What Is a Triple S Data Format?

Bluffers Guide to Triple S Format

Introduction

Trile-S is a format defined by a consortium of Market Research agencies and other interested parties for the transfer of market research data. The format goes back to mid 90’s and has been refined over the years, but is now fairly static, the last major update was in 2006.

Unlike SPSS or Excel it is not a single file which can be opened by an application. However, it is quite easy to work with the two files that make up a triple-s data set as they are created using common formats.

Basics

From the Forsta Visualizations manual:

• The data file must be of “Fixed width”, Triple-s files in CSV format is not supported.
• Multiple choice questions can either be binary or categorical in the imported file.
• If the imported Triple-S file contains multiple languages the first language in the file will be used.
• In the Triple-S standard a third file showing hierarchical structure between groups exist, this file is not supported, instead the hierarchical structure is defined within Forsta Visualizations.
• In the Triple-S standard the values (answer alternative codes) can be alphanumeric, Forsta Visualizations only supports numeric values.
• All case data in categorical variables that have no corresponding value labels in the meta data will automatically be deleted during import and not considered in any calculations
• All variable texts, value label texts and variable type (Single choice, Multiple choice, Numeric and String) are picked from the imported XML file.
• Text can be updated within Forsta Visualizations
• Question type cannot be updated within Forsta Visualizations
• All values/value labels in the XML files will be imported to Forsta Visualizations, even the standard codes like 99999998 and 99999999 which are used as Missing values in Triple-S standard

File Contents

The meta data is stored in an XML file (*.xml) and the case data is stored in a plain text file. (named *.asc). We do not support the comma separated version of the case data file.

You upload of triple-s as a zip file. To work with the individual files, unzip the two files, then you can examine the contents and even make updates.

The XML file contains the definitions of the variables in the survey. This is a repeating loop of the settings for each question in turn in the project. If you open the XML file in a browser it will render the XML in a friendly way.

This is a list of variables with name, label, start and finish position in the flat file and the allowed values

Here is a single choice question with 11 answer alternatives, in the case data this will be a number from 1 to 11 in columns 93 and 94 of the text file:

Casedata

To view the case data it is best to use a “better” text editor than the built in editor with the operating system, a good one to use is Notepad++

To view each respondents record in one line turn off word wrap option:

If you want to view the data per variable, and you have quite a bit of time to spare, you can manually import a fixed width file into Excel.

This is easier using the legacy text import tool in Excel.

To set this up in Excel, Click on “File” –> “Options” –> “Data” and set the corresponding checkmarks for reactivating the From Text (Legacy) in Excel.

Then start the text import by clicking on “Data” –>”Get Data” –> “Legacy Wizards” –> “From Text (Legacy)”.

Select the file and set the column widths as per the XML file.

Here is XML – look at the “finish” values:

Here is the fixed width text import, with delimits set at 4,5,6 and 24 etc

It’s unlikely you would map an entire file this way, but it can be useful to grab specific variables from the file.

Problems

Error in metadata : check the validity of the metadata using an online tool.

File takes too long to upload : the zipped files can have a large number of respondents, create batches of respondent data instead, by splitting the ASC file

Error in case data – for instance unexpected duplicates. Check the ASC file for unexpected characters on the lines for which you receive an error (lines nearly always start with resp id, so search for the respondent id that was reported in the error first, then check for strange characters in that line.) In Notepad++ they will look like this: