There are three key types of business analytics: Descriptive, Predictive, and Prescriptive.In simple terms, these three types of Analytics can be thought of as a way to help businesses identify current trends and patterns within data; make forecasts and determine the best outcome or direction for the business given a present set of circumstances and/or data. Looking at data in this way allows one to assess a business from both a qualitative and quantitative perspective; a fairly new paradigm.


Looking at Diversity and Analytics in concert

Analytics are employed to help businesses gain valuable insights. People Analytics[1] became popular in the early 2000’s and has become a prominent tool for companies who are sincerely focused on Talent management. It is not enough to say you are focused on moving the needle in diversity and not have the numbers reflected in your headcount and have support from the heads of the business.

How many women are you recruiting? Are there any Blacks or Latinx? Are there any LGBTQ candidates? Further, are you advancing the same number of women and men through your ranks each year? Are you able to assess your organization as a whole on an officer level: Associates, VP, MDs; and further reassess each level by gender and ethnicity? Determining these metrics as well as measuring attrition levels across an organization can provide crucial information to be taken into consideration in diversity and inclusion planning.


The Implementation Process

Despite the increased focused on D&I initiatives there remains much work to be done. There are currently five black CEOs in Fortune 500 Companies. With so few women and minorities in overall top management positions, we must be intentional in increasing the number of diverse candidates in our Fortune 500. Goals and Metrics need to be established for a 3-5yr range in order for progress to be seen. It should begin with the recruiting process within a firm and continue throughout an executives succession program. Used correctly people analytics can be overwhelmingly beneficial to the effectiveness of a corporation.

There are different methodologies in which people analytics can be implemented to assist in the evaluation of a candidates and firms progress. We will look at Regression Analysis and also consider People Analytics in terms of Relational Analytics / Interaction Analytics.


Performance Evaluation using Singular and Multiple Linear Regression Analysis

Regression Analysis, simplified, is a method for investigating the functional relationship between variables. For instance, we may wish to examine whether an employee’s compensation may be impacted by their age; we may further wish to examine other considerations such as ethnicity, officer status and gender for data at the employee’s respective company. This type of analysis is extremely valuable for management when looking at compensation across the company to ensure there are no biases.

To benefit readers of the article there will be a summary of the analysis in the main portion, detailed breakdown (including equations and graph with explanatory information) may be found in the analysis at the end of the document. [3]

The technique behind Singular Linear Regression requires one to initially evaluate data visually on a scatter plot. The data set must display linearity [2] to proceed with the regression. If the data is linear, a regression analysis can be run to determine the best fit line (this can be done easily in Excel). The best fit line is the line that minimizes the distance between the data points and accurately describes the relationship of the two variables. Excel will return an equation for this line. Once we obtain the equation, we are able to use it forecast the variable we are solving for.

Example:

We have employee Maria; who works at Company A, we have compensation data for different stages during her career (stages are given by age). We use the data to answer the question “What should Maria’s compensation be when she is xx yrs. old?”. This is single linear regression analysis.

We are using Descriptive Data to make an assessment on Maria’s compensation, this is Predictive analytics.

Taking this a step further; suppose we want to look at Maria and determine if she is paid appropriately based on her gender and her title compared to the others in her company (we will only compare Maria to others who have advanced with her, those on the same officer level, an apples to apples comparison). We would then need to add two additional variables; gender and title. This now becomes a multiple linear regression analysis.

And the question now becomes “Is Maria paid on par with equally qualified men?”.

This additional analysis allows us to evaluate Maria in comparison to her peers, what is our assessment for the business? We have now implemented Prescriptive Analytics.


Relational Analytics / Interaction Analytics

There has been an overwhelming response to Relational Analytics / Interaction Analytics since Google’s Project Oxygen. Relational Analytics is employee attribute data and allows one to better understand opportunities for business improvement. Data is collected across the firm in respect to how employees interact with one another (social networking); this information along with an individual’s personal attributes is now thought to be a strong indicator of work-place performance. By looking at employees interactions with one another; companies can better assess who is the more innovative employee, who has more influence, or who might end up leaving the firm for a better opportunity.

In a recent study of evaluating more than 1500 project teams for Efficiency researchers studied a wide variety of perspectives such as ethnicity gender education and tenure when measuring efficiency. Nevertheless, the results showed that these variables did not make the greatest impact on the teams performance. Looking at relational data provided much greater insight. There were two variables that affected team performance: internal density and external range. Internal density is the amount of connectedness the team members had to one another, how well they related and got along. External range was their ability to network outside of their immediate group to other experts; members must have had distinct contacts from other members outside of the group to have measured well in this category. This greatly increases the teams potential to access information, secure resources and meet deadlines when necessary. These two variables were critical in increasing the overall efficiency of the team more than just diversity and tenure.

Relational Analytics has become much more multidisciplinary in nature and serves business leaders to drive business results. More traditional people analytics in terms of performance management is also a vital tool for talent management, understanding internal mobility and employee retention.

Other firms have also implemented these types of analytics as of late to evaluate their workforce. JPMorgan has enlisted, R squared, a cloud based artificial intelligence and natural language processing platform analyzing employees email, text messages and other digital communication in order to determine the employees level of engagement and inclusion. The company’s offering states that their services eliminate the need for full engagement survey’s thereby resulting in happier employees, as survey’s are often met with resentment, and also aid in succession planning. It is also worth noting that engagement surveys are deemed to be one-off responses and do not predict patterns of behaviors over time. Perhaps the largest benefit in utilizing this platform, in particular to a JPM investment bank type clientele, is the elimination of overhead given the cloud-based technology.

There is no doubt that people analytics is a rapidly expanding component within D&I. The question becomes how to effectively put these practices/insights into place within your organization in a manner to reap the rewards, maintain low cost while safeguarding security and privacy standards of employees. I look forward to bringing you additional content on this subject in the near future.




[1] For the purpose of this article People Analytics is meant to refer to people within one organization or firm; it can be defined in more broad terms as Analytics of interaction within an organization and of those who the firm interacts with (clients /consultants/ sales).

[2] The scatter plot in a regression is a normal probability plot of the residuals; if assumptions hold these data points should be a random scatter of points (uncorrelated to the predictor variables).

[3] Appendix

September 25, 2020

DATE:

STEPHANIE GARLAND

WRITTEN BY: