Sunday, June 10, 2012

Identifying Cognitive Pollution

In the prior post,How To Assess a Piece of Writing, Especially Outside One's Expertise, I created a potential framework for assessing the value of an article or piece of writing in terms of how effectively it advances an argument. The framework consists of five categories each category having five elements.
While in the process of creating this framework, I came across Carl Sagan’s Baloney Detection Kit which is also useful for assessing an argument. Sagan’s is probably the better approach for a rough and ready assessment. The framework I have laid out is useful in that it lends itself to quantification and for being able to pinpoint the nature of the weakness in a particular argument.

Time to take the framework for a test run on Trayvon – Killed by an Idea by Zeta Elliott which was forwarded with the claim that it was worth reading. My sense at the time was that it was actually just another piece of cognitive pollution – material regurgitated without being clearly thought through, argued or grounded in factual reality – but I wanted to put some parameters on that instinct. The above framework is a first pass at a means of diagnosing a document to identify whether something is worth reading.

So off we go. While the framework flows in a logical fashion, rarely are articles written in a manner that mimics that flow. What follows is a high level identification of the argument followed by an assessment of the facts pertinent to the argument and rounding up with a review of the logical integrity of the overall argument. As elements of the assessment are addressed, I will notate them at the end of the pertinent paragraph.

On the face of it, Ms. Elliott’s article looks like it ought to be reasonably robust. There are thirteen hot links to external sources of information or opinion. There are four quotes from others. There are several assertions of implied fact. Without considering consistency, integrity, accuracy, etc. (as from above), there would appear to be at least the hallmarks of a reasonably robust article.


The first thing is to determine exactly what is the argument that Ms. Elliott is advancing. It would appear that the basic argument is that black children face “environments and attitudes that are hostile to our children's survival”; that books by black authors “can help to restructure” that environment and those attitudes; that children are put at risk of death by not being able to read books by black authors (“the lack of books for children in our communities is a matter of life and death”); and that black authors are not being published in sufficient numbers because of conscious or unconscious discrimination by the publishing industry (“I appealed to the overwhelmingly white and middle-class editors, marketers, and artistic directors in the hope that they would embrace diversity and work toward achieving true equity”). A further challenge is that it is unclear whether Ms. Elliott is identifying the danger to young black children as arising from society at large or from within the black community. Either interpretation is reasonable but it would appear that Ms. Elliott, by her referencing of Mr. Staples’ editorial, her reference to the contents of her own letter to the publishing industry and her incorporation of the Tucson City controversy over suitability of materials, is of the belief that the danger originates with the broader society rather than within the black community. I read “At this moment in history -- and in this particular political climate -- children of color are under attack. They are the most vulnerable among us, and they fight daily against forces that seek to distort, marginalize, and silence them” as being consistent with this interpretation that Ms. Elliott believes the danger to black children as being from the broader society, particularly given her linking this to the death of Trayvon Martin whom she appears to believe was killed by a white man. For the balance of this analysis, that is the interpretation with which I will work, recognizing that it is not clear as to Ms. Elliott’s intent. (Integrity of argument: Assumptions and definitions – 0, many points of confusion regarding terminology and assumptions).

There are many necessary inferences to string together this argument and it may not properly reflect Ms. Elliott’s actual intended argument. (Integrity of Argument: Clarity – 0, many inferences necessary).

Let’s reformulate the string of sentences for greater clarity and incorporating the inferences that arise from the assumption that Ms. Elliott believes the source of danger is in the larger society.

The argument now might look like: The publishing industry should publish more books by black authors. Books by black authors will be able to change the environment in which black children grow up, in particular by changing negative stereotypes to which the larger society is exposed and is susceptible to internalizing. Black children will be safer for having read books by black authors and majority culture children will have greater tolerance and acceptance of black children by having read books by black authors. The implication is that if the publishing industry publishes more black authors there will be fewer deaths among black children, in particular young black men.

