Ordinarily I avoid on this blog topics that are obviously political in nature—even though anything connected with child care and child welfare is highly value-laden and therefore has strong political implications. However, I can no longer resist commenting on some critical thinking issues about discussions of killings by undocumented migrants.
In the last year or two, there have been several much-discussed events in which a person later discovered to be in the United States illegally has killed another person who was a legal resident of the country. These facts have been used as part of arguments against admitting most immigrants and in favor of draconian measures against undocumented persons already in the U.S.
Is it legitimate to reason from the killings to the positions taken by the people who advocate for much stricter anti-immigration measures? There are actually two ways we can think about this.
The first approach we might make to the making of a connection between killings by undocumented immigrants and severe immigration policies involves the idea of “the canary in the coal mine”. When little information about a danger is available , it can be the case that unusual attention should be paid to a few anecdotes. Although ordinarily psychologists assign little weight to anecdotal information, an important 2010 paper by Dimidjian and Hollon poited out that with respect to adverse events, attention to anecdotes may be the first line of defense. Anecdotes can alert us to potential problems, which can then be given systematic study. In cases where there are serious but rare problems that we would like to prevent, we might never notice such infrequent events without considering anecdotes, as the great preponderance of the information would “wash out” the problems statistically. This does not, of course, imply that we can generalize freely from a small number of anecdotes—it only means that anecdotes can guide our attention to risks that we might not otherwise notice. Without systematic study, even a large number of anecdotes can only give us proof of concept (as I am trying to argue in a paper I am presently working on, which involves about 50 anecdotes).
The second approach involves thinking through all of the possible events and looking at them in terms of the proportion each makes up of the whole set of events of interest. In the case of the migrant=killer argument stated earlier, there are two variables that can be used to categorize events. People may be classified according to their legal status in the United States (i.e., legal resident or undocumented resident). They may also be classified as to their involvement in a murder, either killing or being killed. This sets aside the great majority of the population, who, whatever their residence status, have not killed anyone, nor have they been killed.
Focusing our attention only on the small proportion of people who have killed or been killed, we see that there are four possible categories they may fall into: 1. Legal residents who have killed other legal residents; 2. Legal residents who have killed undocumented residents; 3. Undocumented residents who have killed other undocumented residents; and 4. Undocumented residents who have killed legal residents. The argument that undocumented residents are killers has been based only on the fourth category, undocumented residents who have killed legal residents. And you can create a table with four categories of interest, each of which needs to have a number filled in if we are going to use them to support any argument. (However, I apparently can't, because every time I try to post what I've done, the lines get scrambled, and I am giving it up.)
The conclusions we could draw—IF we had the numbers to plug into our table—would have to take account of proportions of people, not of absolute numbers. There are a lot more legal residents than there are undocumented residents, so we would expect more of the killings to be done by legal residents, and we would learn something from this analysis only if we look at the rates (number of killings compared to the total number of killers) for each category.
So what are the possible conclusions that we could draw from the completed table, if we had the numbers to fill it in? We might conclude that legal residents kill other legal residents at a higher rate than we see in the other categories-- or that legal residents are more likely to kill undocumented residents—or that undocumented residents are more likely to kill each other—or, what has recently been claimed, that undocumented residents have a high rate of killing legal residents.
Whatever the outcome and the policy argument it supports, we can’t figure it out until we have all the information required. The existence of a few anecdotes may encourage us to do this work, but until it is done, no conclusion is possible.