Remarks by the President on Common-Sense Gun Safety Reform

[opinterph]

I have been more concerned with the map at 164

I think you meant the map at what is currently post #161:

View attachment 1146116

I also do not think it is a coincidence that the high rate areas are colored red as in red for Republican while allegedly low rate areas are democrat blue.

Using two contrasting colors and 8 intervals enables the viewer to quickly determine which areas are above or below the median.

Here’s a totally blue version, if that will soothe your anxiety. (same data, different visual representation)

 

[Kabluey]

Smoothing is intended to be used when accurate data exists for some counties but not all the nearby counties. It may sometimes by reasonable to guess that the counties have similar rates of disease, etc. As the CDC explains in the link: "Geospatial smoothing can be applied to national, regional, or state maps showing county-level detail. When a county-level map indicates many unstable (or undefined) rates, geographic patterns can sometimes be clarified by displaying smoothed rates. For a given county, the smoothed rate is calculated by using."

But, in this case the method is being totally misused, because to fabricate deaths where they did not occur. And, our poor liberals have willingly allowed themselves to be flimflamed into thinking that the desert has huge rates of deaths.

And, yes, this is all done pursuant to an Agenda to show a link between gun ownership and deaths. They explain: "See our longer post on the links between gun ownership and gun death rates."

The smoothing process is similar to a noise-reduction and dust/scratch elimination algorithm used for photo and video reproduction. Missing image data is averaged in from adjacent pixels to approximate the original picture.

The resulting image is in general as close as possible to how it would look without the missing image data.

You might argue that anything was obscured by a hair or scratch, but chances are, the smoothing algorithm is correct.

Also note that isolated counties aren't smoothed.

 

[Benvolio]

I continue object to the clear distortion which colors the unsettled areas dark, by allowing a few deaths to represent the rate for a very large area. Notice that all of Alaska is dark, while Illinois is very light. But it is more attractive blue.

 

[jbrown329]

You do realize how many people live in Alaska? So, if Alaska had 40 shootings (state wide).......... screw it. No matter how simply it's explained, you just don't get it.

 

[Benvolio]

You do realize how many people live in Alaska? So, if Alaska had 40 shootings (state wide).......... screw it. No matter how simply it's explained, you just don't get it.

No you don't get it. With, you say, 40 shootings, the entire state including the frozen north, the uninhabited mountains, are all colored dark red for high rate of shootings. BUT CA. IL, NY with many more shootings and much less territory are light colored. It is a function of the "smoothing"; tarring some areas with the same brush as others. It is misleading you.

 

[rareboy]

Oh.

My.

God.

You still don't understand the concept of gun deaths per 100000 population.

Beyond hope.

 

[rareboy]

Based on this map, I'm sure that Benvolio will now be convinced that all gun deaths were caused by Democrats.

 

[Benvolio]

I do not find the link to interactive to show the map without the "smoothing". It would show a much different picture.

 

[Kabluey]

I do not find the link to interactive to show the map without the "smoothing". It would show a much different picture.

It would still show pretty much the same isolated areas to have higher rates of gun deaths per capita.

You don't seem to mind having vast geographical swathes of the USA coloured as Republican voters...

 

[rareboy]

It would still show pretty much the same isolated areas to have higher rates of gun deaths per capita.

You don't seem to mind having vast geographical swathes of the USA coloured as Republican voters...

I'm sure that for Ben, who thinks the map is better all blue, it carries the suggestion that only Democrats have been killed by guns.

And that would make him hard.

 

[jbrown329]

No you don't get it. With, you say, 40 shootings, the entire state including the frozen north, the uninhabited mountains, are all colored dark red for high rate of shootings. BUT CA. IL, NY with many more shootings and much less territory are light colored. It is a function of the "smoothing"; tarring some areas with the same brush as others. It is misleading you.

https://en.wikipedia.org/wiki/Per_capita

 

[rareboy]

https://en.wikipedia.org/wiki/Per_capita

Matthew 7:6

 

[Kulindahr]

On the concealed carry front, are there any statistics on concealed carry individuals stopping things like mass shootings, versus starting them?

There have been a couple of instances already this year where concealed carriers have stopped what looked to be mass shootings (my favorite being a guy who just opened fire at a gas station supposedly because there were people between him and getting gas, and a concealed carrier took him down -- what a great place to be shooting; to quote a line from Hunt for Red October, "Some things here don't react well to bullets"). At the same time, no concealed carrier (i.e. someone with a license/permit to do so) has ever started a mass shooting.

 

[Kulindahr]

Oregon’s suicide rate is among the highest in the country.

