For many, diversifying your workforce starts at the source. The more diversity in the candidate pool applying for your job, the more likely there will be diversity in the resulting hires. So, if you are not getting the range of diversity you need in your candidate pool, it may be time to change or expand where you are sourcing your candidates.
But, how can you tell if that expansion is working or if you have a sourcing problem to begin with?
Track metrics of success for your candidate source initiative, such as the diversity of candidates and diversity of hires. Keep the data, such as the candidate demographics, separated from the candidate screening process to avoid any real or perceived bias.
So how do you measure metrics around demographics?
Track the source
Track where every candidate is coming from as it enters your process. From a major job board to a referral from John. If you don’t know where your candidates are coming from, you can’t tell which sources are effective and which are not.
For the candidates from each source:
Measure what matters
Common demographics use the EEOC standards – Race/ethnicity, gender identity, Veteran status, disability, age. But there may be other data you need to measure the success (or lack of success) of your sourcing. Sexual orientation, educational background, industry experience, to name just a few.
For any data, EEOC or otherwise, the data collection must be:
- Optional. You must allow your candidates to opt out of providing any personal data that is not pertinent to the role.
- Clear. Make sure what you are asking for and what you are using it for is very clear. Include links to D&I statements, privacy statements, and information about how that data will be protected.
- Inclusive. Watch for language that may be inadvertently alienating or offensive. For terms where there are preferences, use both (Black/African American). When it comes to demographics, avoid colloquialisms and humor (not everyone will find the same thing funny, especially when it comes to terms of self-identification).
- Legal. If it is not EEOC-standard data worded in an EEOC-standard way, make sure legal reviews the questions, supporting privacy statements, and process to make sure everything is compliant.
Measure at multiple points
Measure at multiple points of the hiring process so you can track the demographic distribution by step – the distribution of candidates who applied, passed initial screening, passed the interview, etc. That will give you data on both the quantity and quality of the diversity sourcing efforts as well as expose any adverse impact throughout the process (i.e. giving an unfair advantage to one demographic over another).
Here’s an example of what that data looks like in career.place, (showing just one facet of diversity reporting – race/ethnicity):
In order to avoid any real or perceived bias when measuring demographics:
1) Store the data separately
Store the demographic data in a separate place from the rest of the candidate information so the hiring team actively participating in the selection process has no way to access that data.
In the above example with career.place, the data is only accessible in aggregate by organization, group of jobs, or individual job so the candidates themselves remain anonymous.
2) Do not update demographic data in real time
If the demographic report is updated in real time (i.e. it changes every time a new candidate applies or progresses through the screening process), then it’s possible to know an individual’s demographics.
For example, imagine you check the demographic report and see that there are four women who applied. Then you see a new candidate enter the process. If you check the report again and see there are now five women who applied, you know the new individual is a woman.
Instead, use a method such as batch updates (for example, update the data after every five candidates that apply) or time-based updates (for example, update the data once a week).
3) Do not report on one person
If the demographic report has a data set of 1 (there is one person who applied or one person who made it to the interview stage), then reporting on that individual will expose that individual’s demographics.
For example, imagine you are reporting on the demographics of all candidates who apply and who pass the blind screening process. If there is only one person who has passed the blind screening and the demographic report shows one woman passed, then you know the candidate is a woman.
A simple solution is not to report on data sets of 1. In other words, if there is only one person in one of the reported states, hide it as “No Answer Provided”. At career.place, we automatically hide single data sets.
Interpret the data in context
Before jumping to any conclusions that you don’t have enough [demographic here] applying for your job, take a look at the available candidate pool. Diversity goals that don’t reflect the candidate pool are much harder to obtain and often lead to discrimination rather than the intended purpose.
If you are only sourcing locally, for example, the available pool could be reflected using the population’s demographics or local graduation demographics for relevant fields. If you don’t like the diversity of the candidate pool, focus on diversifying that first.
Consider, if you set a goal to hire 20% Veterans against a candidate pool that only has 10% Veterans for the field, then the goal will more likely drive the process to discriminate against those who have not served rather than promote equity for Veterans. So, if there are not enough Veterans in the field locally, consider remote positions, relocation, expand what it is to be qualified by removing requirements, or considering transferable skills.