There is an issue regarding ambiguity of terms. What does it mean to “embrace diversity” and what does “true equity” mean? In particular what does this mean in the context of a competitive publishing industry that serves a free-choice market, i.e. publishers have to find books that customers are willing to buy. Diversity takes many forms, in fact as many forms as there are people. Race, ethnicity, age, gender, orientation, culture, religion, educational attainment, familial structure, language, income, class, etc. Should a publisher be focusing on making sure that what they publish adheres to any and all of these criteria of diversity, or is their obligation, as a commercial enterprise, to find and publish those books which the public is willing to invest money in purchasing and time in reading? Clearly by law and practical consideration, publisher’s have to focus on what sells even if they are mindful of the legion of criteria making up diversity. (see Ray Bradbury on proportional representation in storytelling)

Ms. Elliott does not define diversity though she does seem to privilege race as the sole criteria of diversity. She also does not define how one would measure diversity. What is the base of comparison? If African Americans are 13% of the population, how would we know whether diversity had been achieved in publishing? For example, if African Americans are 13% of the population, do we expect to see 13% of all books authored by African Americans? Or is the base to be compared to, that of the percentage of aspiring authors? In other words, if African Americans are only 6% of the population of aspiring authors, should we then expect only 6% of published authors to be African Americans? This is the challenge of comparability. Ms. Elliott is apparently concerned about disparate impact (13% are African American but perhaps only 1.6% of published authors are African American) and is attributing that disparate impact to some fault in the structure of the publishing industry without actually being able to identify precisely what that fault might be. But if, for whatever reason, fewer African Americans aspire to be writers, then we would expect there to be fewer published African American authors. So what does diversity mean to Ms. Elliott? We don’t know though we can make inferences. It seems to be a fairly constricted definition – diversity is defined as African American authors to be published in the same proportion as African Americans are to the general population. (see It’s 2012 already: why is opinion writing still mostly male?: by Erika Fry for a sophisticated discussion of sources of disparate impact.)

What about the term true equity? What does that mean? If the publishing industry is close to perfect competition (as defined in economic terms – no barriers to entry, perfect pricing information transparency, no market dominance, etc.), wouldn’t the books purchased by freely choosing agents (consumers) represent perfect equity? People are able to buy what they want. Who is in a position to determine that they ought to buy more or less fiction, more or less classics versus contemporary, more or less books by authors of a particular ethnicity/religion/race/language/culture/ etc. Is it equity in terms of outcome or equity in terms of opportunity? Here it is virtually impossible to determine what Ms. Elliott means by true equity. It appears that Ms. Elliott believes there to be an unrecognized market for books by black authors and wishes publishers to increase their risk profile by creating supply to that market (increased number of books) without an economic basis for being confident that they will be purchased, i.e. she wants them to increase their chance of losing money above the level at which they are already operating. (Integrity of Argument: Assumptions and Definitions – 0, many points of confusion regarding terminology and assumptions).

Having touched on disparate impact, there is a factual Scylla and Charybdis that we have to clear before we proceed much further. Ms. Elliott’s entire essay rests on two factual assumptions both of which are erroneous.

First (Scylla) is that Ms. Elliott appears to have only read the very first news reports of the Trayvon Martin murder which are widely acknowledged to have been materially inaccurate. She seems to assume that a black youth (Trayvon Martin) was attacked and killed by a white man (George Zimmerman). Ms. Elliott appears to subscribe to Brent Staples hypothesis that “Trayvon Martin was killed by a very old idea”, that idea apparently being that George Zimmerman killed Trayvon Martin because Zimmerman feared Martin as a representative of a stereotype. (Quality of Data: Independence of Source – 0, actual conflict of interest). There are a very strange closing couple of sentences to Ms. Elliott’s article. “Imagine what it might mean if they realized that a broader range of images in books might have created a different set of assumptions in the mind of George Zimmerman (or Anders Breivik). I wish publishers recognized that they have the power to help undo the distortions that are destroying our youth.”

It is unclear what Zimmerman and Breivik have to do with one another and what either has to do with “distortions that are destroying our youth.” The most logical interpretation I can put on this is that Ms. Elliott believes both Zimmerman and Breivik were white racists killing people of color because they hated people of color. This assumption would be consistent with Brent Staples’ essay to which she links. It would appear that Ms. Elliott was not aware that Zimmerman is of mixed white, black and Hispanic heritage, was raised in a mixed race household, resided in a mixed race neighborhood and lived an integrated life which included close social ties with members of various racial groups (See Prelude to a Shooting by Chris Francescani).

