concept.
Correlation does not equal causation is a popular quote used in the world of statistics. Simply because you show you that two things have a mutual relationship does not mean one causes the other. Changes in ice cream sales don’t cause a change in homicides in Chicago anymore than changes in my wife’s mood will change whether it’s my job to take out the trash. On the list of other things that correlate to the homicide rate in Chicago, but don’t cause it are alligator attacks, lawn mower sales, visits to a swimming pool, or my personal consumption of Leinenkugel Summer Shandy.
Unfortunately, this goes both ways. When I see a statistic from the gun control lobby that the FBI shows a rise in mass shootings between 2001 and 2013, this is upsetting. While this statistic correlates with a rise in the number of guns owned in America between the same time period, that does not prove causation. There is no statistical proof showing that the rise in the number of guns caused more mass shootings. On the other hand, gun rights supporter love to use the fact that as the number of guns in America has gone up since 1993, the gun homicide rate has decreased in the same time period. They do, of course, correlate. But there’s no data showing that one has caused the other.
The left will use any tactic they can to trick, dupe, or otherwise fool the American public into believing their side of the story. They want what they want and will do anything to achieve it. They believe that the end justifies the means, and on the subject of guns, this is all too obvious. Ever notice how the people telling us that guns are bad are the same people surrounded by private security with concealed carry handguns protecting them?
To combat this, our side has used statistics that contradict the rhetoric that the left employs. This doesn’t justify the action on either side, however. But it does achieve the inadvertent goal of confusing the uninformed voter on what the facts really are, forcing them to work from emotion. Emotion is the means by which liberalism is spread…the concept that it doesn’t matter what caused something, it only matters how you feel about it. Are logic and reason enough to combat emotion? The way I see it, the answer to a question is often easier to attain than the reason behind it.
No. Sometimes I might as well be talking to a carrot.
Just remember: most times you aren’t debating to convert the mind of the opposing team/individual, you’re trying to win over the minds of the moderate and undecided; not the heavily indoctrinated that might only learn after a tragedy occurs to them.
You see Reverend Maynard, tomorrow is harvest day, and to them it is the Holocaust.
And I sprang from my slumber drenched in sweat like the tears of one million terrified brothers and roared, “Hear me now, I have seen the light!
They have a consciousness, they have a life, they have a soul! Damn you!
Let the rabbits wear glasses!
Save our brothers!
No, but it’s still fun to point out how the facts and data contradict their claims.
Aye. Making a liberal so mad that they just start screaming, “Racist!” is pretty entertaining. Bonus points if they cry real liberal tears out of frustration when backed into a logical corner.
Indeed, the whole point of feelings and emotion is to overwhelm reason and logic. If rationality was easy to come by, it would not be as prized as a commodity at it.
Well, we perhaps can’t prove that more guns = less crime, but we *can* prove that more guns does *not* = more crime.
It at least disproves the anti’s causation theory.
That’s half the battle right there…
Knowing is half the battle…
The other half? Red and blue lasers.
I won’t be buying either, they never hit anybody.
Not seeing “more guns not = more crime” proof. The article notes the antis use similar timing and trend vector to ascribe cause and effect. Where are the hard stats that more guns = less crime? I think there is correlation on that meme also, but no direct evidence that rules out any other factor(s).
“More guns != more crime” and “more guns = less crime” are not equivalent statements. Correlation without causation happens all the time, but causation without correlation? That’s pretty hard to swallow.
They’re starting to pick up on this, that’s why they’re moving to the “gun violence” rhetoric. Note the “gun” qualifier. They couldn’t give two flying fcks about “violence” unless it involves a gun.
What? That’s it? Dan, you were really working toward something, but it had to be something other than a surprise transition from statistics to emotion. The submittal needs to be expanded. We almost have something we can use in defense of gun rights. One missing piece is how to translate statistics (after further discussion about cause and effect) into statements, propositions that drive an emotional response to the thrust of the stats.
Please.
Not mine. Douglas wrote this.
My bad, just looked at the byline under the title. Can you forward comments to Douglas?
Sam I Am, I am following the comments.
I remember seeing that cartoon. Piled Higher and Deeper, or was it something else?
edit: Nevermind, I was reminded it was xkcd below.
Agree. This was driving to an interesting destination and got T-boned by the old left/right liberal/conservative drunk driver.
I stumbled in here from a Facebook link. I’m not a gunner, but subscribe to the notion of let people do what they want if its not hurting someone else. If you want to destroy old refrigerators with an AK-47, enjoy yourself. My old boss certainly did. I was hoping at the beginning of the article that it would lead somewhere enlightening but was disappointed.
Yeah I guess-but Chiraq is violent year-round. That’s what 150000 gangbangers get you. No “safe space” moment. And some of those hot weather shootings happen while waiting for ice cream. Unruly crowds and all that it brings. I know-I lived there. Interesting effort…
The thugs are on the street less when it is cold outside. The warmer the outside temperature, the more that the thugs go outside. Hey, the thugs want creature comforts too.
Running an XKCD without attribution? This site should know better.
Absolutely. Mr. Munroe is a good guy, he deserves credit where it’s due.
