The most important rule of self defense - Statistics

Status
Not open for further replies.
Should you walk outside of your house at night without first turning on the lights and looking out the window? Even though there has been no incident after several years doesnt mean it will never happen. In fact, after a series of no incidents tells me that it might be just about to happen...
Um, no. Not unless you are talking about mutually inclusive events, which I don't believe you are.

For example, based on statistics of the sun coming up every day for the past 1460000000000 days (4 billion years of days), I can postulate that the sun will come up every day forever and have a very high statisticaly probability that this will in fact happen. This would be a misapplication of statistics because each day's event of rise isn't mutually exclusive from the past events.

You see, we know that the sun won't last for ever and does have a finite life as a star. So with every day that passes, there is a greater and greater chance that the sun will die and hence not "come up." In other words, each day's event is tied to the events of the past. They are mutually inclusive.

Take the example with tossing a coin and doing heads or tails where it is assumed that the coin is actually (truly) balanced and 'sides' or 'edges' are not a realistic option. Each toss results in a 50% chance of a given outcome. Now say that you had 9 results that were actually the same, say they were all heads. What is the chance the 10th toss will result in heads? It can be argued with statistics that the chances are very good given the 9 previous outcomes. It can be argued with statistics that the chance is very bad as the longer you continue with such a pattern of results, the more likely the pattern is to be broken. Both sets of analysis would in error because they are treating the events as being mutually inclusive.

In reality, the chance of that 10th toss being heads again is the exact same as the original toss, 50-50. That is because each toss is mutually exclusive.

Applying such statistics to human behavior is problematic for this reason. People love to use statistics for things like estimating how much ammo they will need in a gun fight. While it may be most common for folks to only need 2-3 rounds for a gun fight (mean, median, or mode), what has happened in past gun fights does not determine or give a percentage indication of what will happen in any new gun fight in which any of us participate.

In fact, after a series of no incidents tells me that it might be just about to happen...
No. Note statistically, not unless you are doing something to influence that outcome. Mutually exclusive events are not linked through time (hence part of the reason they are mutually exclusive). There is no more chance of you having an event happen today than yesterday simply because of the passage of time just like there is no more chance of me winning the lottery today because I haven't won't it in the past 10 years of paying.
 
So what are you suggesting as an alternative that would make one less likely to come up with a poor plan?

I think reading an academic study about crime and using the statistics within to formulate an action plan of self defense without understanding what the statistics mean and their context and bias might cause you to formulate a bad plan.
My alternative? Training by experienced people supplemented by some reading on the subject (not an academic study). Some advice I see on these forum (some) is more useful to me than reading Gary Kleck's works.
 
Last edited:
It can be argued with statistics that the chances are very good given the 9 previous outcomes.

Actually, that can not be argued with statistics. It COULD be argued by a person that doesn't understand statistics. In order for it to be a valid experiment we must assume that the coin is balanced and that there is no external influence on its' landing. Making those assumptions, there is EXACTLY a 50% chance that the next toss will be heads. The odds of your 9 heads in a row is roughly 19 in 10,000. The odds of getting 100 heads in a row are very low (7.888609X10^-31, approximately) but it could happen. Once it does happen, the odds of getting heads on #101 are exactly 50%. Previous outcomes of random events DO NOT affect future outcomes.
 
Last edited:
One should be a touch careful about basing arguments on 'experienced people' - let me give you a counter example.

Academic research using standard statistical methods have demonstrated the lack of utility of the AWB and the large number of defensive gun usages. These studies have been strong counterevidence to the folk knowledge of some (not all) antigun law enforcement experienced people - like chief of police types who are brought forward to support all gun bans, oppose shall issue laws and the like.

So, you might get what you wish for. Arguing from the one vivid instance and then overestimating the probability of such is a common mistake.

Cho used handguns - experts say handguns kill - ban them.
 
The danger is that if you don't really understand what these mathmatical formulae mean then you might come up with a poor plan.

So what are you suggesting as an alternative that would make one less likely to come up with a poor plan?
good point, John. I suppose one could develop a plan from watching Rambo movies, or reading Sin City comics, or any of a number of other fictional sources. You don't always need to know what the formula means in order to understand the results. You really don't need to understand levels of confidence, multivariate and bivariate analysis, and so on to understand that you don't need to worry much about being attacked by a tiger in your living room when you find out the odds. The real dange, IMO, is developing plans based on incorrect perceptions, bad data, and so on. THAT is where the poor plan is more likely to come from, IMO.
 
People love to use statistics for things like estimating how much ammo they will need in a gun fight. While it may be most common for folks to only need 2-3 rounds for a gun fight (mean, median, or mode), what has happened in past gun fights does not determine or give a percentage indication of what will happen in any new gun fight in which any of us participate.
Sure it does. It lets you know that there is a high percentage of probability that you will be able to solve the problem with a fairly low number of rounds. It tells us that we probably don't need to carry 100 rounds with us all the time. You might not be right in your decision about how much to carry, but the statistics have given you a pretty good idea about how many rounds you will neeed in a gunfight. Past gunfights don't determine what wil lhappen in a new gunfight, true, but they do give us an indication of what is likely to happen and what is unlikely.
 
