Buyer Beware! Correlation Does Not Equal Causation.

The clear majority of dashboards/benchmarks/scorecards present data that is correlated. When two variables change together in some consistent way (e.g. training and improved performance), they appear to be related and are said to be correlated. Survey-based methodologies for evaluating investments are able to establish a correlation between the investment and the recipient’s performance.

Many of you have heard the expression “correlation does not equal causation.” When two things appear to be related, it’s important to investigate whether other factors may be involved. To establish causation an investment’s impact on the business must be isolated from all the other variables that also maybe impacting the outcome.

For example, there is a clear-cut relationship between ice cream consumption and murder rates—as more ice cream is consumed in the United States, more people are murdered. Is it fair to say that eating ice cream increases the likelihood that one’s life will be violently cut short?

These events are clearly correlated, but digging further you will see high temperature is the cause. In hot weather, more people congregate outdoors, increasing the likelihood of conflicts leading to murders. These links are all correlations and thus do not say much about actual causes. To assert that ice cream sales have a causal effect is not only silly, it is irresponsible. 

But in business, causality may be less obvious, leading even the wise man to misinterpret a strong correlation with causation. As we seek to establish causation, we are trying to identify the connection between an investment and a business effect, or determine which metric drives another. Ideally, we can identify the investments that positively impact our business metrics.

Most analytical studies offer an approach that blends statistics with the best-articulated case the chain of evidence provides. In addition to a deep dive into as much data as can be accessed, we look for anecdotal evidence, success stories, and any other supporting evidence that tells the same story as the math and logic. Ultimately, proof is a combination of mathematics and old-fashioned reasoning. 

For more on this topic, read my latest book “Optimize Your Greatest Asset – Your People.” 

Everybody’s Doing It – Analytics That Is.

The adage from your mother about everybody else jumping off a bridge simply doesn’t apply here. The other functions in your organization (marketing, operations, finance, and others) leverage their data and advances in analytics to make more intelligent, strategic decisions. Every time you scan your customer loyalty card at the supermarket, you’re exchanging valuable information about your shopping habits in exchange for that buy-one-get-one deal on Cheez-its. Marketers analyze patterns in the immense amount of data they collect to figure out how to get you to spend a little more on your next trip. This is predictive analytics.

Think about supply chain management. An automotive manufacturer bleeds cash for every minute their assembly line shuts down because they’ve run out of a car part. Using analytics, they manage their inventory to assure this doesn’t happen, while simultaneously avoiding an overstock on a costly component. This, too, is predictive analytics. Imagine having a piece of traditional capital in a manufacturing facility and not realizing its full potential; this is what we do with our talent. The point is, these functions have figured out how to harness the power of analytics to help them work smarter.

Why not apply that same methods and science to the organization’s most valuable asset—its people?

By applying analytics to investments in people development, HR leaders know that they are deploying programs that work. They can fine-tune investments to their population’s needs. Employees appreciate programs that give them what they need and want on the job. When people are given the skills, tools, and knowledge they need to perform at their best, it’s a win for everybody involved. Companies are rewarded because they are maximizing their investments in human capital. Employees are engaged because they are successful on the job. And customers are satisfied because companies that run like a well-oiled machine can fulfill their needs and exceed their expectations. The company is growing because its satisfied customers drive an increase in revenue. And the cycle continues, ad infinitum.

Our KPIs are a F@*king Mess


Words matter.

I discussed it in my last blog on why this is so important.

The human resources industry for decades has been trying to get validation in the C-suite, let alone at the board level. The vocabulary of the CCO, CFO and COO is in business terms and those terms are well established, clear and concise. KPIs (key performance indicators) are not that.

There is tremendous confusion around what exactly is a KPI and a wide variety of definitions, and rightly so. From Wikipedia:

KPIs can be summarized into the following sub-categories:
• Quantitative indicators that can be presented with a number.
• Qualitative indicators that can’t be presented as a number.
• Leading indicators that can predict the outcome of a process.
• Lagging indicators that present the success or failure post hoc.
• Input indicators that measure the amount of resources consumed during the generation of the outcome.
• Process indicators that represent the efficiency or the productivity of the process.
• Output indicators that reflect the outcome or results of the process activities.
• Practical indicators that interface with existing company processes.
• Directional indicators specifying whether an organization is getting better.
• Actionable indicators are sufficiently in an organization’s control to effect change.
• Financial indicators used in performance measurement and when looking at an operating index.

Wow. Clear and concise. Ya think?

When we developed the methodology behind what eventually became Vestrics, we eliminated KPI from our vocabulary. In thinking about following the evidence to an outcome we first looked at leading indicators that linked into business outcomes. Simply stated, leading indicators are those metrics that do not carry a financial value, but show evidence towards a business outcome. A business outcome is defined as one which carries a financial value.

Leading indicators are evidence of things to come such as engagement scores, hire fill rates, performance reviews, numbers of complaints. Business results carry a financial value such as revenue, retention/turnover, workers comp costs, cost avoidance.

