Australian Gun Stats? Do Stricter Gun Laws Help?
*SPOILER* This title is a massive clickbait, if you thought that this was the moment I attempt to answer all the questions you ever had about this, you were mislead. I’m just going to show you why you shouldn’t read charts wrong.
So, I keep getting sent this tweet.
And the thing is, everyone tells me “this tweet is dumb”, and yes, it is a very poor interpretation of the data presented. But let’s all take a moment to talk about why this is a very poor (and basically non-existent) analysis of this graph.
So, our independent variable is time/the passing of the 1996 Australian gun law and our dependent variable is how many deaths by guns in Australia. This isn’t a controlled study or anything so we can’t guarantee that there aren’t confounding variables, but going based on the data presented to us our goal will be to determine whether the passing of the 1996 Australian gun law had an effect on the number of deaths in Australia.
For the sake of statistics, let’s state our null and alternate hypothesis.
H0: The 1996 Australian gun law had no affect on the number of deaths by guns in Australia
H1: There is a significant effect
This is how I organized the data in the chart
I imported this data into R Studio and then created a box and whisker plot
You can see here that there’s a pretty big difference in terms of the medians and the data spread for each group. After the law was passed it appears there is a large drop in the number of deaths compared to before.
But you can’t just trust an eye-test, so I then did a Welch’s T-Test because the groups are unequal in size and I really want to see if there is a significant difference.
At a 95% confidence interval, there appears to be a significant difference in gun related deaths in Australia between the time periods before the 1996 Australian gun law and after.
It looks like based on just this data, stricter guns laws reduce gun related deaths. I tried to find statistics about gun deaths in other parts of the world from 1988–2012, but I can’t find anymore data from that specific time frame and I don’t think it’s a good idea to use other data because that would add even more confounding variables and increase the likelihood of a type-1 error.
But, my point here is that you can’t look at statistics or charts the way the fellow in that tweet did and that we really need to realize when our own bias is affecting our ability to accurately discern data.
This isn’t the best analysis or report, but it’s just a little look. I’m also pretty rusty with R so correct me wherever you can. Thanks have a nice day.