It also appears that Ms. Elliott has chosen to believe the initial faulty reports and based on those reports (which characterized Zimmerman as a “white Hispanic”) has then leapt to the conclusion that if Zimmerman was white he must have killed Martin because Martin was black. It is only with this construction that “Trayvon – killed by an idea” begins to make sense; she is assuming Zimmerman carried negative stereotypes of young black men and that he acted on those stereotypes in a fashion that led to Martin’s death. The chain of reasoning is not faulty but is dependent on the facts being correct.

The implication is that Ms. Elliott believes George Zimmerman to be guilty of murdering Trayvon Martin based on racial stereotypes he harbored. It is important to note that Zimmerman is in fact innocent, though charged with second degree murder, and that all the evidence released by the District Attorney’s office so far is consistent with his account of a fatal shooting arising from self-defense and none of the evidence contradicts his account (see New Forensic Evidence Is Consistent With George Zimmerman's Self Defense Claim by Alan Dershowitz). (Integrity of Argument: Accuracy and completeness of data – 0, data is sharply contested and appears inconsistent with Ms. Elliott’s assumptions).

So Ms. Elliott is advancing a single case (Martin-Zimmerman) as an example of the mortal risks presented to black children. (Quality of Research: Sample size – 0, sample size of 1). (Quality of Research: Representativeness of sample – 0, no necessary attributes consistent with Ms. Elliott’s assumptions). (Quality of Research: Counterfactuals – 0, many counterfactuals).

Second (Charybdis), Ms. Elliott believes the level of violence among young black men is misrepresented. Ms. Elliott claims that “Yet for decades, the U.S. media has done precisely that by misrepresenting young black men as violent, predatory criminals”. What constitutes misrepresentation? African-Americans are 13% of the population but are responsible for roughly 50% of murders, the overwhelming preponderance committed by young men (see Who kills who from political Calculations using DOJ Statistics. If you narrow down the ranges by gender and by age, you have about 3% of the population (young and black and male) committing perhaps 40% of the murders. If a stereotype is grounded in a numeric reality, what, in a fair and integrated society, should be done with those numbers? This tragic reality was captured by Jesse Jackson’s plaint in 1993 “There is nothing more painful to me at this stage in my life than to walk down the street and hear footsteps and start thinking about robbery. Then look around and see somebody white and feel relieved”. The DOJ statistics also indicate that in America, mortal crime is substantially intra-racial; whites kill whites (83% intra-racial), blacks kill blacks (90% intra-racial). Finally, in that smaller population of inter-racial killings, a white is more than twice as likely to be killed by a black as a black is by a white. (for stereotypes versus other root causes, see Europe Agrees: Greece Is the Laziest, Most Incompetent Nation in the EU by Derek Thompson: “Europe's problem isn't stereotypes. It's institutions.”).

Whatever one may think of how young black men are represented, it is easy to see why there would be a perception of young black men as being “violent, predatory criminals” to use Ms. Elliott’s words. It is a stereotype grounded in numbers (3% of population, 40% of murders). Who is to blame for that perception? Ms. Elliott believes that the media is responsible. But if the media comes anywhere close to reporting reality (a tenuous premise), then perforce they would be representing young black men as disproportionately violent. (see Michael Crichton on the Gell-Mann Amnesia effect in Why Speculate, starting paragraph 15). It is logically inconsistent of Ms. Elliott of to make the accusation of misrepresentation if what is being represented is the reality. This contradiction is captured in the linked Staples essay as well. He offers as evidence that whites are overvalued in the justice system: “Over the last three decades, a growing body of research has shown that racial stereotypes play a powerful role in judgments made by ostensibly fair-minded people. Killers of whites, for example, are more likely to receive the death penalty than killers of blacks”. But if blacks are a disproportionate percentage of murderers and blacks are a disproportionate percentage of victims, what this means is that a black murderer is less likely to be sentenced to death than a white murderer. So is that valuing black victims less or is it a greater demonstration of clemency to the black perpetrator? If blacks and whites were to be proportionately sentenced to death, there would be a much larger disproportionate number of black men on death row than there are already. (Integrity of Argument: Logical consistency – 0, many inconsistencies).

