A Multivariate Understanding: Tornadoes & Systemic Racism

A Change in definition

American society has abandoned the MLK notion of racism, which is a reference to the bigoted views that an individual, a “racist,” holds against another racial group or groups. Beginning in the 1960s, various academic departments collectively replaced the individualistic, psychological notion of racism with a more abstract concept that is more concerned with socioeconomic outcomes than the content of human character. These academic departments, which include most of the humanities disciplines and schools of education, were inspired by French postmodern (Derrida, Foucault, etc.) and continental “constructivist” philosophy (Rorty, Butler, etc.), and the school of social thought known as “Critical Theory,” which assumes that societal systems and institutions have a more significant impact on human well-being and outcomes than individuals’ choices and beliefs. 

Linguistically, racism described more than dehumanizing animus towards fellow citizens; it described the myriad factors that determine racial differences in any outcome variable. While the adjectives “structural” or “systemic” started to precede the noun “racism” in a sentence to differentiate this new concept of racism from the MLK definition, the vast majority of conversations today are concerned with “systemic racism.”

umbrella terms cannot be measured individually

“Systemic racism” is an umbrella term that refers to an immensely large number of social determinants, institutions, and other confounding variables that influence the probability of socioeconomic outcomes. “Systemic racism” cannot be measured. Rather, each of the variables that influence socioeconomic outcomes can be measured individually.

To say that “systemic racism” can be measured is the same as saying that “meteorological phenomena” can be measured. Sure, we can measure barometric pressure, the number of tornadoes during spring, wind speed, temperature, and the level of precipitation. However, that is not functionally the same thing as producing a quantitative measure that succinctly describes “meteorological phenomena” – as that term is a category that various natural events and forces are grouped under. We can measure the individual events and forces, but not the category itself.

Tornado or no tornado?

For some reason, I have always been fascinated with tornadoes. I saw “Twister” in the theater during the summer of 1996 (I was six at the time) and obsessively watched National Geographic documentaries on the subject as a kid. 

Since “Twister” was released twenty-five years ago, the methods of tornado research have evolved. Instead of placing a container of sensors in the path of a tornado, researchers measure the physical dynamics of the storm from afar with Doppler radar. The multivariate data is then plugged into complex computer models that predict the probability of tornadic activity in various scenarios.

Unlike many other elements of the natural world, our understanding of tornadoes is still quite lacking. We understand which atmospheric conditions make a tornado LIKELY to occur, but we do not understand what distinguishes tornado-producing supercells (thunderstorms with a mesocyclone) from non-tornadic supercells.

If we had this knowledge, better warning systems could be developed, and human lives could be saved. Meteorologists, assuming a null hypothesis of “No Tornado,” have the difficult job of choosing to err on the side of failing to issue a tornado warning when they should have (Type 2 Error), or issuing a warning when a tornado did not actually touch down (Type 1 Error). The former risks the loss of human life while the latter risks a loss of public trust.

Our lack of understanding limits the efficacy of our solutions. And climate change is leading to more violent storms.

The 2013 El Reno tornado was the largest tornado ever recorded with a funnel diameter of 2.5 miles. It killed more storm chasers than any other tornado in history.

doing the actual work

Any viable solution to a problem requires a correct understanding of the root causes of that problem. If the problem is not adequately understood, then solutions will either be ineffective or sub-optimal – as all or most of the problem’s root causes will not be considered when formulating the solution. Better tornado warning systems will only be discovered through more a comprehensive understanding of tornadic phenomena.

Similarly, effective policies that equalize socioeconomic opportunity across racial groups will materialize ONLY if the determinants that affect those outcomes are understood. Lamenting the deleterious effects of systemic racism is a ubiquitous form of virtue-signaling practiced by journalists, politicians, and legions of social media users. It is true that structural forces do produce negative externalities. Nevertheless, public declarations of Mea Culpa – where individuals from “privileged” racial groups acknowledge their “complicity & fragility” and repent for their “role” in the perpetuation of systemic racism – do not improve our understanding of racial disparities and socioeconomic phenomena at all. Nor do they help lead to viable solutions to the problem.

This flaccid response to one of the defining challenges of our time reminds me of the common GOP reply to a tragic mass shooting: “We send thoughts and prayers to the victims.” Those vacuous thoughts and prayers are just as effective at moving us toward a viable solution to American gun violence as the public admissions of privilege and complicity are at moving us toward an equilibrium of comparable socioeconomic opportunity across racial groups. In fact, the opposite generally occurs; simplistic understandings of social phenomena, which tend to conveniently conform to ideological narratives, move us toward emotionally gratifying solutions, not effective ones. As linguist John McWhorter has pointed out, these performative struggle sessions are spiritual in nature and eerily mimic religious concepts, such as Original Sin (“whiteness”) and atonement (admitting “white privilege” or what Robin DiAngelo calls “doing the work”).

