Can CV Survive Against The Sanctity Of Blind Recruiting?
Fair consideration is the first step to true inclusion
Posted on 09-18-2018, Read Time: Min
Share:
Quickly browsing a CV to find candidates is a pressure many recruiters are used to. But with more and more businesses vying for attention in ever-crowded marketplaces and, in turn, boardrooms are determined to attract candidates from the widest pools, is it possible to continue to be competitive and to ensure the very best service for their clients without bias when reviewing a CV?
The recognition of such a need is clearly out there already. In its most recent Global CEO Survey, 77 percent of CEOs told PriceWaterhouseCoopers they already have a diversity and inclusion strategy or plan to adopt one in the next 12 months. And the talent they want to recruit supports this view. Additional research by PwC research shows that 86 percent of female and 74 percent of male millennials consider employers’ policies on diversity, equality and inclusion when deciding which company to work for.
However, in the eyes of some, current recruiting processes are imperfect, elitist and exclusionary. These concerns make it important that recruiters consider new ways to attract people from all backgrounds. To do this, there is a need to ensure that nothing in the recruitment process puts up barriers that prevent the best talent from joining a firm.
By removing the candidate’s name and other personal information, such as their nationality or the university they attended, an employer can ensure that people will be judged on merit and not on their background, race or gender. Used well, blind recruitment can help ensure demonstrable insights that show differences as a strength– but is this enough? Or does the recruitment industry need to take a closer look at the way language is being analyzed during the recruitment process, especially when it comes to influencing invites to interview?
WCN partnered with University College London to conduct the first large-scale statistical linguistic analysis of male and female resumes across multiple job sectors (financial services, information technology, management consulting and retail and buying). The research ran under the premise that even if blind recruitment removes the name and gender of a candidate, is there evidence to suggest that candidate CVs s may contain lexical, syntactic or semantic differences that sustains gender inequality?
A previous study published in the Journal of Personality and Social Psychology into gendered wording in job advertisements had found evidence that subtle but systematic wording differences influenced job appeal negatively and increased gender inequality. We wanted to go one step further and consider if such differences also exist in candidate CVs,and perpetuate biases either unconsciously through the human recruiter or through machine learning representations.
This work differs from the previous work in that the dataset available is much larger, containing more than 200,000 CVs s across multiple job sectors for analysis. However, similar lexical, syntactic and semantic analysis can be carried out on the corpus to discern differences (if any) between male and female candidates.
The first objective of this work is to establish features that may differentiate a male candidate’s CV from a female candidate’s resume. Are there lexical, syntactic and semantic differences in the text that distinguishes male and female CVs? Are these differences then perpetuated or amplified in machine learning representations of the resumes?
The results of this examination is important in determining if disparate impact on the gender minority may occur in the hiring process due to statistical differences in the underlying male and female CVs, manifesting as redundant encodings of gender information in machine learning representations of CVs and resumes.
For each job sector scrutinized by Oleeo, the top-10 words for each gender from the discriminant word analysis results are listed below in which you can see clear differences between the genders as to the drivers they showcase in their CVs:
The recognition of such a need is clearly out there already. In its most recent Global CEO Survey, 77 percent of CEOs told PriceWaterhouseCoopers they already have a diversity and inclusion strategy or plan to adopt one in the next 12 months. And the talent they want to recruit supports this view. Additional research by PwC research shows that 86 percent of female and 74 percent of male millennials consider employers’ policies on diversity, equality and inclusion when deciding which company to work for.
However, in the eyes of some, current recruiting processes are imperfect, elitist and exclusionary. These concerns make it important that recruiters consider new ways to attract people from all backgrounds. To do this, there is a need to ensure that nothing in the recruitment process puts up barriers that prevent the best talent from joining a firm.
By removing the candidate’s name and other personal information, such as their nationality or the university they attended, an employer can ensure that people will be judged on merit and not on their background, race or gender. Used well, blind recruitment can help ensure demonstrable insights that show differences as a strength– but is this enough? Or does the recruitment industry need to take a closer look at the way language is being analyzed during the recruitment process, especially when it comes to influencing invites to interview?
WCN partnered with University College London to conduct the first large-scale statistical linguistic analysis of male and female resumes across multiple job sectors (financial services, information technology, management consulting and retail and buying). The research ran under the premise that even if blind recruitment removes the name and gender of a candidate, is there evidence to suggest that candidate CVs s may contain lexical, syntactic or semantic differences that sustains gender inequality?
A previous study published in the Journal of Personality and Social Psychology into gendered wording in job advertisements had found evidence that subtle but systematic wording differences influenced job appeal negatively and increased gender inequality. We wanted to go one step further and consider if such differences also exist in candidate CVs,and perpetuate biases either unconsciously through the human recruiter or through machine learning representations.
This work differs from the previous work in that the dataset available is much larger, containing more than 200,000 CVs s across multiple job sectors for analysis. However, similar lexical, syntactic and semantic analysis can be carried out on the corpus to discern differences (if any) between male and female candidates.
The first objective of this work is to establish features that may differentiate a male candidate’s CV from a female candidate’s resume. Are there lexical, syntactic and semantic differences in the text that distinguishes male and female CVs? Are these differences then perpetuated or amplified in machine learning representations of the resumes?
The results of this examination is important in determining if disparate impact on the gender minority may occur in the hiring process due to statistical differences in the underlying male and female CVs, manifesting as redundant encodings of gender information in machine learning representations of CVs and resumes.
