Once you've created a basic version of your data dictionary, you're ready to add variables to it. For NSRR, all variables and domains are stored as individual JSON files, which look something like this:
[
"id": "variable_id",
"display_name": "variable_name",
"description": "A description of the variable",
"type": "vartype",
"units": "optional",
"calculation": "optional",
"domain": "optional",
"forms": [
"variable_form1",
"variable_form2"
],
"labels": [
"label1",
"label2"
],
"commonly_used": true/false
]
Of all the modifiers above, the ones that are always required for a variable is id, display_name, description, and type. Depending on the variable type selected, units or a domain will also be required. Valid variable types are listed below:
As specified above, variables that are type: choices require a domain. Domains lists of possible options for variables that would have pre-defined choices (like a dropdown). Domains are stored in a separate JSON file, and get referenced in the variable file.
[
{
"value": "1",
"display_name": "Option name 1",
"description": "A description of the option"
},
{
"value": "2",
"display_name": "Option name 2",
"description": "A description of the option"
},
{
"value": "3",
"display_name": "Option name 3",
"description": "A description of the option"
}
]
Any variable, regardless of type, can be associated with labels or forms. Labels are used to add additional tags to variables that are not found elsewhere in the JSON file that you would search for on sleepdata.org. The forms field allows you to group variables together, such as variables from the same CRF, when creating a codebook. More documentation and examples of variables and domains can be found at the bottom of the Spout README on GitHub