The Survey Interchange Format

1

• The Survey Interchange Format
• Standard
• Keith Hughes, Merlinco Ltd
• Stephen Jenkins, Mercator Ltd
• Geoff Wright, Arbitron UK Ltd

2
Design Goals

• Encapsulates the definitions and data of a large
  number of common surveys.
• Reasonably simple to implement for both export
  and import.
• Does not prejudice further development by
  incorporating more than is needed.

3
Text file benefits

• Files can be inspected using a suitable text
  editor which speeds the development of both
  import and export programs.
• Import functions could be written before
  equivalent export functions (if any) by
  hand-coding suitable Triple-S files.
• Introduce corrections if particular problems
  arose in the field.

4
XML

• Extensible Markup Language

5
HTML to XML

• ltFONT FACEArialgtWhen was your business
  founded?lt/FONTgt
• ltQUESTIONgtWhen was your business
  founded?lt/QUESTIONgt

6
Variable Types

• SINGLE - categorical data with one response
  allowed
• MULTIPLE - categorical data with any number of
  responses allowed
• QUANTITY - open numeric value (integer or real)
• CHARACTER - character data
• LOGICAL - individual Yes/No or True/False data
  values

7
Standard Names

• consist of 1-8 characters
• composed of only characters A-Z, a-z and 0-9
• start with a letter
• are case insensitive
• all characters are considered significant
• are unique within the survey

8
Definition File Overview

• lt!DOCTYPE SSS SYSTEM "c\sss.dtd"gt
• ltSSS VERSION"1.2"gt
• ltDATEgt2/MAR/1999lt/DATEgt
• ltTIMEgt1832lt/TIMEgt
• ltORIGINgtProg v1.42lt/ORIGINgt
• ltUSERgtAnn Otherlt/USERgt
• ltOPTIONSgtSTANDARD NAMESlt/OPTIONSgt
• ltSURVEYgt
• ltTITLEgtHistoric House Exit Surveylt/TITLEgt
• ltRECORD ID"V"gt
• details for each survey variable
• lt/RECORDgt
• lt/SURVEYgt
• lt/SSSgt

9
Single Type Variable

• ltVARIABLE ID"1"gt
• ltNAMEgtQ1lt/NAMEgt
• ltLABELgtNumber of visitslt/LABELgt
• ltTYPEgtSINGLElt/TYPEgt
• ltPOSITION START"1" /gt
• ltVALUESgt
• ltVALUE CODE"1"gtFirst visitlt/VALUEgt
• ltVALUE CODE"2"gtVisited before within
  the yearlt/VALUEgt
• ltVALUE CODE"3"gtVisited before
  thatlt/VALUEgt
• lt/VALUESgt
• lt/VARIABLEgt

10
Multiple Type Variable

• ltVARIABLE ID"4"gt
• ltNAMEgtQ4lt/NAMEgt
• ltLABELgtTwo favourite attractions visitedlt/LABELgt
• ltTYPEgtMULTIPLElt/TYPEgt
• ltPOSITION START"41" FINISH"42" /gt
• ltSPREAD SUBFIELDS"2" /gt
• ltVALUESgt
• ltVALUE CODE"1"gtSherwood Forestlt/VALUEgt
• ltVALUE CODE"2"gtNottingham Castlelt/VALUEgt
• ltVALUE CODE"3"gtquotFriarsquot
  Restaurantlt/VALUEgt
• ltVALUE CODE"4"gtquotMaid Marionquot
  Cafelt/VALUEgt
• ltVALUE CODE"5"gtMining museumlt/VALUEgt
• ltVALUE CODE"9"gtOtherlt/VALUEgt
• lt/VALUESgt
• lt/VARIABLEgt

11
Character Type variable

• ltVARIABLE ID"3"gt
• ltNAMEgtQ3lt/NAMEgt
• ltLABELgtOther attractions visitedlt/LABELgt
• ltTYPEgtCHARACTERlt/TYPEgt
• ltPOSITION START"11" FINISH"40" /gt
• ltSIZEgt30lt/SIZEgt
• lt/VARIABLEgt

