Thinking Data
Is crime correlated to wages, and if so how?
Are rising wages the reason for falling crime rates? Read more to find out
Crime rates have been falling since the 90s and people can’t always agree on why. However, one reason that people cite a lot is the better wages from a better economy. The logic says that if people are paid more they will have less of a reason to turn to crime. This makes sense. But is this what happens?
The data I used to get the wages for each state were from the Bureau of Economic Analysis of the US department of commerce. I acquired data for state crime rates from the Corgis-edu dataset run by Austin Cory Bart, Ryan Whitcomb, Jason Riddle, Omar Saleem, Dr. Eli Tilevich, Dr. Clifford A. Shaffer, and Dr. Dennis Kafura. To get the relevant data I wanted I made two filter functions
I got the data I wanted for income by filtering the crime and wage datasets using two functions I made. The crime dataset had extremely erratic behavior between 2012 and 2015 where crime rates crashed from around four thousand to a few hundred. This does not make any sense especially since rates went back up following 2015 to normal values. I isolated these values and turned them into nulls. After that, I used the Numpy function interpolate which generates values to take the place of null values using the other data points and Machine Learning. Following this, the graphs looked much more normal and made sense. My next step was to convert each wage value into 2015 dollars. I did this using the CPIs of each year which I got from the OECD inflation dataset. I plotted both against time to see if they looked similar or followed any trends. Then I plotted them against each other and added a regression line to see the correlation. I also calculated the exact correlation coefficient to get some quantitative data.
You can interact with the graph above by using the home, search, and pan buttons above. The buttons appear when you hover your cursor over a graph. Click on the search or pan icons and drag on the graphs to interact with them. Press the home button to reset the graphs.
​
The graph of USA crime rates shares no correlation with the graph of USA wages. The graph of crime rates for the USA shows an extremely rapid rise between 1960 and 1980 where crime rates rose from 2500 to around 6000. There is a small dip around 1985 where crime rates drop to approximately 5000 before returning to around 6000. The graph then declines from around 6000 to 2500 between 1990 and 2019. but at a moderate almost constant rate. The graph of wages in the USA is even quite consistent with only a few noticeable jumps and falls. There was a small dip around 2008 in the wage graph due to the recession. The two graphs of crime and wages look utterly unrelated. This thought is only reinforced when looking at the last graph which plots crime and wages against each other with a line of best fit. The crime rate strays far from the line of best fit and follows its own pattern without any regard for wage values. The exact correlation coefficient for this graph is -0.02 which is so small that it can be ignored.
The California graph of crime is similar to the USA one in that they both are peaking between 1980 and 1990. The crime graph for California shows a sharp incline between 1960 and 1980 much like the USA graph. Crime rates rose from around 3500 to almost 8000. Crime then fell slightly following 1980 to around 6500 before plateauing for around a decade. Crime rates then declined sharply between 1990 and 2000. Crime fell from around 7000 to approximately 4000 which is an extremely sharp fall. The USA also had a fall in crime following 1990 but it was not nearly this steep. Crime then continues to decrease at a moderate pace after 2000. The wage graph shares no correlation with the graph of crime rates as it rises gradually from 1960 to 2019 like the USA wage graph. The wage graph shows a small dip in 2008 during the recession but seems mostly linear otherwise. This is in jarring contrast to the erratic tendency of the crime graph. The two graphs look completely unrelated once again. The third graph shows even more erratic tendencies with the line of best fit being ignored. The wage and crime rates had an exact correlation of -0.53 which is a decent negative correlation! This does not make sense and a few other examples will show that it is just a coincidence.
I also analyzed the data for a few other states. Texas has a correlation of 0.07, Florida has -0.02, New Hampshire has 0.02, and Pennsylvania has 0.02. All of this shows that wage and crime are completely unrelated. Crime rates being correlated to wages is a common misconception that even new companies have made in the past. I analyzed this because I was curious whether this was an actual correlation. The logic is sound but the theory that wages affect crime rates is false. I used data from the government, professors, and the OECD institution to come to this conclusion. The graphs showed no correlation and the value of the correlation coefficient proved this for each state and the USA. I hope you found this article informative. Thank you for reading.