The tragedy of bias in technical hiring

Synopsis  The Birthplace of Bias – and how to combat it  How bias manifests in: − Job descriptions − The Interview Process − The Hire or No-Hire Decision... Meet Julie Ette: • BS in

2014

Why are there so few women in tech?

Why are there so few women in tech?

1 Industry doesn’t know how to recruit and hire women.

(Hint: Industry must hire women before retaining them!)

Synopsis

 The Birthplace of Bias – and how to combat it

 How bias manifests in:

− Job descriptions

− The Interview Process

− The Hire or No-Hire Decision

Act 1: The Players

“All the world’s a stage, and all the men

and women merely players.”

- William Shakespeare, As You Like It

Meet Julie Ette:

• BS in CS from StateU

• 5 years work experience with two

mobile software companies

• Looking for a new job

Will Julie find a match with Monty’s team?

Let’s find out…

Act 2: The Birthplace of Bias

“Wisely and slow They stumble that run fast.”

- William Shakespeare, Romeo and Juliet

The Two-Systems Model of Judgment and Choice

(Kahneman, Daniel Thinking, Fast and Slow Farrar, Straus and Giroux 2011.)

System 1 Source Data

Comes from the cultural soup we experience every day since infancy:

• Role models - parents, teachers, siblings, and

caregivers

• TV, books, music, and cultural memes

• Peers and their own source data!

System 1 creates a meaningful story from our senses and experiences!

(Efforts to fix The Pipeline change the next generation’s patterns.)

The Birthplace of Bias

Cognitive bias happens when

System 1 decides without System 2 helping to catch errors,

assumptions, biases, and mental

short cuts!

(We all do this! Don’t feel bad It’s part of being human!)

Tech Company Culture Exacerbates Bias

“People who are cognitively busy are also more likely to make selfish choices, use sexist language , and make superficial judgments in social situations… but of course cognitive load

is not the only cause of weakened self-control A few drinks have the same effect, as does a sleepless night.”

- Dr Daniel Kahneman, Thinking, Fast and Slow, pp.41

− First impressions influence later experience

− People answer an easy question with System 1

instead of a harder one with System 2

Common Biases in Hiring

 Confirmation Bias

− Seeking data that confirms our ideas

 Fundamental Attribution Error, or the Negativity Effect

− Over-emphasizing traits in others while

under-emphasizing situations (luck) in ourselves

− Unconsciously assuming that others share our own

perspectives, thoughts, and values

So… How do we overcome our biases?

improve judgments and decisions, both our own and those of the institutions that we serve and that serve us?

The way to block errors that originate in System

1 is simple in principle: recognize the signs that you are in a cognitive minefield, slow down, and ask for reinforcement from System 2

Unfortunately, this sensible procedure is least

likely to be applied when it is needed most.”

- Dr Daniel Kahneman, Thinking, Fast and Slow, pp.417

System 1 in Interviews

“ The optimal time to make a decision about the candidate is about three minutes after the end of the interview … I ask interviewers to write immediate feedback after the interview, either a “hire” or “no hire”, followed by a one or two

paragraph justification It’s due 15 minutes after the

interview ends.”

“Never say “Maybe, I can’t tell.” If you can’t tell, that means

No Hire It’s really easier than you’d think Can’t tell? Just say no! If you are on the fence, that means No Hire…

Mechanically translate all the waffling to “no” and you’ll be all right.”

- Joel Spolsky, The Guerrilla Guide to Interviewing v3.0, Oct 25, 2006 http://www.joelonsoftware.com/articles/GuerrillaInterviewing3.html

Act 3: Attracting Diverse Candidates

Subtle Cues in Job Descriptions

The purpose of a job description?

1 Internal: communicate hiring requirements

2 External: promote the job and company

How can the job description project bias?

(See also: Gaucher, D., Friesen, J., & Kay, A C (2011, March 7) Evidence That Gendered Wording in Job Advertisements Exists and Sustains Gender Inequality Journal of Personality and Social Psychology )

Bad (and real!) examples

“Do you have a passion for quality gaming and auto racing ? XXXX Game Studios is hiring!

