Yesterday was a numbers day. When I first went out, I went to the bank with an installer to whom I had given a cheque that bounced. I had deposited a money order – that alone shows that I belong to an older obsolete age – from another account in another bank to cover the amount of the cheque to the service company. I did not know that banks could or would hold off certifying a deposited money order because I thought that a bank money order was the equivalent of cash. I learned that I should have just taken cash out of one account in one bank and deposited it in the other; after all, the banks were directly across the street from one another. For I was wrong. Banks can hold back crediting money orders to your account. Instead of cash, I could also have obtained a cashier’s cheque or implemented a direct electronic transfer.

That chore resolved, I then went to the dentist to have a crown put on one tooth. Talk about numbers and dollars!

I had a time gap where it did not pay to go home because I was going on to hear the keynote speaker for the Walter Gordon Symposium that I planned to attend the next day (today) on: “Making Policy Count: The Social Implications of Data-Driven Decision-Making.” The subject of the keynote address was, “The Ethics of Counting.” The presenter was Professor Deborah Stone. In the interval between the dentist appointment and the lecture, I was reading the 26 March 2018 issue of The New Yorker and, as I sat in the auditorium waiting for the lecture to begin, totally coincidentally, I was nearing the end of the magazine and was reading the section on “The Critics.” It was an essay called, “The Shorebird: Rachel Carson and the rising of the seas.” The writer was Jill Lepore whom I had gone to hear deliver the three Priestley lectures the week before on, respectively, “Facts,” “Numbers,” and “Data” and about whom I have already written extensively.

As we all know, Rachel Carson’s book, Silent Spring (1962), first published as a three-part series in The New Yorker, alone is credited with launching the environmental movement. Jill Lepore took a different tack. Though mentioning the revolution in science and policy of correlating data on the use of DDT and the disappearance of birds, the focus of Lepore’s essay began with Carson’s personal biography and her lyrical writing about birds, fish, shad and the sea. Why? Because Sandra Steingraber, editor of a collection of essays called, Silent Spring and Other Writings on the Environment, had omitted any reference to that lyrical oeuvre because, though sometimes alluding to environmental threats, those essays failed to call for any specific social action. Lepore was determined to balance the books in her review essay for, as she claimed, Carson could not have written Silent Spring unless she had clambered down rocks and waded in tidal pools and written about what she saw and studied. For her earlier books were not just about molluscs or turtles or, a major concern, shad, or about kingfishers and redstarts, but about placing those creatures within an environmental context. Those earlier books, The Sea Around Us and Under the Sea-Wind became national best-sellers.

Those studies and writings led Rachel Carson to question government policy and the practice of eliminating “career men of long experience and high professional competence and their replacement by political appointees.” There seemed to be some correlation, not only between DDT and aerial spraying and the death of species, but between the emerging practice of dealing with social problems through the lens of power politics rather than the microscopic analyses of the skilled work of the products of The Sanctuary of Method. The mistreatment of the natural environment and of the research environment had similar roots, a concern with exploitation rather than exploration and understanding as we find ourselves located “in an instant of time that is mine…determined by our place in the stream of time and in the long rhythms of the sea.” Very soon after the publication of Silent Spring, Rachel Carson died of cancer before she could write a new envisioned book on the rising and warming of the oceans.

Deborah Stone’s most famous book is her classic study, Policy Paradox: The Art of Political Decision Making. Her lecture on counting was intended to introduce those attending to the question of how to build policy in a data-driven, more than simply a numbers-driven, world, a world of proprietary and indecipherable algorithms and not just numerical correlations. For an earlier stage in the stream of intellectual time, a key issue, which Stone played a significant part in unpacking, was the hidden assumptions and built-in norms behind the statistical evidence and correlations used to produce policy. In a previous blog, I had offered a simple narrative example of the time I got on the university pension committee to question the use of the gender category to doll out different pensions to women than men. Based on such false categorization, Blacks and handicapped professors should get higher pensions.

Other works have driven home similar points: Michael Wheeler’s (1976) Lies, Damned Lies and Statistics: The Manipulation of Public Opinion in the United States. The clever phrasing allegedly went back to Mark Twain who viewed statistics as the greatest source of lies for he had lived in the nineteenth century rather than at the end of the twentieth when data-driven analyses prevailed and superseded statistics in that accusation. In history, however, the reference was initially made in the context of allocating pensions in 1891 in Britain. A more recent work, Cathy O’Neil’s Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy (2016), carries the argument forward into a data rather than simply statistical-driven age. Mathematical algorithms can be tweaked and formulated to serve interests and power as she illustrated the effects on the financial crisis of 2007-08.

