How Power Begets Power in Statistics
21 September 2020
We rely on statistics as a yardstick for good decisions based on their supposed objectivity. Statistics now play an ever more integral part in shaping our lives by influencing decision-making at every level, from government policy formulation to business analytics, and even down to daily mundane decisions. Its ubiquity raises concerns over the power wielded by those involved in its production process, specifically the amount of information they can obtain and how it can be used for manipulation. Seeing how much of a driving force statistics can be in our lives, to the extent of shifting democratic sentiments as in the case of Cambridge Analytica, it has never been more important to understand how statistics are used as a tool of power and not simply harmless objective aids. In his book Trust in Numbers, scientific historian Theodore M. Porter (1995) explains how statistics have been intertwined with the exercise of power since its conception. I define power is the ability to influence events and outcomes. To understand how power operates in statistics, we must not only examine how power is exerted but also what is needed for exertion in the first place. This essay argues that the integral process of categorisation in statistics exerts power by shaping identities, and that power itself is needed to produce statistics as it influences how a statistic should be counted.
Statisticians can exert power by creating categories which end up defining the identities of those being categorised. Categories are integral to process of collecting and processing numbers in the production of quantitative statistics (p. 42). To quantify people in meaningful ways, statisticians invent labels and descriptors to highlight distinctions between individuals and cluster them into groups with common denominators. The Napoleonic French Bureau de Statistique even went as far as to get “local authorities to introduce new categories” (p. 36) as they saw fit so that useful information may be obtained from counting their populations. The problem lies in that these categories might not describe us finely. Individuals are polarised into distinct man-made categories, when reality is not as clear cut. We fall more along a spectrum and parts of our descriptors are smudged by statisticians with this generalisation.
This smudge would otherwise be harmless if categories were just what they were. Power is exerted when these seemingly contingent categories become real and shape our identities. Every person that discusses and uses a statistic lends legitimacy to its categories, which gradually become more official. “Having become official, then, they become increasingly real” and “form the basis for individual and collective identity” (p. 42), as all of us incorporate them into the perspectives and judgements that affect our decision making. One example he provides is of cadres in France. While the French public was unacquainted with the term in the 1930s, it was made a category in official government post-war statistics to count middle-class engineers and managers. In 10 short years, a new cadre social class emerged with their characteristic way of thinking, dressing and reading (p. 42). The smudge in our identity, mentioned in the previous paragraph, is then absorbed into our identity as it alters our behaviour. Therefore, those who define categories in the production of statistics will have the power to influence the way people behave and live by dictating what they should relate to. This power is bolstered by the fact that it may be hard for outsiders to challenge the choice of categories. Those wanting to discuss statistical results obtained from such categorisation “have very limited ability to rework the numbers” (p. 42). Once the data is compiled and the individual details are abstracted away, it can be hard for someone to modify these statistics to generate a new set representing their choice of categories. Thus, the ability of others to challenge this form of power is curtailed.
At the same time, the production of statistics in turn requires power as it influences the consensus of how a statistic should be counted. While perceived as objective, statistics are still subjective to a point in that biases can lead to varying measurements and results of even the same parameter. This leads to a contest over which measurement is the most “accurate” and, by extension, whose result everyone should accept and use. In setting standards in any field, influence and dominance are required. The debate over which measurement should become standard in statistics is no different, in a process Porter describes as “a massive exercise of social power” (p. 33). In fact, all of us contribute to this phenomenon. When we look for statistics, we turn to the powerful. For instance, when looking up statistics on a certain country, most people turn to sources like the Central Intelligence Agency World Factbook, the World Bank or the Economist. These are all powerful organisations that had sufficient influence and dominance to propagate their way of counting. Therefore, those in power are more likely to have their interpretations of statistics accepted by others.
I must concede that the influence of power in altering statistics is limited. Statistics must ultimately gain widespread acceptance for them to be valid (p. 34). While biases do result in differing statistical values that grant favour to their producers, these values cannot be preposterous. Should stakeholders assess that the “measurement process is unreliable or, worse, biased, it may well break down” (p. 33). This limits the extent to which the powerful can distort measurements in their favour and thus the power they may gain from doing so. Furthermore, some impactful statistics are so heavily scrutinised by watchdogs that any attempts to use it to exercise power will certainly be contested, as in the case of population statistics in the United States due to how it affects the “apportionment of political power and of federal reserves” (p. 34). In such cases, even the most powerful may not be able to use statistics as a tool to gain more power.
In conclusion, power begets power in statistics. We tend to subscribe to the statistics of the powerful, and their subjective interpretations are used in decision making as though objective truths. This authority of the powerful to collate and organise statistics subsequently enables more power whenever we lend use to their statistics and allow their biases permeate society. This leads to a feedback loop that self-reinforces existing power structures.
Given the omnipresence of statistics in our lives, the stakes are high. As technology continues to advance and more tools are introduced for data collection, the impact of statistics on our lives can only be more pervasive. It is important for us, then, to be wary of this perpetual cycle of power and be mindful of the effects that statistics can have on us. We must more heavily scrutinise statistical processes to keep those in power in check. We can achieve this by exercising due diligence and critical thinking when working with statistics and not blindly accepting the narratives of the powerful.
Bibliography
Porter, T. M. (1995). How Social Numbers Are Made Valid. In Trust in Numbers: The Pursuit of Objectivity in Science and Public Life (pp. 33–48). Princeton University Press.