Remedies to racial opportunity inequality are not easy. It is hard work to understand multi-variate macroeconomic dynamics. Being born again into “wokeness” is easy. Understanding how variables interact and formulating nuanced policy solutions is hard.

The God of the gaps

There are undoubtedly structural forces (many – but not all – are remnants of the institution of slavery and pre-Civil Rights white supremacist policies) that inequitably affect racial groups. Many activists refer to systemic racism like a sentient miasma that permeates all aspects of American society. The way they talk about systemic racism reminds me of how Jedi describe “The Force” in the Star Wars saga. In the 1977 classic space opera, Obi-Wan Kenobi explains the force to young Luke Skywalker:

“It’s an energy field created by all living things. It surrounds us and penetrates us; it binds the galaxy together.”

I do not believe that systemic racism governs life in a similar pseudo-religious sense. If you want to say that systemic racism is an umbrella term that comprises a virtually infinite number of determinant variables, I am fine with saying that systemic racism exists. My concern is not to haggle over semantics. Rather, my concern is that the constant, automatic appeal to systemic racism as the root cause of all racial inequalities severely limits both our understanding of complex social phenomena and our ability to achieve equality of opportunity for all racial groups.

The appeal to “systemic racism” is a contemporary variant of the “God of the Gaps” argument, which referred to the pre-Enlightenment attribution of unexplained physical phenomena (i.e., phenomena that science could not YET explain) to God or another supernatural entity. In this sense, “systemic racism” is a term that describes statistical residuals. It refers to the variation in a dependent variable, which is generally a socioeconomic indicator (wealth, income, health outcomes, test scores, etc.), among racial groups that cannot be explained (or cannot YET be explained) by the available data.

Suppose we stumble across a large cross-sectional dataset (and a free subscription to a statistical software) and set up an OLS (Ordinary Least Squares) multiple linear regression that explains expected lifetime wealth accumulation (y) in terms of all the relevant independent factors (DOB, # of kids, etc.). For the sake of brevity, we collapse all these control variables into a vector (X). 

Since we are concerned with differences in wealth across races, we also include binary variables for each racial group (Black, White, etc.) up to the final Kth racial group, which is Hispanic/Latino in this case. The binary variables will be turned on (=1) or turned off (=0) depending on the racial group of a particular observation in the dataset.

Note: A competent econometric analysis would require more advanced techniques but for conceptual purposes this model is sufficient.

We would find that there are differences in expected lifetime wealth conditional on race after we apply the controls.

Suppose the estimated marginal impact (Beta 1-hat) of the “Black” binary variable on the dependent variable is substantially negative and statistically significant, but our research indicates that there are variables missing from the equation – as the right-hand factors only explain a modest portion of the left-hand dependent variable’s variance. Technically speaking, our model suffers from omitted variable bias, which is a serious problem.

From our regression output, we can estimate the residuals (u-hat=y-yhat), which are the actual ‘y’ values minus the model’s estimated values. The residuals are the leftovers; they are not explained by the right-hand variables

Critical Race Theory (CRT), which has become the intellectual wellspring of social justice activists in America, teaches that ALL disparities in any socioeconomic outcome variable are byproducts of systemic racism. The implication is that, in the absence of systemic racism, there would be equal outcomes or very close to equal outcomes across racial groups.

The critical race theorist would argue that any variance of the dependent variable across racial groups NOT explained by the independent variables is “systemic racism.” They would point to the residuals of our model as EVIDENCE of systemic racism. It is the same “God of the Gaps” reasoning that Medieval clerics would employ to explain the universe and maintain their grip on power – “it was the Divine.” In this case, “it’s systemic racism.”

In many cases, it is obviously true that multi-factorial structural elements are driving racial inequality. However, pointing to the umbrella category and failing to explain the underlying dynamics of the variables that comprise the umbrella category is an exercise without any empirical utility.  

Because the critical race theorist does not define systemic racism in a way that can be coherently measured, there are no hypotheses that can be employed to test their claims; all their arguments regarding systemic racism are not falsifiable. Instead of asking “how do we capture the omitted variables?” or “how do we construct a model that better explains the causes of racial inequality?”, the critical race theorist simply points to the residuals. Instead of doing the hard work of formulating hypotheses and testing them with data, they engage in circular logic and point back to their preferred narrative.

This moral pontification is a naive and anti-empirical impulse that will not improve the status quo or move us closer towards racial equality of opportunity in America, which is a goal that we must achieve. We will develop better tornado warning systems by gathering data, generating models, and testing hypotheses. We will not achieve anything by lamenting the destructive nature of weather.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.