Academic research using standard statistical methods have demonstrated the lack of utility of the AWB and the large number of defensive gun usages. These studies have been strong counterevidence to the folk knowledge of some (not all) antigun law enforcement experienced people - like chief of police types who are brought forward to support all gun bans, oppose shall issue laws and the like.

I think using statistics to make policy decisions (like your example) has merit but that is off topic.

The real dange, IMO, is developing plans based on incorrect perceptions, bad data, and so on. THAT is where the poor plan is more likely to come from, IMO.

Precisely my point. Many academic studies have all those faulty elements; poorly formed premises, measuring the wrong thing, not taking bias and other factors into consideration, etc. As I have stated earlier, just pulling some statistic from a study or report and then making a plan based on that without understanding what the stats mean or how they were gathered can cause you to form a bad plan. Just because some academic publishes something surely doesn't make it so. In fact if you read that stats from here
You probably would plan to have no gun around ever.

But if you read this one:
you might have ten guns.

Do I really need to read some mathematical formula to plan how to protect myself rather than use acknowledged experts in the field and my own common sense to filter their bias? I think not. BTW the references to Tigers attacking you is strawman stuff.

So, you might get what you wish for. Arguing from the one vivid instance and then overestimating the probability of such is a common mistake.

I have not proposed that Glenn, that is not my position. However, if the vivid instance as you call it matches what I encounter it might have more relevance than a linear progression chart.

Cho used handguns - experts say handguns kill - ban them.

I think the Volokh conspiracy calls that an intuitive argument.;)


It COULD be argued by a person that doesn't understand statistics.

My point here is that most people don't understand statistics (and some who claim to don't either) other than in a very rudimentary way and I am dubious that this is really helping someone make informed decisions with their self defense tactics?:rolleyes:
 
Last edited:
I am dubious that this is really helping someone make informed decisions with their self defense tactics?

No argument there.;) It's a fun and interesting discussion though.

The main area that I think statistics can be useful is when we have little personal experience for a particular event but which might be more common than we expect. It's fairly dangerous to say "Well, it's never happened to me..." when you might be the exception rather than the rule. On the flip side, it can prevent us from being overly paranoid about things that we seem to think we need to be concerned with but are statistically highly improbable, like getting hit by "ice" dropped from an airliner.
 
In my experience the statistically correct amount of ammo to carry depends on such variables as how much I can fit in my gear and who's got extra mags... :D

Milspec
 
I dont think a tiger or space alien will be in your living room anytime in the future. The chances of that are zero.

However, there is a chance of someone forcing entry into your house to get into your living room. Although the chance is small, there is still a chance.

I recount a story that was told to me by an ex-girlfriend of being the victim of a home invasion incident. A group of three guys broke in and made the family get on the ground while they ransacked the place. The family was not hurt.

So, no, a tiger isnt coming to your living room, but someone else might...
 
I think reading an academic study about crime and using the statistics within to formulate an action plan of self defense without understanding what the statistics mean and their context and bias might cause you to formulate a bad plan.
My alternative? Training by experienced people supplemented by some reading on the subject (not an academic study). Some advice I see on these forum (some) is more useful to me than reading Gary Kleck's works.
Given that a lack of understanding of the data collection method and the context & bias seem to be your objections, how would one be better served by your solution?

I believe people have as much or more confusion about how to interpret a single person's experiences/anecdotes (experience as a data collection method) than they do about statistics. In fact, to some extent they go hand in hand. Similarly, determining the context & bias of a trainer or a person posting on the internet would seem to be just as difficult (if not more so) than determining the same for an academic researcher.

I agree that a person needs a basic understanding of the data collection method, and also needs to consider context and bias, I just don't see how that argues more strongly against statistical data provided via academic study than it does against experiential (anecdotal) data provided via word of mouth/internet.
The real dange, IMO, is developing plans based on incorrect perceptions, bad data, and so on.
I believe that this is what it comes down to regardless of one's sources.
 
Sure it does. It lets you know that there is a high percentage of probability that you will be able to solve the problem with a fairly low number of rounds.

No, simple percentage historical data only tell you what the chances where in the past. It is your interpretation of the historical data, your assumption, that they project into the future.

Part of the way in which the percentage average is kept low is due to gun capacities. Low capacity guns introduce a bias into the data that keep the percentages lower.

You have also assumed that the problem will "solved" with that number of rounds. In reality, that number of rounds average includes those where the shooter, good or bad, lost of the fight.
 
