A prominent discrimination case in Canada involves a firefighter named Tawney Meiorin.  Meiorin had successfully performed her duties as a firefighter for many years.  She lost her job after the introduction of mandatory testing to determine her fitness for the position.  The testing measured aerobic capacity, and it was developed in a manner that many would regard as scientific; that is to say, it used a highly quantitative and analytic approach.  However, the Supreme Court of Canada used a critical lens to interpret the data.  It found that the tests results did not seem related to Meiorin's actual ability to perform her job.  Moreover, it seemed that most women were unable to pass the test.  What I describe here as institutional data arises from the actions of institutions - done to a large number of people by a large number of people - using collective normatives.  I would argue that in the Meiorin case, a key normative is that only healthy young men should be firefighters.  The imposition of scientific testing and dismissal of those that failed reflect the use of institutional data; the facts and figures existed to achieve specific collective objectives.  Data in this context can regarded as an instrument of control.  In this blog post, I discuss the dangers of making use of data without considering its institutional intent and collective consequences.

Normally after that type of introduction, I would probably be expected to delve deeper into the human rights component of the Meiorin case.  Actually I want to underline the adverse business impacts of using data to usher in strategies that are disassociated from the surrounding realities of the market.  We have a scenario where there was a great deal of data both during the development and application of testing.  However, the data served a purely instrumental purpose.  It defined rather than explained reality.  In past blogs, I have used the example of sales figures.  Taken simplistically, a measurement of sales provides almost no explanation about the underlying contributing factors.  So if sales start to decline - as they surely must at some point in a competitive environment - an organization finds itself with little guidance.  We have to recognize the institutional nature of sales:  it can provide a basis of comparison between the performance of different companies; it might help an organization determine its rate of growth; and it could offer some indication of future inventory requirements.  It might also be completely wrong.  This is not to say that the quantification is necessarily faulty, but the organizational inferences might be poor or inaccurate due to the deeper nature of the data.  A measurement of sales is likely relevant only in specific contexts to particular parts of an organization.  Consumers could care less about the numbers; nor is the data all that connected to the underlying needs, desires, hopes, and aspirations of consumers.

A large US retailer recently started operations in nearby Canada.  Canadian sales have declined 46 percent in a single year.  Of course during early planning, it seems unlikely that any analyst would factor in a drop of almost half of revenues after the first year of operations.  The institutional data likely provided a reasonable growth profile thereby legitimizing entry into the Canadian market.  I was listening to the CBC on my way home, and I was left with the impression that the retailer expects Canadians to adapt to its business model.  So organizational decision-making was quite insulated from the consequences of its actions.  Organizational pathology has tended to indicate to me the presence of disassociation between perception and phenomena in the data itself.  The context is sometimes imposed by an institution over its data such that there is reinforcement of illusion.  A powerful force can emerge to maintain behaviours that are possibly irrelevant or counter to the underlying organizational needs.

I have used special terms in the past to describe the underlying themes in this blog, and for the sake of consistency I will mention some of them here.  Data can be used to serve the instrumental objectives of an organization.  For instance, data can be collected simply to confirm the extent to which particular processes have occurred.  (Whether those processes lead to the intended outcomes might fall outside the metrics.)  The metrics of criteria support a regime of compliance through the use reductive data.  This is data that informs us only of a specific contextual reality.  The measurement of sales is reductive as it tells us nothing about why sales increase or decline.  We simply gain an indicator of impacts to revenue.  Whether there is a small amount of data or an abundance, the existence of that data is within the context of control dynamics that define the meaning of the data.  Organizational perceptions can become separated from the underlying phenomena of the data including outcomes and consequences.  Many philosophies and behaviours can emerge during good times that some might claim to be related to sales; but when sales decline during bad times, the irrelevance of the reductive data becomes more apparent along with its disassociation to the surrounding realities of the market.

When an organization collects data about customers and future demand for its products and services, there is always a danger of the data concealing rather than describing reality.  So it is important to be critical - to question the underlying meaning of the data.  An organization can pathologically insulate itself from reality - and later express great surprise about its unexpected journey.  It is possible to have a great deal of data that, rather than causing an organization to engage reality, leads to its disengagement, detachment, and alienation.  When I use alienation in an organizational sense, it is usually self-imposed.  If we manage what we can measure, then the construction of data can effectively disable managerial efforts.  Not only this, but the institutionalization of data fosters a decision-making environment in which only certain managerial efforts take pace.  I find that organizations actually share many parallels with humans.  Drinking and driving come to mind - the so-called impairment related to the chemically induced separation of reason from reality.  As we come to terms with our ability to collect and make use of large amounts of data, I think it is important to sit back and question the meaning of the data.  The institutional response to a situation is not about the search for meaning but rather its imposition.  When we are unable to understand the underlying phenomenon, applying an institutional response does not help us deal with the problem, but rather it insulates our perceptions and the actions we take.

Views: 419

Tags: accuracy, alienation, construction, critical, disablement, disassociation, impairment, inferences, instrumentalism, lens, More…management, metrics, organizational, pathology, reliability, social, strategic, validity


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Phillip Burger on November 25, 2013 at 8:15am

There are so many discussions that can come of this topic.

A constant is that with rights come responsibilities. This is true at both the personal and the institutional levels. And true prior to data science and now, with data science and analytics. A worry I harbor is that without the intervention of a rational decision maker, i.e., the human, the algorithms will be too complicated for any body to explain why someone didn't make the force. Then what?

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service