[December 09, 2000] Triple-s is an XML-based "open survey interchange standard" for the encoding and interchange of survey data collected and analyzed by social science professionals. The standard "defines a means by which both survey data and meta-data (variables) may be transferred between different survey programs running on different software and hardware platforms." The domain problem is typical: "Increasingly, users of survey software are demanding that data be exchangeable between survey software systems from different vendors and possibly running on different hardware and/or different operating system platforms. The transfer may be required because an client wants to perform some more detailed analysis of aspects of a survey originally conducted by an agency and the two parties use different survey software." Thus, triple-s has been designed "as an interchange format; it was not conceived as a native survey definition format, nor is it a replacement for the many proprietary survey definition languages currently in use. The triple-s XML format provides for the cross-platform transfer of both survey data and survey variables using universal industry standard protocols. The syntax of a triple-s XML document is described by the freely available triple-s DTD. triple-s XML provides for the description of the five most common types of variable: (1) SINGLE variables interpret categorical data with one response allowed; (2) MULTIPLE variables interpret categorical data with any number of responses allowed; (3) QUANTITY variables interpret open numeric value (integer or real); (4) CHARACTER variables interpret character data; (5) LOGICAL variables interpret individual Yes/No or True/False data values. triple-s XML allows for both integer and real coded values to be represented. Two formats for the representation of multi-response data are supported. Where standard coding has been used to represent special values -- for example, where '9' is used to represent 'Not Answered' -- that coding is maintained through the transfer operation rather than being closed down on a question by question basis. Furthermore, the fact that a particular code is 'special' in some way can be represented and thus indicated to a survey importer. A triple-s survey is described in two text files. One, the Definition File, contains version and general information about the survey together with definitions of the survey variables. This is used to interpret the contents of the Data file..." [From the 1999 article]
"triple-s is a language for describing survey metadata, that is, data about the survey data: question and response texts, the location of individual data items within the record, valid ranges for responses, and so on. triple-s is a specification language, like Quantum or Snap in that does the same job -- describing the data for a survey -- and the information is stored in a text file, but triple-s files aren't written by spec writers: they're normally generated automatically by a program. If one program (say BellView) can translate into triple-s format, and another program (say Merlin) can translate from triple-s format, then the survey metadata can be translated from one language program to the other automatically -- without the need for a specwriter to do the job manually. [The XML specification addresses the problem of] writing imports from each of the packages, and having to change the import each time the competitor changes his syntax. Now the standard has been adopted by more than 50 software producers worldwide, we want to bring it into line with the XML metadata exchange format which looks likely to will become the standard for Web data interrogation and extraction over the next few years..." [Adapted from the FAQ document]
"The initial version of the triple-s standard (version 1.0) was devised by Keith Hughes, Stephen Jenkins and Geoff Wright, and published in 1994. The impetus was a paper by Peter Wills. During 1996 the same group of people met to enhance and extend the standard, based on comments from implementers and users. An interim result of these meetings was presented as a paper to the ASC (Association for Survey Computing) International Conference in 1996. The preliminary specification for version 1.1 of the triple-s standard was agreed in December 1996 and published in March 1998." [From the 2000-02 spec]
Project principals originally included key developers from commercial survey software houses: Keith Hughes, Stephen Jenkins and Geoff Wright. In early 2000, Paul Peterson (SPSSmr) and Ed Ross (from OpenSurvey) joined the technical committee.
References:
triple-s XML: The Survey Interchange Standard." A standard for moving surveys between survey packages on various hardware and software platforms. Version 1.1, February 2000. [cache HTML, see also .DOC]
Standalone XML version 1.1 February 25, 2000. Posted July 2000. [cache]
"triple-s XML: A standard within a standard." by Keith Hughes, Stephen Jenkins, and Geoff Wright. Paper given to Association of Survey Computing International Conference, Edinburgh UK, 1999. See following entry. [cache]
"Triple-s XML. A Standard Within a Standard." By Keith Hughes, Stephen Jenkins and Geoff Wright. In Social Science Computer Review Volume 18, Number 4 (Winter 2000), pages 421-433 (with 3 references). "The triple-s standard defines a means by which both survey data and variables may be transferred between different survey programs running on different software and hardware platforms. It thus represents a first real attempt to resolve what should be a simple, but was previously a complex process. This article gives a brief history of the development of the standard, outlines the formal aims of the original specification, and introduces key aspects of the standard and the revised version (1.1) published in 1998. The authors also look at the development of the Internet as a communication mechanism and the recent rise in the use of XML as a means of describing and categorizing the content of documents. The authors show how this aspect of XML leads to its logical choice as a metastandard for describing survey information and they describe the contents of the proposed triple-s XML standard."
"The triple-s Survey Interchange Standard: The story so far." By Stephen Jenkins' Paper given to Association of Survey Computing International Conference, London UK, 1996. [cache]
"Data Use and Reuse: The Movement and Use of Survey data across Different Hardware and Software Platforms." By Peter Wills. Paper given to International Conference of The Study Group on Computers in Survey Analysis, Bristol UK, 1992.
See also: Data Documentation Initiative: A Project of the Social Science Community
See also: "XML-Based Survey Server." Article by Scott Isaacs. "...we create a Survey Server where for creating multiple choice user surveys. Like our content server, we use XML to manage all the survey-related information. A subtle, but important difference from the content server is that we are now using XML to store data. As you read this article, consider how combining XML as a data store and data description format with XSL to present the data can revolutionize how you interact with your databases. Our Survey Server is a browser-based application. You create and edit surveys all from within your browser. Since the survey presentation is defined using XSL, you can quickly create new survey and result presentations. As we take you through the Survey Server, we will demonstrate the following techniques: (1) Storing multiple presentations in a single XSL file; (2) Querying XML documents for specific information; (3) Manipulating and transforming queried elements; (4) and manipulating XML using the document object model..."
