Analyzing a tracking study
A tracking study is a survey that is used to monitor change over time. This type of survey is typically carried out as a series of “waves”, where each wave consists of one or more collections of response data. This section describes how you can use data management scripting and other UNICOM Intelligence and UNICOM Systems, Inc. technologies to analyze the response data from your tracking studies. The
Sample scripts used described are included with the UNICOM Intelligence Developer Documentation Library so that you can run them yourself.
Analyzing the response data from a tracking study often includes coding open-ended responses into categories and adding weights to the data. Reports are typically run against copies of the live data and might include reports for the wave that is currently in progress and reports for all completed waves.
The analysis process
The process described in this section uses the following products and technologies:
▪Responses are collected using UNICOM Intelligence Interviewer - Server.
▪IBM SPSS Text Analytics for Surveys is used to code open-ended responses.
▪Data management scripting is used to transfer data, merge data, and weight the data using the Weight component.
▪UNICOM Intelligence Reporter - Survey Tabulation is used to run reports.
However, you should be able to adapt the process to suit your own needs. For example, you might want to use a manual-coding tool instead of IBM SPSS Text Analytics for Surveys, or your survey might not require any coding.
The process uses three main data sources:
▪The live data, which is stored in an UNICOM Intelligence Interviewer - Server database.
▪The Wave database, which is created by a data management script and contains coded and weighted responses for the current wave of the survey.
▪The Report database, which is created and updated by data management scripts and contains coded and weighted responses for all complete waves.
The following diagram shows the data flow for the process:
When coding is not required
Some tracking studies do not include open-ended responses or do not require them to be coded. This section also describes an alternative process that only adds weights to the data. The following diagram shows the data flow for this process:
See
