Far be it from me to question those who have a study to back their claims, but may I ask a question about either the data or the conclusion? In the American (other countries have their own version) version of debate over immigration, one side has convinced local and state legislators to pass “sanctuary laws,” the nature of which appears to be peripheral to the belief that regardless of their status innocent and good-willed people are no threat to the well-being of the nation. However, not everyone is good-willed. So, for many there is a need for an explanation when the well-being of citizens and legal immigrants might be threatened. Now, a study purports that the passing of SB54, which bars police from enforcing federal immigration laws, has had no effect on crime in California.* Essentially, the study says crime statistics are about the same as they were before SB54 was passed. I suppose that is good news, but is there a catch?
A steady-state crime environment is apparently okay for some, but is it possible, since those who did the study haven’t released the specifics, that immigrants committed crimes while citizens committed fewer crimes, thus keeping the numbers the same? In other words, who wants to accept a steady-state crime environment if American criminals are decreasing and undocumented criminals are increasing? Wouldn’t statistics on crimes per number of undocumented immigrants vs crimes per number of legal citizens be more revealing of the effects of sanctuaries? Just askin’.
Today, we are immersed in a sea of data, much of it used to manipulate us. In addressing how we might be manipulated, Cathy O’Neil exposed the widespread misuse of data and statistics in Weapons of Math Destruction.** O’Neil, writing about the use of teacher-evaluation statistics, cites a common phenomenon called Simpson’s Paradox: “when a whole body of data displays one trend, yet when broken into subgroups, the opposite trend comes into view for each of those subgroups” (136).* An example she uses comes the 1980s, during the Reagan administration when a report entitled Nation at Risk revealed that SAT scores were declining. The authors of that report ignored the effect of increased numbers of people taking the test and assumed the decline applied across the board. But when others examined the data for each socio-economic group, they found that the SAT scores had actually risen within each group, “from the poor to the rich.”***
We are easily swayed by numbers because we’ve been taught that they “don’t lie,” “are irrefutable,” or “are nothing if not objective.” Assuming I’m like you, I don’t want crime to increase; assuming I’m like you, I want it to decline. Assuming I’m like you if you are an American citizen, both of us can describe ourselves as a descendent of immigrants, mine having come through Ellis Island in that great wave fin de siècle and early twentieth century.
I know that hardships and crime always plague large migrations as some take advantage of the vulnerable among immigrant groups and native populations, and I’ve read about or even seen movies on Irish, Italian, German, Russian, and on just about every nationality’s gangs and criminals. Lots of good people have suffered and continue to suffer at the hands of a few bad people. And those bad people can be either homegrown or imported.
But my generalizations aside, I think we need to know what statistics truly say about any of us. And with regard to “groups” of any kind, from educational to religious to political, I would like to know that the statisticians have accounted for Simpson’s Paradox. Have the researchers who published the study on the effect of SB54 taken into account that paradox? I will acknowledge that one of the authors of the study, Charis Kubrin, plans other related and more specific studies. I hope that when she does conduct those further studies that Simpson’s Paradox plays no role in her conclusions. Let me ask again: Is it possible the reason crime statistics haven’t changed in a sanctuary state and its cities, is that citizens have committed fewer crimes while undocumented people have made up the difference? Just askin’.
Immigrant communities aren’t any more or less violent on average than any representative native population according to other studies, but even that fact, which is often tossed into the mix of research data on crime and immigrants, doesn’t eliminate the potential incorporation of Simpson’s Paradox into conclusions. I know that whenever data collectors tell me about a trend of any kind, I’m going to keep the paradox in mind lest I be tricked.
But a final word on the matter. No victim of a crime committed by an undocumented alien can be comforted by any trend in crime statistics just as no victim of a crime committed by a citizen can be comforted just because the perpetrator is a citizen. Relative changes in crime statistics mean nothing to victims. And whereas it is true that a citizen might be the victim of a crime by another citizen, the undeniable reality is that if the undocumented criminal were not in the country, the particular crime that person commits would never have occurred. There’s no better example of this than the tragic death of Kate Steinle whose life ended when five-time deported and undocumented alien shot her as she walked with her father. One wonders how any of those legislators who backed SB54 might have voted if they had been Kate’s parents. California experienced 1,661 homicides in 2015, the year Kate was killed. Had she not been shot, would a citizen have made up the difference by shooting another citizen? That, of course, would keep the crime statistic the same, showing neither a rise nor fall in violence. If SB54 were in effect at the time, would the substitute murder convince some that sanctuary cities have no effect on crime? Simpson’s Paradox.
*Shultz, David. 15 Feb. 2020. Crime did not surge when California became a ‘sanctuary state.’ Posted in Scientific Community. AAAS Meeting. The work by Kubrin was unpublished research at the time of this writing. Shultz’s report is online at https://www.sciencemag.org/news/2020/02/crime-did-not-surge-when-california-became-sanctuary-state Accessed February 20, 2020.
