Discussion Forum

hi Mignote -

These are excellent questions.

Specific meanings for 8 (88, 888) or 9 (99, 999) in coded value lists

It's a common convention to use 9 for "Unknown" whenever coding a list of possible responses. If there are going to be more than 8 possible responses, use 99 or 999 instead of 9. This makes it easy for anybody else working with your data set to pick out the unknown values.

It's almost as common a convention to use 8 (or 88, or 888) for "Other." If appropriate, this value could open up a text box allowing the data entry person to give a brief free-text explanation of what the "Other" value is.

You can find out more about these conventions and others in our document "Coding and Naming Conventions" - see the link under Thursday July 29, or follow this link.

In general, you should never use spaces or special characters (other than _underscore_) for a field name. These can make problems in calculating fields; there are other reasons too. Many conventions (REDCap among them) also discourage capital letters in field names.

But be sure to distinguish "field names" from "captions." The caption is the text that the user sees describing the field - e.g. "How many living children does the participant have?" or "Age at intake". The field name is the system's internal reference to the field - e.g. number_living_children or age_at_intake. The end user doesn't see those field names, but you'll need to use them if you're developing and managing the data system.

For the sake of simplicity, we didn't have you use the "Complete?" field in the REDCap assignment. In your own system, it's entirely up to you whether the options here (Incomplete, Unverified, Complete) have meanings, and how they fit into your workflow. If you're working in a larger team, it can be useful. If you're the only user of the system, and if you know you'll never save a record until it's complete, it might be irrelevant.

In Assignment 2, you use repeatable instruments to capture longitudinal data points in REDCap. In Assignment 3, you use "longitudinal data collection with defined events." As described in lecture 3, either one can do the job. If there's a small fixed number of instances for your longitudinal measurements (e.g. perhaps a large number of data points, but always collected at four distinct time points in your study), then the longitudinal approach might be best. If there's an unknown or arbitrary number of time points for data collection, repeating instruments might be best.