Where Facts Don't Matter

When a group of people get an idea in their head and refuse to accept any answer that does not align with their preconceived notions, at what point does the attempt to communicate no longer make sense? This is a question I've asked myself a couple of times over the last few years as some colleagues reach out to complain about how a certain group of statistics is unfair and biased against them personally. The idea that math is being used to unjustly punish people who earn the money that pays my salary is absurd and I have conveyed this back to the aggrieved person in as open and complete a manner as I can muster because anything less would be disrespectful. Yet it seems that not a month goes by where someone doesn't openly reject the explanations and claim bias against themselves or unnamed former colleagues.

The issue revolves around a single question on a survey that is offered to some of the customers at the day job. The answer to this question is used to determine, in aggregate, how well a colleague lived up to expectations. It's entirely subjective on the part of the customer and there are cases where someone's had a bad day and a response is later revised or removed completely. Several years of data has shown that people who come to work with a positive attitude generally score an average of 1.13 points higher1 than those who are generally less invested. Employees with a higher score are generally rewarded in some manner. Employees lower in the rankings are not. None of this is rocket science and the numbers cannot lie. Yet some feel there is bias in the single equation that is used for everyone regardless of rank, seniority, competence, or friendship.

This is the gist2 of the SQL query used to generate the number that is upsetting people:

SELECT AVG(score) as avg_score
  FROM Responses
 WHERE is_deleted = 'N' and employee_id = {whatever}
   and response_at BETWEEN DATE_FORMAT(DATESUB(Now(), INTERVAL 1 YEAR), 'YYYY-MM-01 00:00:00')
                       AND DATE_FORMAT(DATESUB(Now(), INTERVAL 2 WEEK), 'YYYY-MM-DD 23:59:59')

In plain English, this is telling the database to collect the average score for all responses that:

  • are not marked as deleted
  • are for a specific employee
  • are received between the start of the current month a year ago today3 and the end of the day two weeks ago today4

Where can bias exist in this equation? It was kept brutally simple on purpose.

But some people don't care. Given the range of concerns and grievances that make their way to my inbox, sometimes I wonder if people want to feel like a victim of a conspiracy in order to justify some aspect of their personal or professional life. Narratives like "I can't get a raise because the system is against me" rarely make for a good story for the sole reason that it's usually not "the system" that's against a person. At the end of the day, we are our own worst enemies When I worked in the classroom I would occasionally feel that "the system" was against me. However, after becoming part of the corporate machine, I quickly learned that there is no system. There are policies, practices, and habits that may not be communicated very well and there are protectionist fiefdoms that get in the way of progress, but there sure as heck isn't "a system" that intentionally hinders one group over another.

This is what I try to convey to people when they complain in my general direction about something that they find frustrating that actually has absolutely nothing to do with them personally. Yet my words are rarely taken seriously. As part of the corporate machine, "the system" has clearly gotten to me and now I must be lying in order to protect … something? My job? My position? Neither of these would exist if it were not for the employees who are spending their days interacting face-to-face with customers5.

But this logic is rarely accepted. Perhaps some people want to be angry. Perhaps some people want to feel downtrodden. Perhaps some people just cannot accept that maybe the negative attitudes they exude at the office bleeds into the classroom. Based on life experience, though, when everything seemed to be against me, it was really myself that I was battling. The rest of the world had nothing to do with it.

  1. This is on a scale of 1 to 5, with 5 being perfect and 1 being a cause for concern.

  2. I'm not supplying the actual code, because who cares? The purpose of the query is to show just how painfully simple it is. There is no place for bias to insert itself in something so primitive.

  3. Today is September 14th, 2019. So the start of the month a year ago today would be September 1st, 2018. This was done so that numbers could be seen rising when employees check their stats during the month. If we were consistently using "a year ago today", then people would see the total number of responses vary every day, which would make the whole purpose of the endeavour suspect.

  4. There is a two week delay between when a person provides feedback and when an employee can see it so that, if there's a problem reported, the customer can be spoken to ahead of time. A lot of the lower scores have nothing to do with the employee, so there's no reason for someone to feel penalized before the reasons can even be confirmed.

  5. What's interesting is that I've pretty much made my living the last couple of years building the tools that support the people complaining. I go out of my way to understand their problems, challenges, and concerns, then solve the issues in an open and transparent manner, often on my own time. Fortunately, the people with the longest list of troubles are few and far between. Most colleagues are consistently far more positive and creative than I could ever hope to be.