Does Ms. Elliott advance any evidence that blacks are at disproportionate risk in general and disproportionate risk from white violence in particular? No. She is offering a single, widely misreported case as the basis for her argument, the facts of which are inconsistent with her thesis. When we look at the DOJ statistics referenced above, we see that 90% of black deaths are at the hands of other blacks and only 8% are from white murderers. Not only does she not advance any data to support her thesis other than a faulty narrative around one tragic encounter, the data that is available refutes her predicate proposition. If you want to address the real dangers to blacks, it has to be done by focusing on violence in the black community. (Quality of Data: Completeness – 0, key premises not supported by sourced data). (Integrity of Argument: Relevant support – 0, no data or authorities cited).

To make it more explicit, were one to manage in some fashion to completely eliminate all murders of blacks by whites, one would save 230 black lives. On the other hand, if one were able to reduce black violence to their proportion of the population (50% down to 13%), you would save 2,040 black lives and 380 white lives. 230 black lives is not to be sneezed at but the real prize is 2,420 lives – a ten-fold increase in lives saved. By focusing on mostly marginal levels of white violence, Ms. Elliott diverts attention from the real opportunity for saving lives.

To be fair to Ms. Elliott, individuals in general have a difficult time estimating demographics. (see Americans Have No Idea How Few Gay People There Are by Garance Franke-Ruta and Americans Stumble on Math of Big Issues by Carl Bialik). The issue of overestimating minority groups (of whatever category) is not unique to the US by any means. However, if one is going to opine with an objective of swaying opinions and influencing the actions of others, there is a moral obligation to be confident in one’s facts and for that confidence to be based on some objective grounding.

So two critical premises to Ms. Elliott’s argument are demonstrably false: 1) This is not a white on black killing, and 2) The proportion of murders committed in the US are disproportionately by black perpetrators and with black victims. Technically, we don’t need to pursue her article further – it is already fatally flawed. However, it is worthwhile to test the assessment framework further.


Clearly Ms. Elliott believes there is systemic and materially impactful negative discrimination against African Americans in the US. That is comprehendible if one does not have a grounding in statistics and the mathematical/sociological nature of measuring outcomes. Many of these issues of interpretation hinge on one’s personal understanding of sources of disparate impact. If one group, by whatever designation (gender, race, religion, income, education level, etc.) is not proportionately represented in some set of outcomes or field of endeavor, is it reasonable to assume that the disparate impact is solely or primarily due to conscious or unconscious discrimination/bias or can it be attributed to other factors? Disparate impact shows up in all sorts of ways: some making good sense (tall people are disproportionately overrepresented in basketball), and some being more mysterious (academia self-identifies as liberal at a rate more than twice that of the population at large; 70-80% in academia versus 20-30% in the at-large population). There is no doubt that within the US and across the globe there is a world of bias and active discrimination. Some bias is driven by factual realities and some is purely based on speculation or assumption. But is bias responsible for the bulk of the disparate impacts seen in practice? There are many reasons both in the US and internationally to ascribe disparate outcomes to other factors, not least being cultural values (see Thomas Sowell’s work for extensive US and international data-based comparisons; The Economics and Politics of Race; Ethnic America; Affirmative Action Around the World; Economic Facts and Fallacies; Race and Culture; Migrations and Cultures; and Conquests and Cultures, etc.) (Refutability: Measured Phenomenon – 0, no measurements). (see also Daniel Kahneman in Thinking, Fast and Slow for a discussion of the importance and necessity of stereotypes, assumptions and heuristics in decision-making.)

This issue of disparate impact and assumptions is not dissimilar to the reasonably settled issue of comparable pay based on gender. Thirty years ago a storm was ignited after early studies revealed that across the OECD, women made only 60-70% as much compensation as men. This was taken as clear evidence of compensation discrimination. This suited various policy agendas and was consonant with various sociological assumptions. However, thirty years of research across the OECD have demonstrated quite conclusively that, when you take into account critical factors such as age, duration of working career, hours worked per year, marriage status, children, continuity of career, educational attainment, career choice, etc. the disparate impact virtually disappears. Depending on country, after accounting for the five most common sources of difference, the gap either does not exist or is between 0-5% potential disparate impact. The more variables of difference that are included, the smaller becomes the gap. In other words, when comparing apples and oranges, it was easy to see a pattern of discrimination in the data. Once the data is normalized so that one is comparing apples-to-apples, the gap is usually non-existent or miniscule. (see Why Women Make Less Than Men by Kay Hymowitz, see An Analysis of Reasons for the Disparity in Wages Between Men and Women from CONSAD Research Corp for the Department of Labor).

