Overview
Surveys are typically fielded in waves, which are distinct periods of time in which answers are collected from respondents. From a data processing perspective, a wave is an entire dataset that is added to a tracking study. These are best represented as Categorical variables in Crunch (rather than date/time variables).
The following article describes how you can add additional metadata to your categorical variables so that they can be used in time plots and time series analysis.
Why do I need categorical date variables?
You need categorical date variables for several reasons. Let's begin by discussing how survey information is captured.
The information relating to when a survey was done is captured in one of two ways:
- Membership to period is captured in a categorical variable.
- For example, Wave 1, Wave 2, Wave 3, or Quarter 1, Quarter 2, Quarter 3, and so on.
- The exact time of the interview is captured in a date/time variable.
- For example, you know the specific end-survey or start-survey time, e.g., 2012-01-03 23:00:20. It may not include the timestamp, just the date.
Date (including date/time) variables give precise time information about each case (respondent). That precision may be counter-productive if the exact dates don’t align neatly with your intended analysis. For example, fieldwork for Wave 3 runs between 2019-12-01 and completes on 2020-01-02, but you really only want to consider it as “the December wave” rather than “December give or take a few”.
This is where a categorical variable with dates associated is better suited to the analytic task, as you can clearly define ranges of time (as categories).
However, categorical variables on their own have no concept of time. There is nothing about a category called Wave 1 or Quarter 1 that tells us when the fieldwork occurred. As a result, categorical variables can’t be used for time plots (e.g., line charts, smoothers, or other time-series analyses). By changing the categorical variable to “Subtype: Date”, you can define a date that corresponds to each category. For example, Wave 1 represents January 2020, with date “2020-01” and Wave 2 represents 2020-02.
How to add date metadata to categorical variables
You can create categorical wave variables using the Crunch web application, R, or with Crunch Automation.
Using the Crunch web application
- Sign in to Crunch.
- Go to the Properties of the categorical variable.
- Change the 'Subtype' to Date.
- Instead of numeric values, there is an input field for Date.
- For each category, enter the desired date in the format YYYY-MM-DD. Alternatively, and more commonly, enter YYYY-MM to represent a month.
- Save your variable.
Using R
See the following dates(Categories)<- Crunch reference:
Using Crunch Automation
Use the SET DATES command. See the following article for more information:
What about date ‘rollup’ with categorical date variables?
Categorical date variables do not ‘roll up’ dates like date/time variables. With date/time variables, you can change the roll-up flexibly between day, week, month, year.
If you want to do a roll-up of your categorical variable, you can create a new categorical variable (e.g., using Combine Categories on the categorical variable) and then add the respective dates to the new variable as per the process described above.