In yesterday’s Washington Post, I read an article on how polling itself – who is ahead and who is behind – influences voting patterns. Reporting that Hillary Clinton was highly favoured to win, rather than data of the percentage of the vote she would likely get, tended to decrease the incentive for supporters to go out and vote. However, Deborah Stone was dealing with an earlier version of such distortions, with numbers and statistics rather than data and algorithms, for the latter are ethically charged models built into the sophisticated mathematics.

Deborah Stone focused on a more fundamental problem characteristic of the transition from the Sanctuary of Method to the Social Service Station in which symbol and numbers were tied to causes and interests depending on the categories used. The latter led to interpretations and decisions dependent often on the negative or positive connotation of the category. Stone in her lecture went back to basics. We can learn to count by focusing only on identicals or by focusing on differences united by a single category, such as counting different kinds of cookies and not just identical glasses of milk. Counting is, thus, not just about identicals, but about categorizing what is different as an identical. In the case of the pension issue that I discussed, instead of treating all professors as equals, they were divided by gender to allocate pensions. In the name of distributive justice, namely that women retirees needed the same money each year as male retirees, such a principle of distribution was unethical.

Deborah offered a ream of illustrations of such a misuse of statistics that led to and supported unjust policies. In collecting numbers on violence against women, the collection depended upon what was classified as violence, who did the counting and for what purpose. For example, did relegating a second wife and child to a small room in the back of the house, expulsion from the house as a form of punishment, rebukes for giving birth to female babies, count as violence as Bangladeshi women contended? Or were European and North American models of violence predominant in the counting. Think before counting was one mantra. Take into consideration the language and concern of those counted was another. Always take into consideration what people wanted to accomplish by collecting such statistics. For numbers carry clout.

Interestingly, Stone referred, but in greater detail, to the same illustration that Lepore used in her lecture, the three-fifths rule for counting slaves built into the American constitution by James Madison in an early attempt to reconcile the paradox that slaves were, on the one hand, property that could be bought and sold, and were, on the other hand, sentient human beings who were held accountable and punishable for their actions. Tax policies and the distribution of votes depended on how slaves were counted.

Numbers count, whether referring to the numbers attending President Trump’s inauguration or to back whether you should take Lipitor to deal with your cholesterol level. Do we ask questions whether you believe immigrants take your jobs in undertaking a survey, or do you ask whether they contribute to create jobs by starting businesses?

Let me take up both issues of the application of statistics and their creation. On the recommendation of my heart specialist, I use Lipitor, the brand name of Pfizer Pharmaceutical that has earned the company $130 billion in sales since the drug was approved for human use in 1996, to lower my cholesterol level and, therefore, to introduce a preventive measure against blood clots. (I once developed a 2.5 inch-long blood clot in a leg vein that went just above my knee.) This in turn would reduce the risk of a heart attack and stroke by lowering plaque build-up in my veins. I have never investigated the categories or methods used in the research behind the drug. I take the drug based on the authority of my physician.

However, when you disaggregate the issue of cholesterol, you find there are different types, some “good” cholesterol and some “bad” – low density lipoproteins (LDL). Further, based on research paid for by the drug companies, what counts as a high cholesterol level has been gradually lowered over the years to the great benefit of the bottom line of Pfizer. Given associated risks – to kidneys and liver, to diabetes and muscle diseases, as Lipitor, a statin, reduces the amount of cholesterol made by and stored in the liver – the lecture implied that research funded by Pfizer based on its economic interests should be questioned.

It was clear that Deborah Stone did not favour collecting stats based on supply and demand and she was sceptical about stats collected by economic interests or those interested in perpetuating their political power. Good stats should be based on building a community and social well-being, on fostering empathy and minimizing exploitation. As the lecture progressed on the ethics of numbers, it became clear that Stone was not just interested in issues, where injustice was perpetuated by the use of statistics, but was positively selling an alternative ethic as the basis for statistical analysis. She was a bleeding heart rather than a possessive individualist. She wanted statistics that fostered empathy and undermined the use and abuse of some people by others. Categories used in statistics can and are used to change hearts and minds – though other stats that she collected indicated that prior prejudices meant that information did not work in changing hearts and minds since biases are almost immune to change by numbers. This was readily apparent in a CBC radio show yesterday on the introduction of a cap-and-trade tax on carbon to combat environmental degradation; a Progressive party defender of the tax dealt with calls, mostly by conservatives, who opposed the tax. Statistics were central to the argument but seemed useless in getting anyone to change their mind.