Limiting the CDC output to only homicide by firearm, the map below demonstrates that most counties in the State of Oregon do not register enough deaths to generate accurate results in the determination of a rate to associate with the county.

---

Including geospatial smoothing adds more data to the otherwise unremarkable counties, which allows the observer to better differentiate the counties relative to each other. A larger data set renders the output more useful by allowing extrapolation to generate a comparison that better quantifies an associated value for each respective county.

Smoothing geographically in this case does exactly what benvolio is pointing out: it results in lies. If a county has no fatal shootings but is surrounded by counties that do have, smoothing could result in the assertion that said county does have fatal shootings.

Smoothing has to be used appropriately. In this case, smoothing should be done over time, so that a population-small county with a pair of fatal shootings one year doesn't show a, say, sixteen per 100k rate for that year when the reality is it was twenty years since the last fatal shooting, so the smoothed figure would be 0.8 per 100k, which gives a better sense of the real rate. Knowing which sort of smoothing to use for a given data set is essential to make sure the result is truth rather than falsehood.


BTW, including suicides isn't helpful in any way in telling anyone how dangerous an area is in terms of being the victim of a shooting, since suicides generally shoot only themselves. It's certainly useful for addressing whether one might wish to live there (Is there something exceptionally depressing about the place?), but irrelevant to going about one's everyday life.

 

[Kulindahr]

I think you meant the map at what is currently post #161:



Using two contrasting colors and 8 intervals enables the viewer to quickly determine which areas are above or below the median.

Here’s a totally blue version, if that will soothe your anxiety. (same data, different visual representation)

Looking at that map I have to wonder how well it correlates with people liking "Old West" movies and themes where shootouts are considered a manly way of settling issues.

 

[Kulindahr]

No you don't get it. With, you say, 40 shootings, the entire state including the frozen north, the uninhabited mountains, are all colored dark red for high rate of shootings. BUT CA. IL, NY with many more shootings and much less territory are light colored. It is a function of the "smoothing"; tarring some areas with the same brush as others. It is misleading you.

It's misleading if you're trying to choose an area for going on a walking journey and want to know your risk, yes. But that is inevitable when the map goes by political subdivision; and the map goes by political subdivision because that's how the data is readily available. To get a map showing actual risk by location, one would have to throw out political subdivisions, plot shootings by precise GPS coordinates, and do something like a contour map.

 

[Benvolio]

It's misleading if you're trying to choose an area for going on a walking journey and want to know your risk, yes. But that is inevitable when the map goes by political subdivision; and the map goes by political subdivision because that's how the data is readily available. To get a map showing actual risk by location, one would have to throw out political subdivisions, plot shootings by precise GPS coordinates, and do something like a contour map.

It would be less misleading if the smoothing technique were not used.-

 

[T-Rexx]

It would be less misleading if the smoothing technique were not used.-

No, it wouldn't.

That's the whole point of smoothing - to make trends truly present within the data more understandable to the human eye.

All real data have statistical error inherent within. Statistical "outliers" (such as counties with no deaths during the survey period) deceive the eye. There is no such thing as an inhabited region with "no" gun deaths any more than there is such a thing as a region with "no" crime. The reality is that some regions with gun deaths just didn't happen to have one during the survey period, because the populations were too small. To claim that such a region never, ever suffers a gun death would be false, but the raw data make that appear to be the case. Presenting such raw data visually would be deception. In fact, the presentation of unsmoothed data is one way that nefarious people sometimes use to deceive their audience into believing something that is not true. The smoothed data is actually a more accurate presentation of the truth than the raw data. That is why smoothing is so common when data are presented visually.

To illustrate, in 2015 the highest rate of gun death in the USA occurred in Renegade Mountain, Tennessee, when 4 people were killed. Only 40 people lived in Renegade Mountain, so that was 10% of the population. It would be absurd to claim that Renegade Mountain, Tennessee has a 10% murder rate, but that is what the raw data imply. This was a statistical outlier. The 2015 deaths were probably the only gun deaths there in 100 years. Smoothing the data with the surrounding area is a better depiction of the truth.

I'm not asking you to believe this - I know you and Kuli don't like science. But that is why these data are presented this way. It is a more accurate presentation of the truth.

 

[Benvolio]

Smoothing in this case makes it appear that sparcely inhabited areas have high rates of shooting deaths because a few deaths are smeared over a large area. But the smear serves the democrat and your political agenda.

 

[T-Rexx]

Smoothing in this case makes it appear that sparcely inhabited areas have high rates of shooting deaths because a few deaths are smeared over a large area.

Again, smoothing shows the sparsely-populated areas to have high rates of gun deaths because those areas have high rates of gun deaths.