When we made this change we saw people from different areas, backgrounds and titles understand these definitions and began to speak the same language, in business terms.

The line of sight to business outcomes becomes so much clearer by eliminating KPI from the vocabulary. Try it in a small group and see what happens – I dare you.

Words matter.

Storytelling is Killing People Analytics’ Credibility

Words matter.

The use of the word storytelling has creeped into our vocabulary and that’s not a good thing.

Words matter.

From Wikipedia: Storytelling is the social and cultural activity of sharing stories, often with improvisation, theatrics, or embellishment. Stories or narratives have been shared in every culture as a means of entertainment, education, cultural preservation and instilling moral values.

Is that how the findings from an analytical study wants to be viewed? I don’t think so.

Words matter.

The human resources industry for decades has been trying to get validation in the C-suite, let alone at the board level. The vocabulary of the CCO, CFO and COO is in business terms. Unless you work for companies such as Disney or Electronic Arts – storytelling is not generally part of the business leaders’ vocabulary.

For decades, the HR industry has been bitching about a “seat at the table,” while also being comfortable collecting evidence primarily around surveys. Correlation or not? Do you remember the movement in the training industry to adopt ROE – return on expectations, because getting to ROI was too hard? No wonder the C-suite doesn’t believe the profession is strategic.

Don’t get me wrong. I’m been an advocate for better analytics communications since I started this work in 2004, and have written extensively about it. We learned early in our journey that communicating complicated information (say that three times) was more effective using visuals with few words vs. reams of words and calculations.

Last year I wrote a blog on this subject.

“I recently read Cole Nussbaumer Knaflice’s book Story Telling With Data, a data visualization guide for business professionals … It is the best book I have ever read on communicating data with clarity.” https://www.linkedin.com/pulse/power-story-gene-pease

I understand the need for the use of the word within our profession, and continue to applaud all efforts to better analytics. However, in recent years just about every analytics presentation I have seen includes some segment on storytelling.

I know some of these slides have been presented to the C-suite. Do we really want them to think we are telling stories with the data?

Words matter.

The Big Data Conundrum

Big data comes into play when you combine data from multiple sources, systems, and departments. As this begins to happen, you gain insights into the organization that you cannot see by looking at a single source. By combining HR demographic data, training data, engagement scores and the like; with operational and performance data, you’re able to segment the workforce to uncover patterns and trends not easily recognized otherwise.

Death by dashboard was discussed in an earlier post, but basically means that organizations can’t practically use all of the information they have. It comes from disparate sources, much of it is irrelevant, and it’s not synthesized in any useful way. An executive in this position, considering a whole bunch of discrete data sources, is actually ignoring big data. It may seem counter-intuitive: we know there are about a zillion data points out there. Advancements in analytics and technology now allows for the ability see useful integration of all these data points. It’s combining a Facebook profile with the information captured by the customer loyalty card to deliver the right advertisement at the right moment (more about the “creepiness” factor of this in later weeks).

And without looking at the operational performance data, you can’t do any of this analysis; furthermore, the operations team can’t do this analysis without the HR data. While there are often political hurdles to obtaining data you don’t own, it’s worth the effort to overcome them. Big data is more than just data in organizations—it’s a significant cultural change to break down data silos, forge relationships across the organization, and brings together information from disparate sources. As a reward for your efforts, you receive actionable intelligence to help you control and prepare for the future.

Alternative Facts: Correlation Does Equal Causation

Most dashboards/benchmarks/scorecards present data that is correlated. When two variables change together in some consistent way (e.g. training and improved performance), they appear to be related and are said to be correlated. Survey-based methodologies for evaluating investments can establish a correlation between the investment and the recipient’s performance. Many of you have heard the expression “correlation does not equal causation.” When two things appear to be related, it’s important to investigate whether other factors may be involved. To establish causation an investment’s impact on the business must be isolated from all the other variables that also maybe impacting the outcome.

For example, there is a clear-cut relationship between ice cream consumption and murder rates—as more ice cream is consumed in the United States, more people are murdered. Is it fair to say that eating ice cream increases the likelihood that one’s life will be violently cut short?

These events are clearly correlated, but digging further you will see high temperature is the cause. In hot weather, more people congregate outdoors, increasing the likelihood of conflicts leading to murders. These links are all correlations and thus do not say much about actual causes. To assert that ice cream sales have a causal effect is not only silly, it is irresponsible. Tyler Vigen in his book Spurious Correlations has multiple correlated examples, such as the example above: The number of people who drowned by falling into a pool, and the films Nicolas Cage has appeared in. It makes no sense.

In business, causality may be less obvious, leading even the wise man to misinterpret a strong correlation with causation. As we seek to establish causation, we are trying to identify the connection between an investment and a business effect, or determine which metric drives another. Ideally, we can identify the investments that positively impact our business metrics.