I couldn’t remember if it was PHD or something else. That’s for jogging the memory.
Again. Still.
Might as well post “The Fun Show Song” again.
Get your shit together TTAG.
oblig
Forgot to add the alt-text, which of course is the best part:
“Correlation doesn’t imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing ‘look over there’.”
As with all of our pics, hold your cursor over the pic for linkage and attribution.
Remember a good tool to use in arguing with the mentally liberal is instead of direct confrontation, ask them questions that force them to think, and defend their positions. Even those that strongly believe certain things, tend to stumble a bit when asked “why” they believe certain things. A typical liberal might say something like, “only the police should have guns.” And you ask, “why do you say that?” It then forces them to back up what they say, which ussually turns out to be emotion driven bullshit or lies. Which you can then correct them on.
The correct saying regarding statistics is
Statistics are like a bikini. What it reveal is substantial, what it hides is significant.
I’m sorry sir but I have been assured that both ice cream consumption and murder are caused by summer heat. And what of the catastrophic feedback loop that occurs when the truck runs out of pushups? That right there probably explains that barbecue massacre in Chicago a few weeks back. /sarc
Truth is you don’t need Statistics or facts. Pro gun must beat the drum from the moral high ground. No one, or group, nor ink on paper removes a citizen’s ability to protect themselves. Yet government and lobbist continue creating policy, legislation and taxes that murder, rape, and assault, for the most part, poor people.
But……..as more Americans now buy and carry guns….the crime rate did not increase, did it? So their main point is shown to be completely wrong….that more guns will lead to more crime and gun murder…so right there their whole belief system falls apart….and we can’t point that out enough, especially to non gun people who don’t follow the issue
Good point. At least it was shown that more guns did not lead to more murder of which this is the Statist position.
While correlation cannot prove causation, correlation CAN DISPROVE causation.
So… Anti-gunners love to preach that more guns = more crime.
The numbers show that since 1993 guns have nearly doubled while violent crime has dropped by half.
While we cannot say that more guns equals / causes less crime, we can certainly say that more guns does NOT cause more crime.
Very good.
No, we can’t. There can easily be uncontrolled variables that have effected the crime rate- higher rates of incarceration for offenders, a decrease in the amount of lead in the water, etc- that could be causing the drop in crime. These same variables could, theoretically, have caused crime to drop much further except for increased guns acting as a confounding variable to bring crime back up (though it still dropped from the baseline).
I’m not saying that’s true, but correlation cannot eliminate causation in the way you are describing unless you eliminate all those variables. And if you can eliminate all those variables you can probably just figure out the causation anyway.
By your definition, most statistical data would be worthless to solve anything. Even in the Process Engineering world, we cannot eliminate all minor variables in a study. We can monitor major variables.
Somewhere along the line someone is going to have to throw out the minor variables which just lead anyone performing research down the rabbit path.
We now have MORE guns than in 1993.
We now have LESS crime in 1993.
Therefore, we CAN prove that more guns does NOT equal more crime.
We are not trying to prove that more guns equals less crime. We do not know nor care WHY we have less crime. We just know that guns do not cause crime, or the crime rate would have necessarily increased.
Bingo!
The fact that (documented) increased gun ownership correlates with (documented) decreased violent crime, does indeed, disprove the causation claimed by antis that more guns equals more crime. There simply ISN’T more crime. There is LESS crime… by about 50%. The decrease in crime may be attributable to MANY things. But, to suggest that more guns equals more crime is untenable no matter how you look at it.
Australia had ALL these. It has an increased incarceration rate, reduction in lead blood levels, etc. Its murder rate dropped 41% since 1993. The US murder rate dropped over 60%. yet the gun control lobby says it was removal of guns that caused the 41% drop.
The article was a good, simple introduction to statistical concepts. That said, it should have included this principle (that correlation can disprove causation), since it is so applicable to the guns vs crime trope.
Another fundamental principle of statistics is the fact that causation can generally only be proved from statistics gathered in an experiment. Only after someone controls for all variables can causation be proven. Since crime is such a multi faceted beast, we will probably never be able to create an experiment that controls for all variables. We will probably never be able to prove any of the causes of crime. That means that we will need to take the approach of using correlations to disprove hypotheses until we are left with correlations that cannot be disproven. Although we still cannot assume causation in those instances, we will art least be operating from a more informed opinion.
Excellent article
Collective statistics are irrelevant; the rights of the individual matter. However, it’s completely obvious that a gun makes it easier to kill someone. But ease of method has almost no bearing on desire or motivation or the heart of a person who wants to kill someone else. If you want to kill someone, a gun is not a requirement. If you introduced legal guns into a relatively crime free, gun free nation like Japan, it’s pretty obvious to me the number of murders and suicides would not change much. It’s also obvious to me that many homicides would employ a firearm as the method of killing because it’s easier to kill with a gun.
Not much difference between Western European American gun crime rates and Western European gun crime rates, even though the Americans usually have much more lax gun laws and more guns than Europeans.