**O’Neil, Cathy. 2016. Weapons of Math Destruction: How Big Data Increases Ineqauality and Threatens Democracy. New York. Crown Publishing Group.
*** Nation at Risk, regardless of Simpson’s Paradox, influenced Americans’ views on public education, leading to a general notion that those schools were failing.
A steady-state crime environment is apparently okay for some, but is it possible, since those who did the study haven’t released the specifics, that immigrants committed crimes while citizens committed fewer crimes, thus keeping the numbers the same? In other words, who wants to accept a steady-state crime environment if American criminals are decreasing and undocumented criminals are increasing? Wouldn’t statistics on crimes per number of undocumented immigrants vs crimes per number of legal citizens be more revealing of the effects of sanctuaries? Just askin’.
Today, we are immersed in a sea of data, much of it used to manipulate us. In addressing how we might be manipulated, Cathy O’Neil exposed the widespread misuse of data and statistics in Weapons of Math Destruction.** O’Neil, writing about the use of teacher-evaluation statistics, cites a common phenomenon called Simpson’s Paradox: “when a whole body of data displays one trend, yet when broken into subgroups, the opposite trend comes into view for each of those subgroups” (136).* An example she uses comes the 1980s, during the Reagan administration when a report entitled Nation at Risk revealed that SAT scores were declining. The authors of that report ignored the effect of increased numbers of people taking the test and assumed the decline applied across the board. But when others examined the data for each socio-economic group, they found that the SAT scores had actually risen within each group, “from the poor to the rich.”***
We are easily swayed by numbers because we’ve been taught that they “don’t lie,” “are irrefutable,” or “are nothing if not objective.” Assuming I’m like you, I don’t want crime to increase; assuming I’m like you, I want it to decline. Assuming I’m like you if you are an American citizen, both of us can describe ourselves as a descendent of immigrants, mine having come through Ellis Island in that great wave fin de siècle and early twentieth century.
I know that hardships and crime always plague large migrations as some take advantage of the vulnerable among immigrant groups and native populations, and I’ve read about or even seen movies on Irish, Italian, German, Russian, and on just about every nationality’s gangs and criminals. Lots of good people have suffered and continue to suffer at the hands of a few bad people. And those bad people can be either homegrown or imported.
But my generalizations aside, I think we need to know what statistics truly say about any of us. And with regard to “groups” of any kind, from educational to religious to political, I would like to know that the statisticians have accounted for Simpson’s Paradox. Have the researchers who published the study on the effect of SB54 taken into account that paradox? I will acknowledge that one of the authors of the study, Charis Kubrin, plans other related and more specific studies. I hope that when she does conduct those further studies that Simpson’s Paradox plays no role in her conclusions. Let me ask again: Is it possible the reason crime statistics haven’t changed in a sanctuary state and its cities, is that citizens have committed fewer crimes while undocumented people have made up the difference? Just askin’.
Immigrant communities aren’t any more or less violent on average than any representative native population according to other studies, but even that fact, which is often tossed into the mix of research data on crime and immigrants, doesn’t eliminate the potential incorporation of Simpson’s Paradox into conclusions. I know that whenever data collectors tell me about a trend of any kind, I’m going to keep the paradox in mind lest I be tricked.
But a final word on the matter. No victim of a crime committed by an undocumented alien can be comforted by any trend in crime statistics just as no victim of a crime committed by a citizen can be comforted just because the perpetrator is a citizen. Relative changes in crime statistics mean nothing to victims. And whereas it is true that a citizen might be the victim of a crime by another citizen, the undeniable reality is that if the undocumented criminal were not in the country, the particular crime that person commits would never have occurred. There’s no better example of this than the tragic death of Kate Steinle whose life ended when five-time deported and undocumented alien shot her as she walked with her father. One wonders how any of those legislators who backed SB54 might have voted if they had been Kate’s parents. California experienced 1,661 homicides in 2015, the year Kate was killed. Had she not been shot, would a citizen have made up the difference by shooting another citizen? That, of course, would keep the crime statistic the same, showing neither a rise nor fall in violence. If SB54 were in effect at the time, would the substitute murder convince some that sanctuary cities have no effect on crime? Simpson’s Paradox.
*Shultz, David. 15 Feb. 2020. Crime did not surge when California became a ‘sanctuary state.’ Posted in Scientific Community. AAAS Meeting. The work by Kubrin was unpublished research at the time of this writing. Shultz’s report is online at https://www.sciencemag.org/news/2020/02/crime-did-not-surge-when-california-became-sanctuary-state Accessed February 20, 2020.
**O’Neil, Cathy. 2016. Weapons of Math Destruction: How Big Data Increases Ineqauality and Threatens Democracy. New York. Crown Publishing Group.
*** Nation at Risk, regardless of Simpson’s Paradox, influenced Americans’ views on public education, leading to a general notion that those schools were failing.
RSS Feed