Correlation Is Not Proof: Gaining Legitimate Insight from Learning Data

Conference Archive

Learning Solutions Conference & Expo 2019 - March 27, 2019

L&D has wrestled with the idea of measuring impact for decades. Unfortunately, most of the methods used as “proof” of impact are simple correlations. And most people know the adage “correlation does not equal causation.” This session will help participants understand different, more practical types of analytics, and how they can provide much more impactful insights than simple correlations.

This session will examine practical examples of data sets, and how insights can differ greatly based on the analysis chosen. Using the same data sets for different analyses, you will learn how simple analyses lead to broad—and often dangerously inaccurate—insights. By learning how to best structure data for analytics—using simple, everyday tools—you will see the benefits of proper analysis and the dangers of a simplistic analysis of the same data. Most importantly, you will then see how logically a strategy for performance improvement flows from proper analysis—and how easy it is to continue improving your organization.

In this session, you will learn:

• About the risks of using correlation to “prove” impact—and how common that is
• The difference between relative, comparative, and distributive analytics (in plain English)
• How to identify statistically significant differences
• The difference between a statistically significant event and a fluke
• Practical steps to identify L&D overachievers and underachievers
• How to easily craft a prescriptive strategy from those findings

Audience:

Managers and senior leaders (directors, VP, CLO, executive, etc.)

Technology discussed in this session:

Microsoft PowerPoint and Excel

Handouts

You do not have access to these files. Please log in or join with a free membership to download these files.

 

 Back to Library