For each job sector scrutinized by Oleeo, the top-10 words for each gender from the discriminant word analysis results are listed below in which you can see clear differences between the genders as to the drivers they showcase in their CVs:
Financial Services
Female: organize, event, volunteer, assistant, social, student, marketing, community, department, plan
Male: equity, portfolio, investment, capital, analyst, finance, market, stock, interests, technical
Male: equity, portfolio, investment, capital, analyst, finance, market, stock, interests, technical
Information Technology
Female: volunteer, event, assistant, organise, analyse, plan, student, social, conduct, excel
Male: php, c, software, linux, c++, computer, have, developer, engineer, network
Male: php, c, software, linux, c++, computer, have, developer, engineer, network
Management Consulting
Female: volunteer, assistant, event, social, organise, write, community, student, communication, research
Male: engineering, sport, investment, finance, analyst, club, cost, financial, technology, technical
Male: engineering, sport, investment, finance, analyst, club, cost, financial, technology, technical
Retail and Buying
Female: art, child, volunteer, shop, assistant, assist, social, design, organise, create
Male: football, play, sport, business, club, technology, computer, mobile, it, leadership
Overall, the analysis demonstrated how small, but statistically significant patterns, such as resume length, readability and use of certain words can easily lead to gender identification and influence a recruiter looking to achieve gender balance subconsciously. Whilst the statistical differences for CVs between different job sectors are very apparent, the differences in a specific job sector between male and female CVs was more subtle but still statistically significant. For instance, the number of sentences, words and unique words used in a CV have distinct differences across job sectors.
However, there is a gender difference in that 90% of the top-10 male discriminant words are proper nouns and nouns, whereas only 68% of the top-10 female discriminant words are proper nouns and nouns.This indicates that proper nouns usage plays a big role in differentiating male and female CVs s, which is not surprising as it encompasses given names which have gender connotations.
Even with blind recruitment methods applied, the redundant encoding of gender in machine learning representations could still be an issue and appears to be amplified, which could lead to disparate impact on the gender minority.
So can this be mitigated? The research offers a positive perspective for recruiters to make use of gender de-biasing methodology that can be applied to successfully remove gender redundant encoding and lower disparate impact scores applied during machine based hiring predictions. This makes it suitable for use in automated hiring screening processes where not causing disparate impact is paramount.
The proposed intelligent automation de-biasing method is not only proven to remove gender redundant encodings, but also maintains a better prescriptive intelligence hiring prediction performance through machine learning. It is also shown to have consistent negligible disparate impact across a range of hiring targets, providing room for adjustment in recruitment screening thresholds without increasing disparate impact.
Technology works because algorithms can replicate your collective decision making, reducing the influence of bias by individuals or process. It’s not just WCN saying this. The Confederation of British Industry has described “name-blind” recruitment was one way to remove “criteria that could unintentionally bias managers, and give under-represented groups confidence that their application will be fairly considered”.
Fair consideration is the first step to true inclusion. How a company then promotes its values to demonstrate this is the next step. Once applications are blind sifted, the recruiter must consider if they can fairly run interviews alone and continue the momentum. Panel interviewing may help avoid doubt and showcase a more transparent commitment to equal opportunities. The power really does lie in the recruiter’s hands to see these values through. Are you ready?
Male: football, play, sport, business, club, technology, computer, mobile, it, leadership
Overall, the analysis demonstrated how small, but statistically significant patterns, such as resume length, readability and use of certain words can easily lead to gender identification and influence a recruiter looking to achieve gender balance subconsciously. Whilst the statistical differences for CVs between different job sectors are very apparent, the differences in a specific job sector between male and female CVs was more subtle but still statistically significant. For instance, the number of sentences, words and unique words used in a CV have distinct differences across job sectors.
However, there is a gender difference in that 90% of the top-10 male discriminant words are proper nouns and nouns, whereas only 68% of the top-10 female discriminant words are proper nouns and nouns.This indicates that proper nouns usage plays a big role in differentiating male and female CVs s, which is not surprising as it encompasses given names which have gender connotations.
Even with blind recruitment methods applied, the redundant encoding of gender in machine learning representations could still be an issue and appears to be amplified, which could lead to disparate impact on the gender minority.
So can this be mitigated? The research offers a positive perspective for recruiters to make use of gender de-biasing methodology that can be applied to successfully remove gender redundant encoding and lower disparate impact scores applied during machine based hiring predictions. This makes it suitable for use in automated hiring screening processes where not causing disparate impact is paramount.
The proposed intelligent automation de-biasing method is not only proven to remove gender redundant encodings, but also maintains a better prescriptive intelligence hiring prediction performance through machine learning. It is also shown to have consistent negligible disparate impact across a range of hiring targets, providing room for adjustment in recruitment screening thresholds without increasing disparate impact.
Technology works because algorithms can replicate your collective decision making, reducing the influence of bias by individuals or process. It’s not just WCN saying this. The Confederation of British Industry has described “name-blind” recruitment was one way to remove “criteria that could unintentionally bias managers, and give under-represented groups confidence that their application will be fairly considered”.
Fair consideration is the first step to true inclusion. How a company then promotes its values to demonstrate this is the next step. Once applications are blind sifted, the recruiter must consider if they can fairly run interviews alone and continue the momentum. Panel interviewing may help avoid doubt and showcase a more transparent commitment to equal opportunities. The power really does lie in the recruiter’s hands to see these values through. Are you ready?
Author Bio
| Charles Hipps is CEO and Founder of WCN. He has worked with many large- and small-sized organizations across a wide variety of industry sectors on e-Recruitment. Connect Charles Hipps Visit www.oleeo.com |
Error: No such template "/CustomCode/topleader/category"!
