Tracker studies are unique in that they allow users to look at trends in data over time. In order for trending to be available for a study, KnowledgeHound must receive all data across all waves in a single data file. Below you will find some best practices for formatting data for tracker studies - make sure to reference our article "Preparing Data for KnowledgeHound" for specifics around formatting of all data files. You can also reference our article "Making Changes to Data Files" for more information about updating existing studies in KnowledgeHound.

KnowledgeHound will work closely with the client/supplier to discuss the cadence in which data should be updated (quarterly, annually, etc.) and will make the appropriate updates on a rolling basis.

Best Practices 

  • Deliver cumulative files with data for all desired waves in a single data file
  • Unless changes to the survey have been made, the format of the file should remain consistent from wave to wave (variable names, question labels, response option labels, etc.)
  • Include a wave variable (whether it be year, quarter, month, etc.) that KnowledgeHound can apply as default breakouts to present trended data visualizations to users
  • For questions that remain consistent wave over wave, make sure all data for all respondents is captured in the same variable(s)
  • Inform KnowledgeHound of any changes to the survey between waves of data collection.
  • In instances where the question wording itself stayed the same, but response options were added or removed, please make sure to keep those response options in the data file/variable - do NOT delete them or create a new variable for them unless, for analysis purposes, they should not be viewed with legacy responses.

Items to Note
Questions with missing trending data: if there is a question that was added/removed or otherwise modified, KnowledgeHound will simply display this as missing data (see screenshot below), so long as the data is coded properly (i.e. set to missing vs. 0s).

Analyzing trended data: KnowledgeHound will default the chart view for tracker studies to a line chart to display trends over time. Because of this, a user cannot go in and break the data out by another variable - they must first either 1. filter the data to view the desired subgroup, or 2. turn the "wave" variable breakout off (resulting in data becoming a "total" view), and then apply the desired breakout - they can then filter the data using the "wave" variable to view a single timeframe of data vs. the total set.

Did this answer your question?