<img src="https://steemitimages.com/0x0/http://www.picserver.org/images/highway/phrases/gun-control.jpg"><br>Source <a href="http://www.picserver.org/g/gun-control.html">Here</a>

<p>So a few weeks ago I made a post finding a negative correlation between the number of gun laws and gun mortality rate by each state but my methods at doing so had some flaws which I will point out here. This isn't necessarily a correction but rather something intentional as many studies follow such a flawed path. This post will outline my flaws with finding that correlation, as there are some. As stated though, my findings are not pt through the rigor of an actual scientific study or approach but some studies have been published using similar methods and falling to similar flaws of my post. The post is <a href="https://steemit.com/statistics/@kryzsec/analysis-for-correlation-between-the-number-of-gun-laws-by-state-and-gun-related-mortality-rate">here</a> and this post is purely criticism. Before we get into the material I will state my bias on the subject to allow for you to dissociate between items that have been manipulated by my bias or otherwise.</p>
<hr>
<center><h1>Bias</h1></center>
<p>If anyone is curious about my position on why I think someone should be deemed safe to handle a gun before they can buy a gun (completely opinionated and not factual so it was left out of original post) it is because if someone whom is mentally unstable cannot buy guns legally then they will go to illegal methods... Now those that sell weapons illegally for a profit make money on repeat business and not going to jail, selling weapons to someone whom is mentally unstable is definitely not repeat business and if caught they cannot be trusted to not rat you out so it really wouldn't be in the best interest for them to sell that person weapons either. Look, worst case scenario we reduce the number of mass shootings (as seen in Canada, UK, Australia) but murders with guns still exist and best case won't happen. Because of my bias (and love for the taste of game meat) I do not agree with prevention of any guns being sold but rather restrictions on guns and equipment (suppressors, high capacity magazines, etc) being sold to those with proper gun handling training, no violent criminal background, and no mental illness.</p>

<hr>
<center><h1>How I did it (brief)</h1></center>
<p>What I did was I took the number of deaths caused by guns in each state per capita and the number of gun laws and I pt them into a range for each and then tallied up each into a chart and did a chi squared test to determine correlation. After determining correlation I looked at the trend to determine if it was a positive or negative (inverse) correlation. The data pointed to negative correlation between the variables meaning when the number of gun laws increased, the number of gun related mortalities per capita tended to decreased.</p>

<img src="https://steemitimages.com/0x0/https://imgur.com/7CAqiUI.gif"><br><p> Made By @gra</p>

<hr>
<center><h1>Problem 1: Control</h1></center>
<p>So this is something that can be alleviated through other means slightly but there is no control group (i.e. groups with no gun control, groups with complete gun control, groups with 0 gun related deaths, etc) and the reason this skews the final answer is because we have nothing to compare it to except for the data on its own. Now there are ways to get around this which brings us to problem 2.</p>
<hr>
<center><h1>Problem 2: Random Sampling</h1></center>
<p>Under certain circumstances there will be no way to have a true control group no matter how hard you try so to alleviate this we use a couple of methods, the first is random sampling. Random sampling is where you take random populations of the data and you use them for the statistical analysis rather than using all of the data at once. This allows for many tests from the initial population to be used in order to determine statistical significance.</p>

<hr>
<center><h1>Problem 3: Sample Size</h1></center>
<p>The second part to this is I had a relatively small sample size of 50. With that few samples we do not have enough for an accurate random sampling or otherwise. You see the larger the sample, the higher the probability is that it will drown out statistical noise. For instance, flips a coin 10 times and you might have it come out 70% heads and 30% tails, is it a rigged coin? Now flip that coin 100,000 times and if you have still around 70:30 split then maybe but if you have a difference of less than 5% or something like that then its likely that it isn't rigged. 50 data points is simply a really low number and the effects of noise on the analysis will play a large role.</p>

<hr>
<center><h1>First 3 Problems Together</h1></center>
<p>The first 3 problems are what I am calling noise groupings because they are used together to drown out statistical noise that might affect the study/analysis and none of them were used. The reason why none of them were used was there there was not enough data to have a large sample size or to do random sampling and I would have no idea how to approach a control group for that. The next set of problems are issues directly with the data.</p>

<hr>
<center><h1>Problem 4: Error Propagation</h1></center>
<p>So there are many aspects of this that should have an error attached to it but don't such as population, number of gun related mortalities, and thus the gun mortality rate per capita. You could also argue that the number of gun laws per state but the numbers are so small there that its relatively concrete although you should still use the smallest figure for error (for instance 11 would be 11.0±1) but what is more important is I did not allow for any error to take place and treated all the numbers as perfect concrete numbers.</p>

<hr>
<center><h1>Problem 5: Data Source</h1></center>
<p>Quite often you have to go off of second hand data but whenever possible you should confirm the data yourself. I took the data from my sources as is though I could not truly confirm these numbers myself easily. For this reason it is reasonable that I didn't but it is still a point that can be used to argue that the accuracy of my analysis is dependent on the accuracy of the data they provided. </p>
<hr>
<center><h1>Problem 6: Not All Gun Laws Are Equal</h1></center>
<p>The entire basis for the initial post assumes that more gun laws also equals greater gun control and this is simply... its stupid to think in such a way. The effect that this causes is unknown without knowing the exact weights of certain laws but it is, for all intense and purpose, impossible to solve that problem with the data presented.</p>
<hr>
<center><h1>How I Could Have Actually Tested This</h1></center>
<p>So now for actually testing this would be a lot more difficult but in an optimal siltation you could test it with 4 groups. First group is a group of states with the sale of 0 firearms, second group has no gun laws. Third and group has some gun laws. These are the 3 control groups and our 4th group we will manipulate the number of gun laws and see if there is a response in gun fatalities. We will use the first 3 groups to determine the statistical noise (remember a group can contain any number of states) and the 4th will be used to actually determine what effect the number of gun laws have when manipulated. You could also have the 4 groups be 4 staggered sets of number of gun laws. Another way to do this more accurately would be to look at these variables over a series of years and watch how they change and how the implementation of new laws affects the values but for the original intentions, that would be very difficult.</p>
<hr>
<center><h1>Conclusion</h1></center>
<p>I wrote that up with the intent of just seeing what happens when doing the test. I didn't have plans to use it as major reasoning to argue for or against gun control but instead did it out of curiosity. So there were problems with it and there were ways it could have been better, and maybe these things should actually be tested (not by a first year student) but that doesn't mean I will completely disregard those results. There were flaws, yes, but the fact that a primitive analysis can find a correlation implies that a more intense study should take place to either confirm or deny the findings. So while I do not say that my post should be used as a proper argument or some source to cite as evidence as that could mean that anything citing it as fact would be poisoned due to the... nature of the post, I am however saying that it should be use to have a discussion and considering that discussion cannot actually be held in the United States, it trly makes you think.</p>

<h1>References</h1>
<p>Picture: www.picserver.org/g/gun-control.html<br>Original Document: steemit.com/statistics/@kryzsec/analysis-for-correlation-between-the-number-of-gun-laws-by-state-and-gun-related-mortality-rate</p>

<div><center><h6><b><i>Kryzsec's Steemit Board</i></b></h6>
<br>
<a href="https://steemitboard.com/board.html?user=kryzsec">
<img src="https://steemitimages.com/50x60/https://steemitboard.com/@kryzsec/votes.png">
<img src="https://steemitimages.com/60x70/https://steemitboard.com/@kryzsec/comments.png">
<img src="https://steemitimages.com/70x80/https://steemitboard.com/@kryzsec/posts.png">
<img src="https://steemitimages.com/130x100/http://www.steemitboard.com/@kryzsec/st-nicholas.png">
<img src="https://steemitimages.com/70x80/https://steemitboard.com/@kryzsec/payout.png">
<img src="https://steemitimages.com/60x70/https://steemitboard.com/@kryzsec/commented.png">
<img src="https://steemitimages.com/50x60/https://steemitboard.com/@kryzsec/voted.png"></a></center><br>
<a href="https://www.steem.center/index.php?title=SteemSTEM"><img src="https://steemitimages.com/DQmf18V3PQenbUzYfkZcWSVGgmV5CtZNymi824SYwAQ2GTp/SteemSTEM-1.gif"></a>
</div>
<p><center>Image <a href="https://steemit.com/dtube/@nitesh9/amir0m2x">Source</a> from @nitesh9</center><br>Do you enjoy reading or writing topics related to STEM (<b>S</b>cience, <b>T</b>echnology, <b>E</b>ngineering, and <b>M</b>athematics) then I would suggest checking out @steemstem! They do wonderful work curating the best STEM related posts on Steemit. For more information check out the SteemStem chat room on steemit.chat or  check out their <a href="https://steemit.com/steemstem/@steemstem/helpful-guidelines-for-crafting-steemstem-content">Guidlines</a> and start writing.</p>