In this instance, were Ms. Elliott’s assumptions correct regarding “this moment in history -- and in this particular political climate -- children of color are under attack. They are the most vulnerable among us, and they fight daily against forces that seek to distort, marginalize, and silence them”, then we would expect to see similar negative disparate impacts across all people of color. In terms of education attainment, income, health, wealth accumulation, etc. that is not true at all. There are distinctively different levels of attainment between different ethnic and racial groups, some exceeding those of white Americans. Even within the black community, there are perturbations in impact that cannot be sourced to discrimination. Blacks of mixed race parentage (one parent white and one black) and children of African or West Indian immigrants represent something on the order of 9% of the general black population but are 40-70% of the black student body at the most competitive universities (see Top Colleges Take More Blacks, but Which Ones? by Sara Rimer and Karen W. Arenson). Given these numbers, the “forces that seek to distort, marginalize, and silence them” cannot be operating against blackness per se. It is logical to look for other root causes of disparate outcomes rather than only race discrimination. Ockham’s Razor suggests something more like individual behaviors, values, decision-making, usually grouped under the rubric of culture. (Refutability: Inferences made explicit – 0, None).

Having done a radar sweep of the article and identified some fairly significant issues of argument formulation and of fact, let us now return to the key elements of Ms. Elliott’s argument. 1) The publishing industry should publish more books by black authors. 2) Books by black authors will be able to change the environment in which black children grow up, in particular by 3) changing negative stereotypes to which the larger society is exposed and is susceptible to internalizing. 4) Black children will be safer for having read books by black authors and 5) majority culture children will have greater tolerance and acceptance of black children by having read books by black authors. The implication is that 6) if the publishing industry publishes more black authors there will be fewer deaths among black children, in particular young black men.

If these are the critical links in her argument, does she advance support for any of them, and even if she does not, is there reason to believe each of those assertions to be true. In fact, there is no data advanced to support any of the six critical steps in her argument. (Refutability: Critical path identified – 0, unclear as to which elements of argument are critical).

The publishing industry should publish more books by black authors – This is a statement of opinion and not an argument. It is Ms. Elliott’s conclusion and it only has relevance in the context of the supporting predicate arguments.

Books by black authors will be able to change the environment in which black children grow up – There is no evidence advanced to support this most critical proposition. There are several variant propositions that have at least some evidentiary backing but in general the robustness of data around the influence of books and reading is ambiguous at best. It is well established that children from homes in which there are many books and in which adults are seen reading are more likely to become enthusiastic readers and that these attributes are also correlated with higher educational attainment, higher income potential, greater tolerance, etc. Correlation is not causation and it is extraordinarily difficult to untangle cause from effect. None-the-less, the unstated predicate in this line of thinking is children growing up in traditional family structures. Again with Ockham’s Razor. If 66% of children are growing up in single parent homes, they are less likely to have the environment in which reading is valued or even optional (see Kids Count Data Center). Having more books by black authors won’t make a difference if there is no audience to buy them (too poor) or read them (too busy putting food on the table). (see Growing Up With a Single Parent: What Hurts, What Helps by Sara McLanahan, Gary D. Sandefur).

There are innumerable individual and anecdotal testimonials regarding how particular titles read by particular children at particular times can have seminal and transformational impacts. There is no study or data that sheds any light on how to identify which books will be seminal to a child at a particular time. In addition, there is virtually no research or data available regarding the long term impact of reading by race vis-à-vis author and reader. The very lack of data in this arena poses some interesting research ideas. For example: is there a difference and what are those differences between the literature read by mixed race/African immigrant/West Indian immigrants, American born African-Americans and whites? And are those differences significant in any meaningful way?

The upshot though is that there is no evidence to support the assumption that the race of an author has any material impact on the environment in which a child is raised. It would appear as likely that it is entirely serendipitous which books will have the greatest impact as it is to argue that there is a pattern of attributes which make some books more meaningful (such as being written by black authors). However, this is simply a matter of being at the knowledge frontier. There is no evidence. There is, however, evidence that culture, behaviors, income, educational attainment of parents, etc. are far better predictors of educational and other life outcomes (as well as likelihood of death by violence).

