Tuesday, September 21, 2010

Does Political Leaning Bend Logic?

I don't know the political leanings of Stephen D. Levitt and Stephen J. Dubner, the authors of Freakonomics, but they make some interesting assumptions in their book and fail to challenge some assumptions they present uncritically that seem illogical on their face. They come out against ideas generally supported by the right (for example, by challenging the utility of capital punishment as a crime deterrent), but they make interesting inclusions and omissions of fact in different situations where one of the factors that changes is the race of the person(s) discussed.

The most striking quote that they present without criticizing its underpinnings is a quote in the book from an economist named Gary Becker who (on page 121) says, in part, "African-Americans and Hispanics commit a disproportionate share of felonies". Note the use of the verb "commit" (noting there is no accurate record of how many crimes are actually committed, since a sizable number of crimes inherently go undetected/unpunished) versus the use of the phrase "are convicted of" (which can be objectively proven with statistics regularly gathered by the Department of Justice). How could one calculate who commits every felony that occurs in America, absent arrests or convictions? You can't, unless you're omniscient. So, how did a pair of economists who spend an entire book undercutting "conventional wisdom" with unblinking logic fail to note that Mr. Becker's claim was blatantly unprovable and, worse, quote it as fact? This is the typical blindside created in a society inundated with racism where certain racial stereotypes become so accepted ("Blacks/Latinos are more likely to be criminals") that when these unproven assertions are repeated even ordinarily rational actors may fail to critique them (and, worse, may repeat them).

For the record, there is no set of data that proves African-Americans and Latinos commit a disproportionate share of the felonies committed in America. This should be patently obvious to anyone who considers the number of felonies that may be committed but never reported or even witnessed by anyone other than the criminal(s) engaging in the act (particularly financial crimes such as insider trading, tax evasion, housing discrimination, lending discrimination/redlining, etc, all of which could result in increased felony convictions if our law enforcement agencies had more resources to combat these ills). You can't determine which group is committing a disproportionate share of any set of acts without empirical data showing the overall number of said acts committed; this is simple logic, the same logic generally applied through much of the book.

Other failures in logic are less startling, such as the decision by the authors (in Chapter 3) to attempt to determine the earning level of every crack dealer in the United States from the income generated by one "franchise" of one Chicago street organization, in a 12 square block area in the South Side of Chicago, over four years. Ordinarily, this would be considered too small of a sample size from which to draw a conclusion. Of course, "experts" regularly comment on what is described as the criminal underclass without relying on the type of factual scrutiny reserved for stories about more "respectable" citizens or sectors of society.

Following this mainstream media tradition, where the crimes happen in "respectable" sectors of society, race is not mentioned by the authors at all. In a chapter entitled "Schoolteachers and Sumo Wrestlers", the authors never once mention the race of any of the employees allegedly stealing bagels sold by salesman Paul Feldman at a rate of roughly 11% (more than one out of every ten bagels!). By contrast, the entire chapter dedicated to drug dealing limits its discussion to black inner city drug dealers on the South Side of Chicago (a predominately black neighborhood) whose race is made clear to the reader because they are repeatedly quoted referring to themselves by the n-word (as if the neighborhood cited didn't already codify their race, since "South Side of Chicago" has been transformed into shorthand for "black and poor" by the Chicago and national media). Even noting the location of the drug dealer story as the South Side of Chicago is irrelevant to the larger point being made (the economics of the drug trade). Interestingly, the researcher who lived with the drug dealers on the South Side is name checked by ethnicity (Indian) upon his introduction, for no apparent reason. Feldman, curiously enough, is not described by race, nor is CPS CEO Arne Duncan. Duncan is white (Google him). Feldman has a German surname. Draw your own conclusions from this limited sample size and paucity of information about how the authors racialize their writing.

Any conclusion you might draw from that sample size and limited data is as credible as the conclusion the authors drew about bagel thieves. When discussing how the criminal occurrence of bagel theft decreased from 13% to 11% after the September 11th attacks (p. 47), the authors pontificate that this drop may have been due to "a patriotic element" or "a more general surge in empathy", although the equally likely scenario is that because "many of Feldman's customers are affiliated with national security", these workers were aware that surveillance of their worksites dramatically increased after September 11th, dissuading potential thieves who feared detection. Why are former criminals (the 2%) without a racial designation denoted in Freakonomics assigned such a beneficial motivation for their decision to cease their stealing ways? Why do the authors fail to identify any of the companies housing these thieves, but the South Side of Chicago, as a locale, is named, when neither location has any correlation to the crime being committed? After all, it would be different the authors had to tell an anecdote about coca leaves, that mainly grow in certain areas, but you can sell crack (or steal bagels) virtually anywhere, so why name one setting but not others?

When the authors make these sort of choices, is there something that can logically be deduced about their political leanings and beliefs?

They did go against the right wing viewpoint about capital punishment, after all. I wonder how they feel about liberals?

Maybe the answer can be found where the authors make a wildly unprovable claim on the bottom of page 125, where they state inexplicably "In some cases, hiring additional police was considered a violation of the era's liberal aesthetic", referring to the time period between 1960 and 1985. This is unprovable for any number of reasons. One, the author cannot cite any source wherein a unified (or any) liberal position against additional police hiring is held. Two, if "liberal aestethic(s)" controlled policy decisions from 1960 to 1985, why did the Vietnam War last so long, since liberals were vehemently opposed to that war? How did the nation's law enforcement apparatus investigate, entrap and/or assassinate so many liberal figures with impunity in this era? Why were Republican presidents in office for the majority of this period (13 years of 25)? In short, what type of inherent political bias could lead the authors to make such a claim without citing to any authority as if it were a truism?