Most analytical studies offer an approach that blends statistics with the best-articulated case the chain of evidence provides. In addition to a deep dive into as much data as can be accessed, we look for anecdotal evidence, success stories, and any other supporting evidence that tells the same story as the math and logic. Ultimately, proof is a combination of mathematics and old-fashioned reasoning.

For more on this topic, read my latest book “Optimize Your Greatest Asset – Your People.”

Death by Dashboard

Do you have so much data that you’re completely overwhelmed to the point of being frozen with a big pile of stuff? This is what I call “death by dashboard.” This doesn’t mean to imply that dashboards are inherently dangerous, bad, or worthless. Quite the contrary.

Dashboards are a critical part of the continuum and repositories of the basic data needed for an analysis (for the purposes of this discussion, I lump scorecards, benchmarks and other such reporting tools in with dashboards). They offer a snapshot of the organization at any given point in time.

 They take their name from the dashboard of a car, which gives you a picture of your vehicle’s state—speed, RPMs, remaining fuel, temperature, etc. What a car dashboard fails to tell you (although they are getting more and more sophisticated all the time) is how all this information relates, and what is happening in the world around the car. The car’s fuel may be low, but only the driver can evaluate whether this is an emergency (stuck in traffic on a steaming hot day) or just a task that must be accomplished soon (cruising down the highway with several gas stations coming up in the next ten miles).

 Our corporate dashboards work in this way. They can show a variety of HR metrics in one convenient, easy-to-read place, but they don’t tell you how these metrics relate (if they are related at all) and whether you need to act on any of them. One of the dangers of looking at your HR data in this way is the temptation to make decisions based on correlations.

Suffice to say correlation does not imply causation.

 The problem with dashboards is that they present information in such a way as to make it seem like they are giving a complete picture and offering answers. Many of our analytics clients tell us that their business units are asking for these dashboards. They receive ongoing requests for ad-hoc reports and customized dashboards, resulting in many streams of information coming from one set of data. One client created over 50 dashboards in five years! When they began to investigate these reports and dashboards, they found they were used sporadically, if at all. In the end, it was a significant waste of time and resources to create and maintain them.

So, are your dashboards being used and help make better business decisions?

Follow Your Gut or Follow the Data?

For the past dozen years I have had the privilege of collaborating with a group of incredibly smart scientists who have helped shepherd advanced analytics into the human resources profession. Look at where we are now a decade later. There is significant proof that applying analytics and understanding how investments are working significantly improves business outcomes. We now use predictive analytics to help navigate rapidly changing work environments. Big data allows us to capture both structured and unstructured data and turn it into information that enables us to make better-informed decisions.

HR is playing catch-up with adopting analytics, as significant research shows. Some of us estimate the HR profession is about where marketing was a decade ago regarding the use of analytics. According to Bersin by Deloitte, only 14% of human resources departments have an analytics function. This compares to 77% of operations departments having an analytics function, 58% in sales, and 56% in marketing. With all the evidence we have on the value of analytics, HR has to do better.

I have also had the good fortune to collaborate with thought leaders and write three books on people analytics published by Wiley and the SAS Business Series. In my writing I have tried to present advanced analytical work with a combination of theory, framework and application. Case studies at leading organizations were presented in the first two books. For my last we conducted interviews with leading HR analytics professionals and sprinkled their experienced insights throughout the book and blog. By showcasing them I am hopefully it will help you overcome any obstacles you or your organization may have to join the HR analytics movement. If you do, I promise you won’t regret it.

So follow your gut, or follow the data?

Gene Pease

 

HR Slim Down, Do More

 

SlimDn

Since the 2008 recession, HR leaders are all too familiar with directives to do more with less. As the economy has slowly improved, many organizations have more resources available to develop their people, but the need to eliminate and avoid waste is still top of mind. In very simple terms, analytics shows you where your investments are working and where they aren’t. It’s critical intelligence for a budget of any size, in times of boom and bust. A.D. Detrick, formerly  learning Measurement Consultant at Xerox Learning, is one of many forward-thinking practitioners using employee data to work smarter:

“By having granular data on both user demographics and user behavior, we can closely follow where institutional knowledge resides within an organization. We can identify clusters of skills and gaps in knowledge. We can foresee threats posed by generational shifts or technology changes and work to remedy them before they actually have an impact. And we can expand our reach instantly across the globe to enact that change.”

Multiple studies by Deloitte, i4cp and Bain have proven that the more advanced an organization’s analytics capabilities, the greater the margins by which they outperformed their competitors.

Suffice to say that HR analytics can directly impact a company’s bottom line.

You can generally assume that your employees want the tools, knowledge, and resources to do their jobs in the best way possible. It’s up to you to figure out how to equip them with exactly what they need. This is the crux of argument for analytics. If you can understand where your investments are working (and where they aren’t), you have targeted intelligence to give your people exactly what they need, when and where they need it, in a format that makes it as easy as possible for them to take advantage of it.

Gene Pease

This subject, and many others, is explored in depth in my third book, released by Wiley.

http://tinyurl.com/q6k6474