“The rights of the individual matter.” Which rights? The right to live free of crime, or the right to own a weapon to defend yourself? There’s a lot riding on the answer to that question. First, the notional difference is huge; on the one hand, an assumption of right which is wholly dependent on externalities. On the other, an assumption of right which is wholly self-dependent. And that speaks volumes to the mindset of the asserter.
Furthermore, there’s no natural “right” to live free of crime, although there is a strong desire to do so – a desire which leads men to form “government” to secure that desire. But there is a natural right to life – in the sense that everyone is entitled to defend against a mortal threat; it is the primary directive of a living thing, that no government can make a credible argument against.
In fact, in the government that men form – at least in our government – one of the declarative principles is that such a right to life exists (the Declaration) and that securing that right is one of the jobs of government – i.e. one of the reasons governments are instituted among men. Here, again, interpretation of that declarative principle speaks volumes about the mindset of the interpreter.
That is, as a “job” of government, does that mean government is responsible to protect one from threat, or is government responsible to protect one from infringement of one’s ability to confront a threat? The former is the mindset of the anti-gunner; the latter of the Founding Fathers. The Founding Father mindset is clear from the fact of existence of the Bill of Rights, the wording of the Second Amendment, and even from the pedestrian fact that police are not legally obligated to come to your defense, and are not liable if they don’t. Our very constructs unassailably echo that mindset.
As Scalia said: “Words have meaning.”
I’m reminded of the graphic made by The Yankee Marshall that showed a strong correlation between Democratic counties and mass shooting locations. I would postulate that there is more than correlation on that map.
Probably a fair amount of causation on that little graphic.
Look at the basic moral premise of the Donkeycrap Party.
The Government takes by force and gives to those it “feels” are deserving.
Apply the same philosophy to the street level and see what happens.
Yes – it’s positive proof that the problem is not guns, it’s Democrats with guns . . . (;-)
You need to attribute the cartoon to XKCD.
Charlie
Yeah, seriously… that’s kinda a basic practice to say nothing of copyright law (this probably wouldn’t apply under fair use anyway)
Hold your cursor of the image for attribution and linkage.
Attribution, yes. Linkage? Not so much. The particular comic can be found at http://xkcd.com/552/
You can preach that correlation does not equal causation but it won’t work on many. Humans are emotional.
Sometimes counter-propaganda has its place, even if it’s not entirely intellectually honest.
This is one of the few sites I frequent where I didn’t expect xkcd to pop up, but its great so what the hell.
I basically suck at math. Statistics is math so admitting that I am coming from the weak side I need those better at this to tell me where I missed the mark…..
If A is equal to B, then writing it as B=A is also correct.
If more guns = more crime then more crime = more guns should be equally correct, but is it?
There are more guns by any measure you can take.
There is less crime overall, the only way to show any increase is to cherry-pick data points.
So doesn’t that fail the first statement? More guns does not equal more crime?
And by that fail negate the implied causation?
More guns does not equal more crime because there are more guns and less crime. This does not prove that guns prevent crime but it very clearly proves the guns are not causing crime.
You could even reach a bit and say the guns aren’t the problem, criminals are the problem, but that might not be easily reflected by the statistics.
You could even reach a bit and say the guns aren’t the problem, criminals are the problem, but that might not be easily reflected by the statistics.
I would think statistics of monitoring career repeat criminals taken against crimes committed would easily show a correlation and at least an implied causation.
The problem with playing the statistic game, is that you still must have intelligent analysts collecting and interpreting the data.
You don’t need to be good at math if you’re good at logic. Your logic is good.
Yeah. I read Freakonomics too………
There are two other issues here, but that fall into the category of logic, rather than statistics: you say: “When I see a statistic from the gun control lobby that the FBI shows a rise in mass shootings between 2001 and 2013, this is upsetting. While this statistic correlates with a rise in the number of guns owned in America between the same time period, that does not prove causation.” That does not prove causation, and in fact there may be no causation – IN THAT DIRECTION. But it may well be the case that the increase in mass shootings has had a causative relationship with gun ownership – people, seeing a crime increase, may choose to arm themselves in response. Sort of like “increased ice cream consumption doesn’t cause increased heat; but increased heat causes increased ice cream consumption. The first step in judging causation from a correlation is to select which is the independent variable, and which is the dependent.
Then you say: “There is no statistical proof showing that the rise in the number of guns caused more mass shootings. On the other hand, gun rights supporter(s) love to use the fact that as the number of guns in America has gone up since 1993, the gun homicide rate has decreased in the same time period. They do, of course, correlate. But there’s no data showing that one has caused the other.” But the obverse conclusion has to be that there IS statistical proof that, since the gun homicide rate has decreased during a period in which gun ownership increased, there CANNOT be a causative effect of increased gun ownership on gun homicides. That is, inverse correlation DISPROVES causation – and may give at least rhetorical weight to the reversed causation argument (that is, that increased ownership has reduced gun homicide.)
Why is this important? Understanding the relationship between correlated events and logic principles helps defend against charlatanism and sophistry of the media and the biases of the emotionally-driven anti-gun crowd who want an outcome and will use any means to achieve it – even harmful ones, such as disarming a besieged public. As the saying goes: “figures don’t lie, but liars can figure.”
Comments are closed.