The authors then go on to conclude, without much evidence, that a 50% decrease in police hiring and "leniency in the other half of the criminal justice system", created a "strong positive incentive for criminals", allegedly leading to a spike in crime rates. The authors talk about increased incentive to commit crime due to decreased police hiring, but cite no corresponding drop in arrests to back their claim that reduced hiring "translated into a roughly equal decline in the probability that a given criminal would be caught" (p. 126). Scrupulous with their math in other areas of the book, the authors offer essentially no mathematical data at all to prove this alleged correlation. The first problem with their assertion is, in order to determine probability that a criminal would be caught, the authors would have to know both the total amount of crime committed and the number of criminals arrested for crimes ("caught"). Without citing either data set (total crimes committed or total arrests), the authors cannot possibly believe they have proven their point about the falling probability of arrest for a criminal between 1960-1985. Further weakening their argument, the authors neither cite what the crime rates were from 1960 to 1985, what crime rates were prior to this period, nor whether crime rates remained at some constant level from 1960-1985 (with demographic changes factored in as variables), in order to establish their central thesis. Do they seriously expect readers to believe there was a spike in crime between, say, 1959 and 1960 without any numerical data? Even when discussing an area limited to the state of New York, the authors fail to note the effects of such landmark acts as the Rockefeller drug laws (which greatly increased the penalty for drug possession in New York) in their calculus of determining what affects crime rates, even though the Rockefeller laws were passed in the middle of their mythical supposedly "liberal aesthetic"-ruled period between 1960-1985.

This is not even to discuss the problem of viewing side by side both the authors' acceptance of the unprovable assertion that African-Americans and Latinos commit a disproportionate amount of the felonies in our country and the authors' argument (from pages 115-145) that legalized abortion post-Roe led to the steep drop in crime in the 1990s. Read together, these two ideas seem to imply that killing a sizable proportion of African-American and Latino fetuses before they are born is a good crime prevention tactic. Freakonomics, indeed.

Post Script: Did I mention that the persons identified as black in the book include a crack gang, an abusive lover who knocks a woman's teeth out and an academic failure? How can such a selection of characters amongst a set of 30-plus million people be justified in a non-racialized way? Ah, I neglect to mention an anecdote about Kareem Abdul-Jabaar (not identified by race, but whose race is obviously known to most of the American population). And the naming chapter identifies five more characters as black: a police detective, a felon with three dozen arrests, a black teen girl arrested for "bringing men into the home while her mother was at work" (the sole black female in the book just happens to be sexually promiscuous, a traditional black female stereotype), and an economist who pushes a still unproven theory that disproportionate poor test scores amongst blacks is explained by their fear of "acting white". So, that's a crack gang, an abuser, a long-time felon, a promiscuous black teen girl, an intellectual failure, a police detective, an economist and a basketball player.

Mathematically, except for the inclusion of the economist, the authors succeeded in including almost every black stereotype in their book (with police detective and the economist being the sole exceptions, although the economist is pushing a borderline racist theory and the black police detective is becoming a "positive" stereotypical role as it appears to be one of the main roles a black lead actor can get in a film: cop or criminal). Out of seven persons in the book identified as black people, five are criminals. The one person identified as a Latino is a drug trafficker (Oscar Blandon). Is it mathematically possible that these were the only Black and Latino people in the entire country that the authors could chose to profile, if they had a racially unbiased approach to presenting their information? Or is something else behind their curious selection of persons to profile? You do the math on that one.


liam said...

Great unpicking of the subtle, but insidious ways that racism (and, I think, a range of other kinds of discrimination) filter into accepted mainstream discussion!

The concept of something being repeated so regularly, that we come to assume it as fact is, without any basis, is a really troubling one - often highlighted in the 'things viewers of FOX news believe' studies (i.e. - that Iraq had WMDs).

Having not read Freakonimcs, I can't argue in any detail about the last example you cited, but to summarise the Roe v Wade case as a negative race issue (when I would say it was a critical win for women's rights, of all backgrounds), felt troubling to me.

I'm not sure what the evidence would be, but shifting race/ethnicity aside (if it possible to do so for the moment), is it too much of a jump to suggest that babies born to mother's who, if given the choice, would not have had them, would become more disposed to a range of negative patterns?

Views on abortion are almost always inflammatory to someone, but I felt the last comment was dismissive of the importance Roe v Wade has played in countless women's lives, by (seemingly) viewing it 1st as an attack on unborn Black/Latino children...

...But I may very well have missed something that was said in the book, or misinterpreted your comment...

Thanks for the post!


soul searcher said...

The book spends 30 pages arguing that Roe eliminated a generation of babies most likely to become criminals.

I disagree with the central thesis, that the elimination of those fetuses alone explains the crime drop in the 1990s.

I'm not touching the politics of Roe with a ten foot pole, as I have never been in a discussion where any middle ground is discussed on the issue. What I am saying is, I don't think that one can seriously argue that legalized abortion is the sole cause of the crime rate drop in the 1990s, at least not with substantially more data than is presented by the authors of Freakonomics.

The inclusion of the racist assumption that Blacks and Latinos commit disproportionate amounts of felonies in combination with the subtle suggestion that the elimination of a certain set of fetuses reduces crime leads to a highly inflammatory conclusion, by deductive reasoning: lower Black/Latino birth rates and you will lower crime.