What Stone did not do was disaggregate areas in which numbers were collected ostensibly to foster care and concern for the displaced resulting in a very different origin of distortion. I had an occasion to audit statistics on those made homeless by the Israeli invasion of Lebanon in 1982. Originally, I went to undertake an actual count, but upon arrival in Lebanon during the war, I had found that there had been twelve different counts of those made homeless, so I simply performed an audit rather than a count. The whole project was stimulated by competing numbers. The Israeli government had issued a report that 27,000 Palestinians had been made homeless by the invasion. OXFAM Britain had published full page ads that 600,000 had been made homeless. The discrepancy was too huge to ignore for a research unit determined to establish objective and accurate figures in dealing with refugees.

As it turned out, the original figure of 600,000 was produced by the International Red Cross, but it was not of those made homeless, but of “those affected” by the invasion. OXFAM Britain had switched the stat to refer to a very different category. Further, of the twelve counts on the ground, all were carried out very objectively with an intention of producing accurate figures. The Israeli figures were too low (40,000 Palestinians had been made homeless in southern Lebanon.) The corrected figure of 40,000 rather than the original Israeli figure of 27,000 was more accurate because the Israeli figure was a product of an arithmetical error combined with missing some enclaves where the displaced had taken shelter.

The most thorough count was undertaken by the Palestinian school teachers who wrote down every name of every person who had lost their homes in typical elementary school ledgers. The figure arrived at was considered too high by about 10% because Palestinians whose homes had been destroyed had been counted even when they had not lived in those homes for years and instead rented them out to others, mostly Bangladeshi itinerant workers. None of the other counts had considered that these Bangladeshis had been made homeless by the war, a bias not only of both sides, but of the humanitarian international community.

Using measures to arrive at a common definition, the city engineers’ counts and all the others could all be reconciled to result in a common figure. The interesting irony was that the tool based on the “worst” systematic method, that of the International Red Cross, which arrived at its figure by counting kitchenware packages that had been distributed and multiplying by three, turned out to be the most accurate even though the IRC was clearly ashamed of using such a rough tool to determine the result.

I want to illustrate two points by this story. First, not only can private economic interests or political power interests produce distorted statistics, but so can the collection of statistics motivated by empathy and bleeding hearts. Second, statistics can and do provide objective information based on agreed categories and even different methods of collection and analyses. When the ethics of counting closely correlated with the Sanctuary of Method as a fundamental methodological tool is distorted for social purposes, either for profit, for power or even for humanitarian purposes, that is, for solving a specific set of social problems, the determination of the problem and the bias of a belief in correcting the problem can produce distortions by the use and abuse of categories and the resultant numbers.

I do not have the time and space to illustrate other more serious cases – the count of the alleged numbers killed in the Democratic Republic of the Congo in 1996 based on a distortion of the base reference figure that fed a narrative of a second genocide, this time against Hutu rather than Tutsi from Rwanda. For years, until corrected by scholars from both sides, the original figure of the numbers of Palestinians uprooted from their homes in 1948 varied from 520,000 (the standard Israeli figure) and 940,000, the UNRWA figure. Later systematic analysis resulted in a figure of 720,000-740,000 which became an objective reference number for both sides. Objective stats can be collected even in war zones when conflict provided agendas are bracketed and systematic means are used to critique categories and correct for errors.

Stats in themselves are not corrupting, but when we begin to suggest that they be collected to solve a social problem in one direction, say for profit or power, rather than another – enhance aid for refugees or enhance compassion for them – then subjectivity begins to displace objectivity as the critical category and the Sanctuary of Method is undermined as an institutional norm in favour of the Social Service Station. Should the latter be used to enhance wealth accumulation in society or for fostering social justice? For stats are not just correlated with power, as Lepore contended, or with economic interests and power, as Stone contended, but to enhance humanitarian causes. The presumption of subjective bias is partly responsible for the expansion of the idea of post-truth.

To be continued