12
Quantity Type Variable

• ltVARIABLE ID"5"gt
• ltNAMEgtQ5lt/NAMEgt
• ltLABELgtMiles travelledlt/LABELgt
• ltTYPEgtQUANTITYlt/TYPEgt
• ltPOSITION START"43" FINISH"45" /gt
• ltVALUESgt
• ltRANGE FROM"1" TO"499" /gt
• ltVALUE CODE"500"gt500 or morelt/VALUEgt
• ltVALUE CODE"999"gtltSPECIAL/gtNot
  statedlt/VALUEgt
• lt/VALUESgt
• lt/VARIABLEgt

13
Logical Type Variable

• ltVARIABLE ID"6"gt
• ltNAMEgtQ6lt/NAMEgt
• ltLABELgtEnjoyed visitlt/LABELgt
• ltTYPEgtLOGICALlt/TYPEgt
• ltPOSITION START"46" /gt
• lt/VARIABLEgt

14
Sample variable in Triple-S 1.1

• VARIABLE 4
• NAME "Q4"
• LABEL "Two favourite attractions visited
• TYPE MULTIPLE
• POSITION 41 TO 42
• SPREAD 2
• VALUES
• 1 "Sherwood Forest
• 2 "Nottingham Castle
• 3 ""Friar Tuck" Restaurant
• 4 ""Maid Marion" Cafe
• 5 "Mining museum
• 9 "Other"
• END VALUES
• END VARIABLE

15
Sample variable in Triple S-XML

• ltVARIABLE ID"4"gt
• ltNAMEgtQ4lt/NAMEgt
• ltLABELgtTwo favourite attractions
  visitedlt/LABELgt
• ltTYPEgtMULTIPLElt/TYPEgt
• ltPOSITIONgt41 TO 42lt/POSITIONgt
• ltSPREADgt2lt/SPREADgt
• ltVALUESgt
• ltVALUE CODE"1"gtSherwood Forestlt/VALUEgt
• ltVALUE CODE"2"gtNottingham Castlelt/VALUEgt
• ltVALUE CODE"3"gtquotFriar
  TuckquotRestaurantlt/VALUEgt
• ltVALUE CODE"4"gtquotMaid Marionquot
  Cafelt/VALUEgt
• ltVALUE CODE"5"gtMining museumlt/VALUEgt
• ltVALUE CODE"9"gtOtherlt/VALUEgt
• lt/VALUESgt
• lt/VARIABLEgt

16
Triple S-XML DTD

• lt?xml version"1.0" ?gt
• lt!ELEMENT SSS (DATE?, TIME?, ORIGIN?, USER?,
  OPTIONS?, SURVEY)gt
• lt!ATTLIST SSS VERSION CDATA REQUIRED gt
• lt!ELEMENT DATE (PCDATA) gt
• lt!ELEMENT TIME (PCDATA) gt
• lt!ELEMENT ORIGIN (PCDATA) gt
• lt!ELEMENT USER (PCDATA) gt
• lt!ELEMENT OPTIONS (PCDATA) gt
• lt!ELEMENT SURVEY (TITLE, RECORD) gt
• lt!ELEMENT TITLE (PCDATA) gt
• lt!ELEMENT RECORD (VARIABLE) gt
• lt!ATTLIST RECORD ID CDATA REQUIRED gt
• lt!ELEMENT VARIABLE (NAME, LABEL,
  TYPE, POSITION,((SPREAD?, VALUES) SIZE)?)gt
• lt!ATTLIST VARIABLE ID CDATA
  REQUIREDgt
• lt!ELEMENT NAME (PCDATA)gt
• lt!ELEMENT LABEL (PCDATA)gt
• lt!ELEMENT TYPE (PCDATA) gt
• lt!ELEMENT POSITION EMPTYgt

17
Benefits of XML

• Version changes easier to manage
• Standard syntax
• Off-the-shelf parsers simplify development

18
Further work

• Incorporate inverted data for each variable
• Describe structured (hierarchical) datasets
• Incorporate calculations and derived values

19
Availability

• Printed standard available now
• DTD and documentation available from ASC website
  http//www.asc.org.uk