Tracking studies follow a topic by repeating the same survey questions to see how data points change over time. In Protobi, you can create a tracker project to compare data across wave with statistical testing, visualize trended charts and even automate the repeatable task of producing deliverables on a monthly, quarterly, etc. basis. This tutorial shows our recommendations on how to keep trackers organized and links to more detailed topics like how to create a new workspace for a tracking study.
How Protobi supports tracking studies
- Show multiple waves of data in one workspace with WoW comparisons and statistical testing
- Flexibility to update survey data as often as you like
- Create and automate deliverables like quarterly PowerPoint reports or large, stylized Excel tables with statistical testing
- Share Protobi with your clients as a dynamic, online tracking dashboard
- Data cleaning/processing capabilities that allows you to document changes to prior waves and apply the same changes to new waves
New workspace versus update existing
Depending on how often you will update data and produce reports, you might wish to update data yourself and retain the same workspace or create a new independent version of the project with each wave.
For trackers conducted less frequently, there tend to be more survey changes and clients will ask us to set up a new workspace and update the data on their behalf:
- Quarterly
- Bi-quarterly
- Annual
At the start of each new wave we create an independent workspace so the project view can evolve as your analysis does, but the prior wave's project workspace stay the same. See Create a new workspace for a tracking study for detailed instructions on how to set up a new workspace yourself.
Clients tend to update data themselves and keep the same workspace for these types of trackers:
- Continuously fielding
- Weekly
- Monthly
See Combine multiple waves of data on how to add a new wave of data into an existing project. You can also choose to upload a stacked data file with all waves.
Manage changes to the survey
Change log
Over time tracking studies evolve. Some waves will have marginal changes and others significant changes; make sure the updates are communicated and documented (i.e. change log). A detailed change log spreadsheet is helpful to instruct the Protobi team in how to set up the new questions in your tracker. We will link the spreadsheet as a Google sheet that is accessible from your project.
Notes
Your team can also add notes on each element to flag changes to the question WoW.
Tips for recoding open-ends in tracker projects
Open-end coding can pose an additional challenge because different waves might have slightly different code frames.
Add new codes instead of renaming previous codes
Make sure to add new codes to the existing code frame versus renaming previous codes. Renaming old codes might causes confusion when looking back at the results from previous waves.
Make sure you're coding values from the current wave
In the past, users have reported changes to the results of previous coding after they code values in the most recent wave. This can happen if there are stray uncoded values from previous waves, and users do not code the values until a later wave.
Code current wave from simple recode tool
To check that all uncoded values are from this current wave, filter on the current wave and view uncoded responses from the simple recode dialog.
You can filter using press to query or global filters.
Code current wave from advanced recode tool
The advanced recode tool does not adhere to filters within the project. If you prefer to use the advanced recode tool, set the current wave's data file as the Primary source for the project so that you only see the current wave's uncoded values.
Tips for the survey programmer
Do not reuse question names
Make tracking questions easier by retaining question numbers across waves. Adding a new question or dropping an old one in new waves is fine. However, you should not re-use previously used question numbers and column names for different questions.
Maintain data structure for legacy trackers from other vendors
If you inherit a tracking study from another vendor, keep the same data structure and layout as the legacy tracker. Question numbers, data values and positions should align with the legacy tracker.
Add a "wave" variable
Program a "wave" data column into your survey that will be consistently used in each wave. This will make setting up wave filters simpler, but if you don't have a "wave" variable in your survey we can create one in Protobi.
Simplify data cleaning/processing
Start from raw data
Start each wave with data straight from the survey engine and apply business rules and data cleaning within the tool. If you already invested time and effort in data processing in some language like R or Excel, we can work with the processed data. Otherwise, you can process data right in Protobi using Javascript. This way all the data cleaning is documented in one source and you can easily carry the code over to each subsequent wave.
Add comments to process code
Add comments with dates to each block of code in your project’s data process so that other analysts know what each line of code does. Review the process every wave to remove code that is no longer used.
