As eLearning relies more heavily on data and adopts AI-based technologies, understanding how to process, analyze, and apply data is essential. Several of the skills that data scientists use every day would nicely round out an eLearning developer’s toolbox. L&D teams should include at least one member with basic data skills to enhance eLearning development and impact.
Defining the problem
Many eLearning design and development projects begin with (or should begin with) defining the problem. Rather than assuming that “a training” is needed, a data scientist would ask questions and get to the problem that the manager requesting eLearning is hoping to solve. Are learners failing to complete assigned or mandatory training? Are they completing the training but failing to transfer the knowledge covered to their jobs? Are they passing assessments but still making the same errors in performance? Or do they need to quickly master a new skill, understand changes to regulations, or gain fluency in discussing new or updated products?
An eLearning designer needs the answers to these questions to determine whether to create training or performance support tools—or whether an entirely different response is needed—and to figure out where the problems are with existing training and tools. A data scientist would convert the responses to those questions into data questions.
The data scientist might then look at some existing data: How many learners complete the training? At what point do learners drop out? Which areas of training are not transferring successfully to on-the-job performance? What are the most common errors employees are making, even after completing training? This process of analyzing the existing data can provide useful insights that improve the design of the new or existing eLearning tools. For instance, the data might reveal patterns that indicate that employees from specific geographic locations or in cohorts that started work at a certain time or that received different initial training are performing better (or worse). It could reveal patterns in the ages or job descriptions of employees who don’t complete their training. These are clues that designers and developers can use to build eLearning that targets and engages underperforming employees.
The data scientist could draw some conclusions and present them—along with suggestions for tweaking the design—to other members of the team. She could also dig deeper and examine the data more closely or gather some additional information. She might interview learners about how and why the eLearning is failing to meet their needs, for example. Whatever path is selected, gathering and analyzing data has added important information that might have otherwise been overlooked in the initial eLearning design.
Measuring eLearning success
L&D professionals increasingly must justify companies’ investment in training and performance support tools. One way to measure eLearning impact is by using data. But as it becomes easier to collect data, especially in xAPI-based eLearning tools, developers might find themselves with a sea of data that they’re not sure what to do with.
With some data science knowledge, though, those developers could identify metrics that would measure, for example:
- Training impact on performance
- Training impact on business results
- Value—ROI—of training
Gathering and analyzing data on key performance metrics enables developers to fine-tune their eLearning products, increasing the value of L&D within the company—as well as the value and productivity of employee learners.
Collecting data is not sufficient. The developer or data analyst must be able to visualize and explain to managers, executives, L&D colleagues, and to learners themselves what the data show. Data visualization skills are the key here. The data scientist identifies patterns and explains them, providing context and helping those who lack fluency in statistics to understand what the data reveal. She can create charts and graphs that present the data visually in an easy-to-understand format.
Specific data skills L&D pros need
Data scientists might delve deeper into programming, statistics, software engineering, or machine learning, but an eLearning developer who’s eager to add data skills could start with:
- Statistics: Data scientists should understand distributions, design and evaluation of experiments, and probability.
- Programming and data tools: Knowledge of SQL and statistical programming languages is a basic skill for data scientists.
- Algorithms: Learning how algorithms work and how to use them to predict performance is important; while software can do much of this, a basic understanding of the underlying mathematical concepts can help the data scientist avoid misinterpreting data.
- Cleaning data: Raw data can be messy. Identifying imperfections and “cleaning” data, for example, standardizing string formatting or identifying missing values, can improve the accuracy of the data—and therefore any information gleaned from it.
- Data visualization: Presenting data in a clear, comprehensible manner to colleagues, whether managers or fellow L&D professionals, makes the data useful and enables the data scientist to extract the maximum meaning and value from data.
In addition, developers should have strong problem-solving and logic skills and be able to figure out which data matters—and which is irrelevant—to a particular goal or problem.
L&D pros can learn these and other basic data skills in any of dozens of eLearning courses or through certificate and degree programs at their local universities and community colleges. In addition, The eLearning Guild published Putting Data to Work, a research report, in June and will host an online Data and Analytics Summit August 22 & 23, 2018, to explore the connections between data, analytics, and eLearning content.