You are a Senior Software Development Engineer with broad game development experience and

world-class software engineering skills You’re the kind of person who drives projects to

completion, sometimes across multiple functions and groups ”

(See any Projection Bias? Casuistry? Representativeness?)

Bad (and real!) examples

“The Application Programmer Analyst plays a vital role

on the ZZZZ Medical Group Support team,

demonstrating our values of patient-centered care and service ; respect, caring and compassion ; teamwork

and partnership ; continuous learning and

improvement ; and leadership.

In this position you will:

Enhance existing computer programs to add

value throughout the organization …”

(Representative stereotypes exist for female-dominated roles too)

Bad (and real!) examples

“QQQ Software runs a developer paradise: the latest technologies

and platforms and an elite team of great developers No résumé needed! Great work speaks for itself We'd love links to your

GitHub or StackExchange profile!

Who wants these as coworkers?

(And why are they all holding weapons?)

To Attract More Diverse Candidates:

 Be aware of the impact of language.

 Write job descriptions that don’t create role

biases!

− Look carefully for values, traits, behaviors, and

motivations

− Find gender-neutral ways to “sell” the job

 Circulate several versions to attract candidates

with diverse motivations and career objectives

 Get people with different perspectives to edit

job descriptions – how would someone in

marketing write an engineering job description?

Monty’s JD Ben’s JD

(Which one would Julie apply for? What about Julio?)

Act 4: The Interview

The Classic Software Interview

 Short call with a recruiter

 A technical phone screen, some coding

 On-site interview with 4-6 sessions, all with heavy coding

 Many tech companies do no training on how to interview

− Some focus on legal areas of questioning

− A few give training but do not monitor how these

techniques are used in interviews

How effective are tech interviews?

“For the record, we don’t think that the way interviewing is done today is necessarily the way it should be done The current

paradigm puts too much emphasis on the

ability to solve puzzles and familiarity with

generally fails to measure a lot of the skills that are critical to success in industry ”

- Mongan, John, Eric Giguere, and Noah Kindler Programming

Interviews Exposed: Secrets to Landing Your Next Job, 2013, pp xxvi

Schmidt, F.L & Hunter, J.E (1998) The validity and utility of selection methods in personnel

psychology: Practical and theoretical implications of 85 years of research findings,” Psychological

Bulletin, 124, 262–274

Personnel Measures Predictive Validity (r)

Work sample tests 0.54 General cognitive ability

Reference checks 0.26 Job experience 0.18 Years of education 0.10

What can this look like in practice?

 Train interviewers about cognitive biases

 Ask some coding questions

 Also ask behavioral, work habits, and job knowledge questions to assess all the other success factors

 A great resource for technical managers to use with their teams is this book:

Rothman, Johanna Hiring Geeks That Fit Rothman

Consulting Group, Inc 2013

Julie’s interview:

Welcome with recruiter Ben Volio

Software lifecycle, Agile, communication style,

personal work habits with Merlin Cutio

Petra Escalus: CS fundamentals - complexity, networks,

threads, databases, OS

Ty Balt and Amy Bram: lunch at Café Verona

and behavioral and culture fit

Cindy Paris: mad programming skillz - languages, algorithms,

data structures, coding

Phil Laurence: Debugging and testing in mobile & embedded

Finish with Hiring Manager Monty Gue

− The workplace environment

− The technology stack

The interview experience will impact Julie’s final decision!

Act 5: The Decision

The Post-Interview Debrief

 Review job requirements first

 Avoid the Affect Heuristic!

for job requirement #1?”

− Don’t ask: “What did you think of Julie?”

 Watch for cognitive landmines:

− personality traits, intuitions, impressions, or stereotypes,

− Dig into possible bias with questions

− Numerical scores engage System 2

− Don’t use “hire” or “no hire” which uses System 1.

Outcomes

Changing technical interviews won’t convert every hire from a tragedy to a romance…

But, it sets the stage for candidates of all kinds to audition on equal terms.

And what of Julie and fair Roam.io?

Thank you!

Got Feedback?

Rate and Review the session using the GHC Mobile App

To download visit www.gracehopper.org