Books by black authors will facilitate changing negative stereotypes to which the larger society is exposed and is susceptible to internalizing – No evidence advanced that books have a material impact on shaping stereotypes or that the race of the author makes any difference in that process. There can be little doubt that under the right circumstances, books can be profoundly influential. In terms of the long road towards civil rights, where might we be without Fredrick Douglass’ Narrative of the Life of Frederick Douglass, an American Slave, Harriet Beecher Stowe’s Uncle Tom’s Cabin, and Harper Lee’s To Kill a Mockingbird. But again there is the profound issue of cause and effect. Did these books help change the direction of events or did they become popular/acceptable because events had already changed and the public was ready for them? The proposition can be tested by a simple two-part thought experiment. Is there a single example where someone has clearly and apparently changed the course of history by deliberately writing a book with the intention that it should bring about a certain outcome? I think the answer has to be yes. Uncle Tom’s Cabin, Silent Spring, The Jungle, Common Sense, 1984, Brave New World, Walden, all would likely meet those criterion and there are many more which could be identified. The second question is the kicker – How many books have been written with the intention of changing people’s perspectives on some particular issue and have not done so? A gargantuan number. Go to any used bookstore and look through the number of books that have a variant of the blurb promising that “it will change how you think about . . . “ and yet which were out-of-print within five years. Critics are always identifying new modern day classics but rarely are their predictions fulfilled. (see Why Experts Get it Wrong by Jim Manzi).

So yes, books can be hugely influential but which books under what circumstances is virtually impossible to forecast. Since book purchases and book reading (representing an expenditure of time) are substantially a voluntary activity, any message that is intended to be transmitted will only likely find an audience that is already predisposed to that message. Can books by black authors (as opposed to any author) help facilitate changing negative stereotypes? Possibly, but there is no data to support that and there is little logical basis to believe that that is a real phenomenon.

Black children will be safer for having read books by black authors – Again, no evidence and no particular logical grounds to believe that that might be the case. Violence is tied to many known and suspected causative factors (see Steven Pinker's The Better Angel’s of Our Nature for a discussion of causes and trends in violence). Reading in general and reading by race of author are not known factors that improve the environment of violence in which children are raised. There are no good reasons to believe this assertion to be true and many reasons to believe that there are other far more important factors behind violence. (Refutability: Explicit forecast – 1, forecast is unmeasured)

Majority culture children will have greater tolerance and acceptance of black children by having read books by black authors – Ms. Elliott does not present it, but at least in this area, there is some research even though it is mixed and ambiguous. Exposure to alternate contexts or unfamiliar ideas can lead to greater acceptance or empathy, at least as indicated by small scale studies and for short time periods. However, the effect does not seem to be dependent on race of author as opposed to the topic, issues, and structure of the story. In addition, it is unclear that acceptance of black children by the majority culture is the issue (see reference above to prevalence of mixed-race and black immigrant groups) as opposed to acceptance of different cultural values. Here, the evidence is more challenging with studies indicating that exposure to contra-cultural values (cultural values in conflict with established values) can exacerbate views. Again this has nothing to do with race of author.

Interestingly, there are numerous studies linking tolerance to economic development and more directly, tolerance as an outcome of trust. Countries with higher reported trust levels and greater cultural homogeneity that incorporates trust also tend to have greater economic growth as well as more generous welfare systems. Trust leads to both tolerance and generosity. (See Does Economic Freedom Foster Tolerance? by Niclas Berggren and Therese Nilsson, and Trust, Social Capital and Economic Development by Patrick Francois and Jan Zabojnik).

If the publishing industry publishes more black authors there will be fewer deaths among black children, in particular young black men – Independent of the logic and the data regarding the linkage between reading and violence, there is the question of supply and demand. Ms. Elliott seems to assume that there is a recognizable shortage of books by black authors, i.e. that there is greater demand than there is supply and that the publishing industry is somehow responsible for that shortage of supply. There is no corroborative evidence that there is a market that is underserved. There is much evidence that other factors are far more critical to mortality than the number of books being published by African American authors.

