Washington Lawyer - June 2020 - 22

SIDEBAR

A SCIENTIFIC SOLUTION

to the Law Firm Diversity Crisis
By Thomas Latino

iversity is lacking in American law firms; this much is undeniable.
The development of diversity and inclusion initiatives has been hailed
as the way forward to achieving a more racially diverse legal profession.
The consensus, however, is that even with all the workshops, theories,
and specially designated directors throughout some of America's largest
law firms, we are still grappling with this issue on a system-wide basis.
So what is the problem?

Law firms and our industry as a whole have
devoted countless dollars and working hours to
analyzing the problem in an attempt to formulate
a solution. Thus far, there has been little progress
in obtaining one. As frustrating as the issue is, the
solution might not be as opaque as we once
thought. I believe science holds the answer
to every law firm's diversity issue - science
grounded in non-biased, practical application
and objective psychometrics. The law firm hiring
process must remove the import placed on GPA
and class rank and instead focus on the collection
of personality traits that have been identified as
predictive of career longevity and success.

process more effective and efficient. The study
did usher in a new breed of legal hiring tools
previously unseen in America's law firms. For
example, "predictive analytics," made famous by
the book and major motion picture Moneyball,
began to be incorporated into law firm hiring
models. Selection algorithms have become the
latest idée du jour in today's hiring landscape.
The problem with this methodology is that it's
rife with implicit bias - the unconscious attitudes or stereotypes that affect our understanding, actions, and decisions.3 Implicit bias
has been defined as one of the root causes of
the legal profession's diversity crisis.4

In 2008, a massive study analyzed the predictive
value of GPA and class rank as they related to
attorney performance. Unsurprisingly, the study
found that these two variables, which have long
defined the law firm hiring model, yielded only
moderately predictive results.1 "Moderately predictive" is another way of saying the variables
explain things on a 50/50 basis, like flipping a
coin. In some instances, GPA and class rank were
predictive; in other situations they were less so.2
"Moderately predictive" is also nowhere near
being good enough in addressing the diversity
crisis facing law firms. This leaves law firms at the
mercy of random chance to find talent who are
not only a good fit but also high-quality and
diverse.

How does implicit bias render ineffective most
of the predictive modeling used in legal hiring?
The answer is in what's entered into the model
at the very outset. An algorithm can only
compute the numbers fed into it by a neutral
third party. If the input is biased, then the
output, logically, will also be biased. Consider
the predictive algorithm designed to project if
an applicant is going to be a high-performing
employee. When using a predictive model such
as this, staff members are asked to complete an
assessment asking them what they think contributes to an employee being highly effective.
The problem is, what bases do the participants
have to determine what qualities make
a successful employee? They most certainly
have their own experiences, but what's true for
one person might not be true for another.

Twelve years have passed since that study, and
law firms are still struggling with how to modify
their hiring models to improve diversity, save on
turnover costs, and generally make the hiring

22 WASHINGTON LAWYER

JUNE 2020

Furthermore, who is to say these are the only qualities of a potentially high-performing employee?

The very act of differentiating high performers
from low performers often reflects subjective
evaluations that are notoriously discriminatory.5
Models based on these practices can mirror
undesirable social patterns. Even when these
tools accurately predict traits that successful
employees share, they could easily turn away
equally talented candidates who don't happen
to share those characteristics.6 Inferred traits
may not actually have any relationship to performance and, at worst, may be largely due to
chance, leading to a purely random result.
The problem with some predictive models and
selection algorithms is that they rely on
employee performance data, or on employee
interpretation of their performance data. For
candidates who pass the initial screening round,
these numbers could possibly create the illusion
of statistical significance and accuracy that
could influence, subconsciously, how hiring
managers view candidates.7
The diversity issue is only going to become
more prominent as clients increasingly demand
to work with law firms whose professional staff
is representative of the population at large.
Some of America's largest companies are
making it clear that if law firms do not hire more
diverse attorneys, they will spend their dollars
elsewhere.8 Hewlett Packard has gone a step
further, invoking a "diversity holdback" mandate
for all outside legal vendors. According to this
policy, HP will hold back 10 percent of the
invoiced amount due to any outside counsel
that does not meet or exceed HP's minimum
diverse staffing requirements.9
HP isn't the only one either. MetLife gave
outside counsel a deadline of July 2018 to
provide a formal talent development plan
on how to make their firms more diverse.
Microsoft offers 2 percent bonuses to its
premier provider firms if they meet certain
diversity goals. Additional bonuses are available
for hitting other benchmarks. More recently,
Goldman Sachs put the world on notice when
it stated it would no longer engage companies

Washington Lawyer - June 2020

Table of Contents for the Digital Edition of Washington Lawyer - June 2020