Put simply, data literacy is the ability to derive meaning from data. That seems like a straightforward proposition, but, in truth, finding relationships in data can be fraught with complexities, including:

• Understanding where the data came from, including the lineage or source of that data.
• Ensuring that the data meet compliance requirements to eliminate the possibility of bias.
• Preventing the introduction of personal information into business-critical models.
• Knowing how to track the data’s use.
• Having a clear understanding of what the data is being used for.
• Providing a visual of the data that is easily interpreted.

These are just a few of the current demands associated with achieving data literacy, a critical role in delivering accurate and timely results for your business.

Comparing literacy types

Since September is National Literacy Month, I’d like to compare data literacy to other forms of literacy. I’m married to a writer, so I know a first draft of writing looks nothing like the final piece. Likewise, in the first draft of any kind of data analysis, the early use of data sources may have inaccuracies and could require further data cleansing.

Ensuring that data is accurate and meets business requirements can be compared to the editing process in writing. Also, making sure your analysis meets the stated business goal before sharing it for consumption can be compared to a final edit.

Similarly, delivering new versions of a data model based on trend analysis is like the process of revision. It’s iterative and designed to make the output even more relevant.  And just like any other form of information delivery, data literacy should include a process for proper analytics, accountability and metrics.

Ideally, schools should teach data literacy in the same way they teach reading and writing literacy. As our world becomes more data driven and complex, we can all benefit from the skills needed to interpret data at work, at home and in the news.

Growing with data literacy

What does the future hold for data literacy? Like all processes, there’s a maturity cycle that develops around data management. As a subset of data literacy, data management includes data versioning, data utilization, data lineage, data cleansing, scanning for data bias and data nutrition. Recent MIT research describes data nutrition as a process that incorporates many of the requirements above into a rating similar to the way in which food is rated.

Converging data management with model management represents one of the most exciting business opportunities of the future. Model management is the process of creating, validating, deploying and monitoring analytics models. The merging and blending of these two disciplines should further eliminate bias, which will give businesses a much clearer picture of model output.

These types of approaches are designed to give decision makers a more complete understanding and comfort with the results that are delivered by data scientists and data engineers.

As information consumers, we all want to be confident that our analytics results are fair, accurate and unbiased. Part of the onus is on us to interpret and understand data, but the more advanced we get with data management and model management, the more faith we can put in the data and the results themselves.

To read more about data literacy, read A skeptic’s guide to statistics in the media or check out the inspiring work in GatherIQ, a SAS application that uses data and analytics to support world efforts to achieve sustainable development goals – and teaches data literacy in the process.

Celebrate data literacy with free SAS e-books