In Performance Evaluations, Subjectivity Is Not Random

Employment policies associated with unions – e.g., seniority, salary schedules – are frequently criticized for not placing the highest premium on performance. Detractors also argue that such policies, originally designed to protect workers against discrimination (by gender, race, etc.), are no longer necessary now that federal laws are in place. Accordingly, those seeking to limit collective bargaining among teachers have proposed that current policies be replaced by “performance-based” evaluations – or at least a system that would make it easier to reward and punish based on performance.

Be careful, argues Samuel A. Culbert in a recent New York Times article, “Why Your Boss is Wrong About You." Culbert warns that there are serious risks to deregulating the employment relationship, and leaving it even partially in the hands of the employer and his/her performance review:

Now, maybe your boss is all-knowing. But I’ve never seen one that was. In a self-interested world, where imperfect people are judging other imperfect people, anybody reviewing somebody else’s performance ... is subjective.

This viewpoint may sound obvious, but social science research reminds us that the whims of subjective human judgment are not random. The inefficiencies that Culbert mentions are inevitable, but so is the fact that bias tends to operate in a manner that disproportionately affects workers from traditionally disadvantaged social groups, such as women and African Americans. What’s worse – it’s just as likely to occur within as between groups, and we often do it without realizing.

The argument that employment protections, such as tenure, are unnecessary has at least two weaknesses. First, discrimination still occurs in today’s workplaces (also here), despite federal law, and union protections serve a vital role, both in deterring unfair treatment and helping its victims take action when it occurs. Those who believe that the dismissal process for teachers is too burdensome in many places have a point, but it is naïve to believe that federal laws preclude the need for additional protection.

The second problem is that “performance” is not all that gets assessed during “performance-based” evaluations. As Culbert points out, nobody is objective or infallible in assessing others, but there is more to this than personal idiosyncrasy. There is actually a rich base of social science literature explaining how and why our biases become systemic. In other words, human subjectivity is not a randomly occurring event, but quite the opposite – stereotypes about social groups shape perceptions of competence in predictable ways. For example, since women and minorities are stereotyped as being less competent than men and whites, their performance is seen as less indicative of skill (e.g., more attributable to luck), even when they are equally qualified and produce the same results as their white male peers (also here).

Stereotypes are best thought of as one of the many tools we automatically "clutch" and implicitly use (also here and here) from our innate cognitive toolbox as we go about our daily lives. Without this toolbox, humans don’t function as smoothly or as quickly. Measures such as seniority or tenure limit the degree to which we “need” to resort to this toolbox. This means that, the more we move away from policies such as seniority, the more we open ourselves up to our innate tendency to stereotype the world around us.

So, there are good reasons why teachers worry that moving from measures like seniority to more “performance-based” systems would lead to biased decisions affecting pay and other employment conditions (including firing). This is not just a matter of simple favoritism or hostility based on personality, but also the unconscious influence of stereotypes without ill-intent or awareness. Seniority, salary schedules, and tenure may not sufficiently differentiate between high and low performers, but they also do not differentiate by gender, race, or other non-job-related characteristics. And no matter how well we design performance-based measures, they will always open this door.

Even though some people oppose seniority and tenure – understandably, since they can become excessively rigid – these systems have proven to be effective in minimizing differences in pay and other outcomes by sex, race, and other characteristics. In our education debate, we rarely discuss this, because one can so easily be accused of “putting the interest of adults above those of children." In other words – who cares about possible bias against teachers if it improves student performance?

I could not disagree more strongly. For one thing, discrimination-prone measures used for pay, security, etc., by their very nature, threaten efficiency and effectiveness as much as anything. If stereotypes about social groups – and not just merit – always color our views of teachers’ performance, that is bad for students.

I don’t pretend to have an easy answer here but, at the very least, we should only move toward “performance-based” evaluations with caution and a clear understanding of the complexity and subjectivity built in our assessments of others. As for the utility of tenure and seniority, I’ll borrow a phrase from another debate about discriminatory effects: "Mend it; don’t end it."