Top 3 ways to address systemic bias using workforce data
In an uncertain market where customer needs, organisational outcomes and metrics keep changing, data provides a stable baseline. Business leaders today are expected to be data-driven and leverage data as an important source of input in their decision making — whether it is strategic or operational. This skill, combined with (not replacing) the leader’s experience and intuition, is required to create a meaningful change for customers and the workforce.
In this article, I explore the role of data in making DEI (Diversity, Equity and Inclusion) decisions, the underlying challenges in leveraging employee data and finally, the top 3 ways to address systemic bias using the data in hand.
The role of data in DEI decisions
While changes to individual behaviour, mindset and ways of working is important, it is the systemic change that significantly impacts the workforce. An organisation’s DEI efforts cannot be fragmented and launched as a separate “project” or “initiative.” The principles around inclusiveness, equity, fairness and belonging have to be integrated into all workforce practices, processes and policies within organisations.
This calls for leaders to reframe their thinking from ‘launching DEI initiatives’ to ‘measuring DEI outcomes’ to achieve systemic change.
Identifying, measuring and reporting on DEI metrics is an area where leaders can leverage people analytics before making $$ investments.
For example, a popular study says X% of employees who perceive that they are being paid fairly are less likely to look for jobs outside their organisation. But based on this data alone, it may not be possible for leaders to identify the specific workforce interventions required. They will have to use their own workforce data to understand:
- Is ‘Pay’ the primary reason why employees are quitting over the last 12–36 months?
- What % of our workforce are being paid ‘unequally?’
- Which workforce groups are impacted by pay gap?
- Is the actual ‘base-pay’ the issue here, or are there associated reasons for turnover such as lack of career progression, lack of other benefits etc?
- Are people appreciative of the level of pay transparency within the organisation? If not, why so?
- What is the cost of not ensuring ‘fair pay’ across the board now?
- Have we tried to address this issue in the past? If yes, what are the lessons learned?
Diving deep into these questions confirms the issue exists. It enables leaders to pinpoint that the issue persists potentially due to inconsistent hiring practices or biased performance evaluation or the lack of transparent communication or the combination of it all. Data will guide the leaders to identify the root cause more accurately, and without it, any intervention might result in little-to-no return on investment.
Challenges with workforce data
The fundamental challenge that organisations face today in adopting a data-driven decision making is the absence of data and it’s collection itself. Following are some considerations for leaders to leverage what might be already available to them, to draw insights.
1. Data is distributed across multiple systems
The HR Information Systems (HRIS) contain all the necessary data about the employees from a recruitment, learning, payroll, performance processes perspective. They are considered to be the source of truth for all employee data in an organisation. Using the previous example as a reference — questions 1, 2, 3 and 4 can be easily determined using the HRIS data. The downside of these systems is that they are classified as transactional, since their primary function is to store and update records and may not necessarily have built-in capabilities for analytics and reporting. Reporting becomes particularly complicated because organisations tend to use multiple external systems to store different sets of information.
2. Data quality is inconsistent
When we ask employees to disclose certain information, there has to be a rationale why it is needed and how that data will be used by the organisation. Most employees will want to check that “prefer not to say” box when it comes to race, ethnicity and sometimes gender, since they want to avoid discrimination at any cost.
Gender, Race and Ethnicity are not binary values. Even for the sake of reporting gender pay gap, getting employees to group themselves into male and female might be considered systemic bias.
Similarly, employees may choose not to disclose their ethnicity, or LQBTQ affiliation or disability, to avoid discrimination. Nudging employees to declare their ethnicity or disability and storing that information in HRIS might provide the right data to conduct the analysis of ethnic & disability pay gap. However, it does not systemically guarantee that their ethnicity or their disability will not be used as an excuse for not giving a promotion or pay rise or induce other non-inclusive behaviour within the organisation.