Statistics

Statistics are just that, Statistics. I've read artical after artical of statistics. one thing that is the same in all. What can go wrong will go wrong. Plans are great until you have to act. I've been in law enforcement sense '92. I have planned for the worst but hoped for the best. I started shooting at age 9. I have won pistol matches too. I stopped a 15yo kid one night that was carring a 357 mag rev. Before I could put my car in park, he exited walking back at me. Plans are great, but that night my plan went out the door. My gun jamed. The same that I had trained with and shot matches with. Statisticly I should be dead. Turns out his gun was unloaded. I don't care for statistics. Alway plan for the oh s..., what now. And hope your plan works.
 
I think reading an academic study about crime and using the statistics within to formulate an action plan of self defense without understanding what the statistics mean and their context and bias might cause you to formulate a bad plan.
Failure to use something right is not an indictment of that thing. Using ANY source of information without understanding what it means may cause you to formulate a bad plan. And using the stats available, even if used wrong, is apt to be more accurate than watching the news, or listening to a trainer, or "I just think, or "better to have...", etc.
My alternative? Training by experienced people supplemented by some reading on the subject (not an academic study).
First, many of those "experienced people" have little or no actual experience. Second, where do you think those "experienced people" come up with what they feel works, or what should be taught, and so on? It is by analyzing events. Some do it better than other becuase they actually look at the research. And academic studies are apt to be far more accurate and rigorous in their findings than are non-academic works. Academic studies get peer reviewed most of the time, while other readings are frequently done for profit.
 
Past gunfights don't determine what wil lhappen in a new gunfight, true, but they do give us an indication of what is likely to happen and what is unlikely

Hmm. No, not really. Not for any particular single gunfight.

This is a quite common mistake for people to make about statistics; even people who have had training and should know better.

It is very common, for example, for physicians (most of whom have had some training in statistics) to quote statistics on medications and procedures in such a way as to communicate to the patient that: Because a certain medication/procedure has been shown to be effective in 78% of cases, then, should the patient choose this medication/procedure, she has a 78% chance of success.

Not so, of course.

In fact, no probability value for success/failure (in this one case) can be assigned with any confidence. I suppose a person could be excused for saying something like; “I think you’ve got a pretty good chance of success here” (not mathematically sound, but excusable).

Back to the main point; while statistics can be useful in the general sense of preparation for the average occurrence, it would be a serious error to depend upon them to predict how a particular situation will develop, let alone how it will work out in the end.

Stay safe and let your statistic be counted in the alive and breathing set.

Best,

Will
 
There can be a tiger in your house. A few years ago, San Antonio had a major flood. Just a few miles from me, it washed out the fences on an animal rescue preserve. The lion got out. It ate the ostrich and then wandered down the road. It's distance and path was not in our direction. However, if it had chosen to walk in another direction, it would have reached our house.

True, it would have had quite a few other houses to choose from.

Years, ago - my mother-in-law moved into a neighborhood where some dirt bag kept a lion in the backyard as a pet.

Thus, I do worry about large predator attacks and plan accordingly.

Ok :D - one problem with this debate is that some of you (no offense) really don't understand statistical methodology usage. I keep seeing the error of planning as if the central tendency is guaranteed to happen and not understand confidence intervals and the various errors / risks associated with it.

It's very easy - what value to you give to the extremes in planning?
 
No, simple percentage historical data only tell you what the chances where in the past. It is your interpretation of the historical data, your assumption, that they project into the future.
No, and this is where people continually get it wrong, IMO. The historical data (descriptive stats) does not tell you what the chances were in the past, is tells you what actually did happen in the past. Statistics then allow you to take that known data and make assumptions (inferential stats) based on certain proven mathematical formulas. Those allow you to determine, with a varying degree of accuracy, what the chances of something in the future will be.
Part of the way in which the percentage average is kept low is due to gun capacities. Low capacity guns introduce a bias into the data that keep the percentages lower.
That is not a bias, that is a factual piece of data to be included. It's like saying the average age of people dying is kept low because people don't live long. Gun capacity is fairly low, but that doesn't change the fact that most gunfights are solved with a very low number of rounds.
You have also assumed that the problem will "solved" with that number of rounds.
No, you have assumed there is an acceptable chance of success with that number of rounds. You accept the fact that you might be wrong, but decide the chance of being wrong is so low it does not justify any further investment in reducing that chance. Again, the tiger in the living room problem.
In reality, that number of rounds average includes those where the shooter, good or bad, lost of the fight.
That can be a reality. Depending on how you choose to use the data and what data you collect you can have averages from win and lose, those who win only, those who lose only. As a matter of reality, most studies that I have seen show all three of those numbers being very close together.
 
Hmm. No, not really. Not for any particular single gunfight.
Yes, they do. That is the essence of inferential statistics.
In fact, no probability value for success/failure (in this one case) can be assigned with any confidence.
Sure it can. It is done all the time in all sorts of activities. I have seen nothing to indicate that DGUs would be any different.
 
Status
Not open for further replies.
Back
Top