Ms. Elliott cites the data collected from the Cooperative Children’s Book Center which indicates that of the 5,000 children’s books published each year, only 79 (in 2011) or 1.6% were by African American authors (see Children's Books by and about People of Color Published in the United States) as opposed to African Americans being 13% of the general population. It is somewhat difficult to know what to make of this information. Average new titles published each year usually ranges from 25,000-35,000 according to Library and Book Trade Almanac™. The gap between CCBC’s estimate of 5,000 books published each year and LBTA’s reported 26,522 (for the most recent year 2009) is material so it is not possible to be confident that the 1.6% figure is meaningful. Regardless of what the actual numbers might be, and taking as stipulated that African American authors are underrepresented, the question becomes one of why there is such an underrepresentation. It is interesting to note Asians are 5% of the general population but have the same ratio of 1.6% of author’s per the CCBC numbers. Still underrepresented but not nearly to the same degree. This goes back to the issue of disparate impact – what causes disparate impact and can it be attributed solely or substantially to bias or discrimination? (Quality of Data: Transparency of collection – 0, no methodology available to reconcile contradictions).

Brent Staples and Ms. Elliott both seem to subscribe to the idea that the discrimination is not necessarily conscious but that it is an unconscious discrimination based on stereotypes. We have already discussed the challenge that arises if the stereotype is warranted and meaningfully useful. But what of this argument of unconscious discrimination. It is irrefutable that all decisions are made on some mix of conscious and unconscious decision-making and that in general, decisions are arrived at unconsciously and then retroactively justified consciously. (see How We Decide by Jonah Lehrer). While all that may be true, it offers no capacity to determine whether a decision was arrived at irrationally. Unconscious bias as an explanation of disparate outcomes is untestable and unfalsifiable. If people shy away from hoody wearing young black males, is it because they are consciously biased against blacks in general or because they unconsciously associate young black males with violence, or because they accurately associate young black males with higher rates of violence? Claiming unconscious bias is an easy explanation that can be applied to any socially constructed phenomenon but it is not a rigorous explanation. Staples and Elliott offer no information to refute the assumption that disparate decisional outcomes may be based on many other factors than negative and false bias. (Refutability: Forecast, useful, reliable, and non-obvious – 0, No).

In an industry that is immensely competitive, where profit margins are small, where there are massive numbers of participants and where the barriers to entry are low, it would be surprising to discover that there were material unmet needs in terms of any type of book whether it is by topic, style, issue, protagonist demographics, author demographics, etc. It is also characteristic of the industry that there is little predictability in terms of which books will catch on and which will fail. It is also the case that the market for books is crowded with backlist titles. 50-70% of books sold in a year are backlist titles. The reality is that any new author is competing not only with other new authors, not only with all other sources of information/entertainment, but also with all writers who have gone before – that’s a lot of competition. (see This Graph Is Disastrous for Print and Great for Facebook—or the Opposite! by Derek Thompson for a highlight of the small percentage of time spent reading)

In order for Ms. Elliott’s explanation to hold water, we would have to believe that mainstream publishers as well as specialty publishers (African American publishers, multicultural publishers, self-publishers, etc.) are collectively and uniformly failing to recognize that there is a commercial demand for books by African American authors and that their blindness is due to unexplained unconscious bias. Ockham’s Razor would suggest any of several alternative explanations would be more likely.

Given that the CCBC itself indicates that there are increasing numbers of ethnic/multicultural oriented independent publishers, including African American publishers, who presumably are both better able to identify and respond to targeted market demands and who also are presumably more receptive to African American authors, it is difficult to reconcile those trends with the argument that there is an untapped demand for books by African American authors. (see CCBC Small Presses of Color). It would appear that there is a low demand but one that may be growing and as it grows, the supply is growing with it. With these market characteristics (competitive, low barriers to entry, supply greatly exceeding demand, etc.), is it more likely that the problem is some institutional characteristics of the publishing industry creating barriers to African American authors or is it more likely that there is low demand for the particular works by particular African American authors? I would guess that the issue is too low demand rather than too low supply. Hard numbers are not available but anecdotal estimates are that the publishing industry pulps 20-40% of all books printed; clear evidence, if accurate, that production exceeds demand by material amounts.

If this is the case, then one has to do a root cause analysis for why there is too low demand for the work products of African American authors. Possible candidates might be, in no particular order: low general public interest in topics African American authors wish to write about, differences in writing style, inexperience of authors, indifference on the part of the public as to authorial demographics (they don’t care about the race of the writer), low or ineffective marketing efforts, etc.

