Meeting an error screen is inevitable for programmers. While there is plenty of information on how to solve your coding and analytic problems, there isn’t as much information on how to prevent them. In this blog post, I’d like to offer an argument for using data dictionaries to prevent basic errors.

When I started working with CMS claims data, I had a bad habit of assuming variable formatting after previewing raw data. I remember assuming that provider identifiers were numeric after reviewing 50-60 rows. Confidently, I ran my code against the entirety of the data and, when it finished running the next day, I found plenty of error messages waiting for me. Had I known to check the data dictionary first, I would’ve seen that the variable field was character formatted.

Data dictionaries serve as an excellent planning and troubleshooting resource. Beyond general naming and content information, they can also contain important information on data nuances and changes over time. For example, in the CMS RIF data dictionary, the diagnosis code’s comment section includes information on the mid-year 2015 transition from ICD-9 to ICD-10 codes. New programmers may not know this and accidentally use one ICD version, which would mean missing out on diagnosis for part of the year.

Although we can’t stop every error screen from coming our way, I hope that my short anecdote helps you consider adding the data dictionary into your resource toolkit while you are preparing to analyze data. Don’t be afraid to double-check before you send a job. Use your data dictionaries!