3. Data accuracy is questionable
Skills play a crucial role in pay gap analysis as it speaks to why a certain employee may be paid higher for the same role/ same job title than another (skills are only one factor, there are other dimensions to this such as years of experience, location etc.) But employee skills data has its own considerations.
- If the skills are entered by employees themselves without any form of assessment, then there is a question of inflated projection.
- If employees take up various training programmes, it might guarantee theoretical knowledge but may not prove applicability.
- Assessments are widely available for soft skills which might provide some legitimacy to the data collected.
- External certifications for hard technical skills by various solution providers can be considered as acceptable.
Organisations need to ensure that there are underlying principles and approach in place, to govern the mechanism (credentials or testimonials etc.) that are considered acceptable to attest to employee abilities.
4. Data might be unreliable to a certain extent
Survey as the option for deriving insights might be a bad idea since employees are extremely prone to survey fatigue. Even if leaders want to launch surveys there’s a whole layer of bias to watch out for:
- Selection bias — Is the chosen sample set broad enough?
- Response bias — Is the respondent neutral to the majority of the questions?
- Interviewer bias — Is the interviewer biased and are they influencing the respondents towards a particular direction?
Some employees choose to provide more honest response to anonymous surveys. Some react indifferently to anonymity as they would like be recognised for sharing their point of view, and share even more insights over direct interviews/focus group discussions. The anonymity factor can remain optional while launching surveys, but it completely depends on the objectives and the outcome we want out of these surveys.
#1 Pay equity eliminates bias in the way people are paid
Ensuring Pay Equity is one of the best ways to eliminate systemic bias. It addresses the practice of paying employees differently for the same work they do at the same level, by removing factors of discrimination such as — gender, race, ethnicity, generation, age from the equation.
By conducting frequent pay audits, organisations can ensure that no workforce groups are being treated unfairly in terms of pay, access to other opportunities and benefits within the organisation. Pay Equity enables organisations to become fair, increase workforce happiness, retain key talent and contributes to company’s ESG rating.
#2 Flexible working options eliminate bias in the way people are valued
According to the Women @ work study conducted by Deloitte, women faced the most discrimination at work during the pandemic. Even outside of work, they bear the greatest responsibility for household tasks, and often feel they need to prioritise their partners’ careers. Despite working — and, for most respondents (88%) working full time — nearly half of the women polled are primarily responsible for domestic tasks such as cleaning or caring for dependents.
Polls/pulse checks as a form of data collection can be really beneficial for leaders to understand the state of current employee wellbeing and burnout insights. Using this indicative data to compare efforts required against their jobs, leaders might be able to provide flexible working options to their diverse workforce — empowering employees to take ownership of their own performance, than to enforce outdated policies that impacts their engagement and performance adversely.
#3 Inclusive rewards design eliminates bias in the way people are motivated
Reputed organisations around the world get exclusive bragging rights on how employees benefit by working with them, through their lucrative rewards and benefits strategy. But how relevant and useful are such rewards within the organisation for ALL DIVERSE workforce groups?
If an employee is a “local,” then they are entitled to X number of weeks of paid maternity break whereas if they are a foreigner, they are entitled to half that number of weeks of paid leave.
For employees above Managerial level, they enjoy premium insurance benefits with a leading provider for their childcare and maternity hospital care.
Dads of newborn babies enjoy X number of days of paid parental leave, which is drastically different from paid maternity break, with leave days non-transferable to their partners.
The above statements are just an example of different organisational policies around the world, centered around childcare with varying levels of biases. Is childcare, maternity and paid paternity breaks even applicable to employees who are not parents?
Rewards/benefits is another area where people analytics can ensure that such programmes are free of bias and discrimination.
A points-based reward strategy is a good way to let employees redeem incentives that are meaningful and motivational.
To conclude, workforce data is vital for leaders to act on systemic bias and DEI impact measurement. Data has its own challenges but with People analytics, leaders can develop meaningful interventions that improves workforce and organisational outcomes.