This concludes an exploration of Ms. Elliott’s article using the new proposed framework. While it has taken a long time to write, it actually only took about ten minutes to assess informally.

The upshot of applying this new framework for bringing greater critical rigor and more objectivity to reviewing an article is that Ms. Elliott advances no evidence to support any of the critical assumptions that she makes. Independent of whether she deploys evidence in her article, in fact for each of her critical assumptions there is substantial data against the assumption (Books by black authors will be able to change the environment in which black children grow up; Books by black authors will facilitate changing negative stereotypes) or simply no evidence to support (Black children will be safer for having read books by black authors) or ambiguous evidence (Majority culture children will have greater tolerance).

The reason for characterizing this article as cognitive pollution is manifold.
• The author fails to explicitly establish the nature of her argument or the critical elements supporting it, leaving it up to the reader to infer much and fill in the blanks.
• The author advances a self-serving recommendation (she is a black author and she wants publishers to publish more books by black authors).
• The author justifies this by making an argument based on a recent tragedy without actually understanding the details of the tragedy.
• The author does not generate alternate hypotheses and test them. For example the author never addresses why publishers might not be publishing more books by black authors (Ockham’s Razor – low demand)
• The author does not engage in any sort of root cause analysis (why are there so few published black authors?)
• The author is insufficiently confident in her argument to make any concrete forecasts or predictions.
• There is no linkage of causes to outcomes.
• The author indulges in subtle ad hominem attacks without linking accusation with evidence. She implies that the disproportionate presence of whites in the publishing industry predisposes them against publishing books by black authors because of biases or prejudices but without providing any evidence that anything other than standard commercial judgment is at play.
• The author does not appear to be aware of basic crime statistics which refute her position.
• The author nowhere advances any reputable research or data to support her position in whole or in part.
Out of a total of 50 possible points, she manages in a 900 word article to amass a single point (there is an implied, if not measured, forecast that more reading of books by black authors will lead to a reduction in deaths among black children.).

So there you have the first trial run. Not a pretty picture. The outcome was so bad that I wondered whether the flaw was in the framework or in the article. I subsequently applied the framework to a couple of other argumentative articles to see if the outcomes would be materially different. China’s ‘Bad Emperor’ Problem by Francis Fukuyama; Thank God for the Atom Bomb by Paul Fussell; and a couple of others. The numbers are much higher: 14-32. I will play around with this for awhile. So far it appears to be useful even if the numbers are a little lower than I would expect. I suspect that that is a function of the first samples being shorter newspaper and magazine length articles. I will try it out on a couple of research type articles and longer magazine articles and see if the numbers don’t go up somewhat.

What I do like is that, using it as a check list, it is easy to get a first pass assessment pretty quickly. If you’ve just read the article and it is fresh in mind, you can basically give it a score within thirty seconds. It is also a good prompt: One article that I thought to be very good, I realized was only a matter of my agreeing with the author, not that the author had made a particularly good case.

So back to the original mystery, why did a handful of intelligent individuals pass along this article with the endorsement that it was worth reading when it is patently not worth spending time reading. I am guessing that there are possibly three or four causes, but these are pure guesses.
• The referrers are friends with the article’s author and they have allowed friendship to blind them to quality.
• The referrers agree with the conclusion (publishers should publish more black authors) independent of the argument, i.e. they are endorsing the conclusion not the article per se.
• The referrers responded to the truthiness of the article (quotes and links) without actually considering the truth of the article.
• The referrers failed to recognize that this article was not actually an argument but was essentially an advocacy essay dressed up as an argument.
Having invested the time in rigorous review of the quality of the argument, it does expose the danger of sloppy thinking and writing but also the opportunity to redirect attention to more productive lines of action. If Ms. Elliott is concerned about violence against black children, then the analysis redirects attention to the primary source of such violence (black-on-black deaths) rather than the marginal issue of white-on-black murders. If Ms. Elliott is concerned about low rates of publication of black authors, then the analysis suggests that there are alternative sources of that low rate which might have nothing to do with unconscious bias such as more effective marketing, better topic selection, better writing, increasing generic demand for reading materials, etc.


1 comment:

  1. Thank you for a very interesting post, which I admit I am still digesting. Simple fact checking is always of tremendous importance, and so frequently neglected; confirmation bias hits us all.

    ReplyDelete