Criminalizing immigrants from Mexico has been a regularly strategy used by politicians and reporting media to motivate the construction of a wall on the U.S. Southern border. While Donald Trump was still a presidential candidate, he stated that "When Mexico sends its people, they're not sending their best. They're sending people that have a lot of problems, and they're bringing those problems with us. They're bringing drugs. They're bringing crime. They're rapists. And some, I assume, are good people." Is this remark reflected in states with high levels of Hispanic and Latino people? If so to what extent? This short study will take a look for relationships between crime and immigration, primarly by analysis at the state level, with an emphasis on the states which share a border with Mexico.
Is there any clear relationship between total crime and the percentage of immigrants per population?Is the number of Mexican Immigrants to U.S. increasing?
No. The number of Mexican immigrants to the U.S. began to increase rapidly beginning around 1970, but plateaud around 11.5 million in 2010. This number stayed roughly constant until during Trump's presidency in 2017, when we see a drop of around 300,000. Of course, any interesting trends will be buried in the noise around the last decade, which is what we will zoom in on shortly. First, what about the distribution of that population?
So that's what the overall count for the country looks like, but what about the distribution in states?
This animation shows that between 2000 and 2017, the border states New Mexico, Texas, California, and Nevada (in that order) had the highest populations of Hispanic and Latino people by percentage. NM Has consistently ranked first, with over 40% Hispanic/Latino population, while the other border states have stayed in roughly the 30-35% range.
Is there a relationship between the number of immigrants in the U.S. and crime?
This question paints too broad a stroke. As a 0th order analysis, let's consider
the total number of Mexican immigrants in the U.S. and total crime in New Mexico over the years. New Mexico has the highest Hispanic/Latino population percentage, right around 48% in recent years, so I will treat the total immigrants in the U.S. as a rough proxy for the level in New Mexico (don't worry-- I'll refine this analysis later). At first glance, the rate of total crime in NM and the total Mexican immigrants in the U.S. appear to be anti-correlated, with the exclusion of the last two data points, which fall in Trump's presidency. To see whether there is any relation here, let's calculate the correlation.
What do crime trends look like in the other border states?
Again, we're using a fairly wide-toothed comb here, and looking at total crime in each of the border states. It is clear that with the exception of New Mexico, all four border states exhibit a declining crime trend on average over the 12 year period shown. Notably, this data begins in the middle of Bush's second term, spans both of Obama's terms, and extends into the middle fo Trump's term. This data however, only shows the temporal change of crime in these states, but does not show the corresponding change in Hispanic/Latino population. Let's look at that next.
Can any significant correlation be drawn between Hispanic/Latino population and crime rate in border states?
Apparently, yes, at least for Texas and New Mexico. This plot shows the data from the previous plot, grouped by state and fit to a line,
where R in the legends is the correlation coefficient. Notably, both New Mexico and
Texas show correlations of 0.98 and -0.99 respectively, between total
crime rate and percentage of Hispanic/Latino population. That is, these two states
see strong relationships between crime and Hispanic/Latino population, but with oppositely
sloped trends. While a 1% increase in Hispanic/Latino population in New Mexico corresponds
roughly to crime rate increase of 1000 per million, the same population increase in Texas
corresponds to a crime rate decrease around 1250 per million! The other states exhibit a
more complicated relationship between crime rate and Hispanic/Latino population, as the
correlation magnitudes are lower, but a downward trend is still apparent. The conclusion
here is that the impact on crime rates from Hispanic/Latino population is state dependent,
relying on factors that are not accounted for with blanket statements regarding the raw
percentage of Hispanic and Latino people in any given place.
Is there an effect of Hispanic/Latino population on specific crime categories?
Digging into the particulars of crime by category can lead us into the weeds pretty quickly, and is beyond the scope of this study, but let's breifly constraint our attention to rates of rape by state. Among the descriptors for Mexican immigrants that Donald Trump has used, calling them rapists seems to a particularly strong statement, and is potentially very hurtful. The bar plot above shows the top 20 states by rate of rape (both 2013 revised and legacy definition) occurences, sorted left to right descending, alongside the Hispanic/Latino population percentage by state in 2015. In 2015, the state with highest number of rape cases per population was Alaska,
ranking in around 120 cases per 100,000 (or 120/100k) population. In second place was New Mexico,
with 80/100k. This is about 1.5 standard deviations above the mean for all states that year.
While New Mexico happens to be the state with the highest percentage of Hispanic and Latino
population, the plot below shows no clear relationship between the percentage of Hispanic and
Latino in a state and the rate of rape cases. Notably, Texas and Arizona, which have the second and
fourth largest percentage of Hispanic/Latino populations respectively, both fell over half a
standard deviation below the national average for rate of rape cases in 2015. Finally, California,
which ties with Texas for second largest Hispanic/Latino population percentage, does not even rank in this top 20 list. If bringing the rate of rape cases down is a goal -- and it ought to be -- let's work toward a real solution, rather than use severity denoted by the mention of rape as a way to sell immigration policy.
What about violent crime?
As above, there is no apparent link between Hispanic/Latino populatio levels and violent crime. Note that violent crime is defined to include murder and nonnegligent manslaughter, rape (2013 revised definition), robbery, and aggravated assault. We do see New Mexico in the top three again, which may point to the need for state specific legislation to better handle the crime there. Also of note, Alaska was first place by both rape and violent crime rate -- if the argument for regulating Mexican immigration is really about crime, I expect we should be hearing about crime policy in Alaska and day now.
Conclusion
We have extracted some meaningful results in this short study, despite the analysis being done with an admittedly wide-toothed comb. First, there is no evidence of a significant correlation between rates of crime and Mexican immigrants, in general. Second, among the four states which share a border with our fellow humans in Mexico, only New Mexico shows a significant positive correlation with the level of Hispanic/Latino population, with a slope of roughly 1000 crimes per 100k per percentage point increase in Hispanic/Latino population. In stark constrast, however, Texas sees roughly a 1000 crimes per 100k fewer for the same Hispanic/Latino population change! This demonstrates that the effect of crime due to people of this ethnicity is highly region dependent. Factors such as the local culture, law inforcement, and a myriad of other factors are at play. A blanket statement about crimes and troubles a particular people group are bringing with them isn't helping anyone or solving any problems. Calling a group of people criminals, and then voting for a wall to keep them out of "your" country is easy. Caring to dig deeper and learn the nuances of why people are emmigrating here in the first place, and whether and in what ways they are actually helping or hurting the societies they inhabit, is not. Please vote with a validated opinion.