Run code on new, unprocessed data only
For tracking studies we always stack the new wave of data with processed historical data that is already in Protobi. There's no need to run the same process code on historical waves, you only need to carry over the code if you with to run it on the new wave of data.
Avoid duplicate blocks of code
If you need to add new code that applies to the current wave as well as historical data, add the code after the Protobi.stack_rows function instead of repeating the code for the wave 1 loop and the wave 2 loop separately.
Example process code
This is an example process for the second wave of a tracker in a new workspace.
/**
* @by tiffany@protobi.com
* @created 2023-06-01
* @see (enter URL of Freshdesk support ticket or other reference)
* This process assigns wave values and stacks the W1 processed data with the W2 data
* The W1 data table has the already processed W1 data from the W1 project workspace, the W2 table is the raw W2 data file
*/
var W1 = data["process_w1"] //define prior wave file
var W2 = data["main"] //define W2 (current wave) file
//TL, 2023-06-01, data from W1 is assigned wave = 1
W1.forEach(function(row){
row.wave = 1; //assign a wave value if it's not already programmed in the survey
})
//TL, 2023-06-01, data from W2 is assigned wave = 2
W2.forEach(function(row){
row.wave = 2; //assign a wave value if it's not already programmed in the survey
//TL, 2023-03-01,Code to create productx_enthusiast flag (Carry over each wave)
if (row.S10 >= 6 && row.S6_5 == 1) row.productx_enthusiast = 1
else row.productx_enthusiast = 0
})
var rows = Protobi.stack_rows([W1 ,W2]); //TL, 2023-06-01, stack prior and current waves of data
//TL, 2023-07-13
//Create new prescriber variable (Carry over each wave moving forward)
rows.forEach(function(row){
if (row.s1 == 3 && row.s4 > 0) row.prescriber = 1
else row.prescriber = 1
})
return rows; //return the resultYou can add processes to:
- create a new variable
- remove respondents
- combine waves
- merge in translations
- stack patient cases, choice cards, etc.
- combine
monthandyearinto a single date
Use Protobi as an online deliverable
For your clients that like to dive into the data, set up a polished dashboard with trended charts and tables to share.
Trended charts
Set up charts in Protobi that exactly parallel your PowerPoint report, so that your clients can easily navigate them, and your analysts can easily update and QC the report each wave.
Stylized tables
You can also display questions as large tables that are comparable to a traditional tabs package.
Use Protobi to create offline deliverables
Protobi can export your entire survey or a subset of questions to Excel or PowerPoint. Excel export includes standard tables and large stylized tables. Online charts export to PowerPoint as native objects.
Export to PowerPoint
Most charts will export as native, editable chart objects with the data embedded inside an Excel file.
Automate updates to PowerPoint slides
You can even choose to update the data in an existing PowerPoint report with the numbers in Protobi without changing other aspects of the PowerPoint, allowing you to automate reports for every wave, market, etc.
Upload a polished PowerPoint report from the previous wave and use it as a template for later waves, linking charts so that Protobi updates PowerPoint charts and tables in place. Contact support@protobi.com to learn more.
Export as stylized tables
You can selectively export questions as stylized Excel tables, or export your entire survey for a full crosstabs report.
Export as standard tables
Exports to Excel as tables will generate data tables in both counts and percents, but these tables do not support as many crosstab banners as the stylized tables.
Cost
Costs are based on number of workspaces and support hours.
The typical baseline wave of a client-facing project with stylized charts is $3,200 (Extended support bundle) which provides 10 hours of expert support.
A typical baseline wave of a project for internal use will start at $1,400 (Expert support bundle). This includes a shared workspace and 2 hours of expert help in setting up the project.
Charges for subsequent waves depend on the survey changes WoW and whether or not you want to create a new workspace for every wave.
Tracking reports with custom charts specific to your report and automated PowerPoint will have additional costs. Contact support@protobi.com for a quote on extended support costs.