REDCap: Don’t try to combine multiple questions into one field

Double-barreled questions

Separate questions need their own fields. Avoid using double-barreled questions—these make it difficult for users to provide good data, because users must try to provide one answer to two questions. This means avoiding multiple questions within one field like, “Have you seen a doctor in the past month and had a glucose test? Yes/No”. This field attempts to capture two data points at once: did you see a doctor in the past month? Did you have a glucose test in the past month? A participant might have an easy time answering yes if they saw a doctor and had a glucose test, but what about if they saw the doctor (yes) and didn’t have a glucose test (no)? Or had a glucose test at a health fair (yes) with no doctor present (no)?

Takeaway: Split up multiple questions into multiple fields and use branching logic if needed to make them flow logically.

Mixed units of measure

Similar to double-barreled questions, mixed units of measure within a single field can be confusing for data entry and can require lots of cleaning before you can even think about analysis. Think about capturing a pediatric age in years and months within one field: allowing sometime to enter “1 year 4 months” does not lead to easily analyzable data, especially if some data points are entered as “1”, “1.4”, “1:4” and “1 year 4 months”. This can result in false precision, in which it appears the subject who is marked as being 1 year 4 months is a different age from the kid whose age is marked as “1”, when really both could be the exact same age and simply have had data entry staff recording their ages with differing levels of precision. Capturing gestational age is another common headache- you might see an entry like “40 weeks 2 days” within one field.

Just as double-barreled questions ask two questions within one, these mixed units questions are trying to capture two variables in one. Don’t! If you need a level of precision at years AND months, then have two separate fields, one for years and one for months. Any well-trained statistician or data integration analyst can easily combine two or more fields in order to conduct analysis. If that method is not preferred, you can consider capturing all ages in one consistent unit. For example, if you have one participant who is 4 months old and another who is 5 years old, you could record all ages in months so that data would be entered as 4 and 60, respectively.

Takeaway: If you need a certain level of precision, require it! Make it hard to mess up data entry. Do this with separate fields for each different unit of measure. Or, decide on one unit of measure and stick with it.

Discover More!

Want to peruse other REDCap topics? Check out: