Triple-S is an open standard for data files, primarily surveys. This command reads a fixed-format or csv file conforming to this standard. It can also generate and save a syntax file for reading the data.
- IBM SPSS Statistics 18 or later and the corresponding IBM SPSS Statistics-Integration Plug-in for Python.
- Open IBM SPSS Statistics
- Navigate to Utilities -> Extension Bundles -> Download and Install Extension Bundles
- Search for the name of the extension and click Ok. Your extension will be available.
Read a file in Triple-S format version 2.0.
STATS GET TRIPLES
METADATA ="file"^*
DATA = "file"
SYNTAX = "syntax file"
DATAENCODING = LOCALE^** or UTF8
EXECUTE = YES or NO^**
STRMVCODE = "string missing value code"
MAXDATALENGTH = number
/OPTIONS
MDVALLABELS = "label for 0" "label for 1"
FULLLABELATTR= YES^** or NO
REMOVEHTML=YES or NO^**
SCORERECODE=YES or NO^**
/HELP
^* Required
^** Default
/HELP displays this help and does nothing else.
Example:
STATS GET TRIPLES METADATA="c:/data/mysurvey.xml"
SYNTAX = "c:/data/mysurvey.sps".
The output may be an open dataset, a syntax file that reads the file, or both.
Triple-S format requires two files. METADATA specifies the xml metadata data that defines the formatting and any multiple response information. HIERARCHY files are not supported.
DATA specifies the data file described by the format. By default, the data file is assumed to have the same name and location as the metadata file and extension txt.
DATAENCODING specifies that the character data are encoded as UTF8 or LOCALE. The Triple-S standard expects code page 1252 or equivalent, but other encodings might be encountered in the wild. LOCALE specifies the current Statistics LOCALE setting. The metadata file is expected to be in utf8.
If the record length is greater than 50000, use MAXRECORDLENGTH = number. Otherwise, bytes beyond that value are lost. It does not matter if the stated length is greater than the actual.
SYNTAX optionally specifies the name of a syntax file that will contain the syntax to read the data file.
EXECUTE specifies whether the syntax (saved or not) should be executed. Note that by default the generated syntax is not executed.
At least one of SYNTAX and EXECUTE=YES must be specified.
STRMVCODE optionally specifies the code to be used for string variable missing values. The default action is to examine the highest value code for a string variable and increment the last character by one character.
Multiple dichotomy sets are defined with variable values of 0 and 1. The value labels default to "No" and "Yes". A pair of alternative labels can be specified with MDVALLABELS.
If variable label text exceeds the SPSS variable label length limit, which is 255 bytes in current versions, the text is truncated. Specify FULLLABELATTR=YES to create a custom attribute with the full text.
The Triple-S standard does not support HTML markup, but in
practice it may be present. Specify REMOVEHTML=YES to remove
HTML markup from text. That causes text of the form
<...> to be removed, but this could match text that is not
actually an html tag to be deleted.
SCORERECODE=YES causes the input values for variables that include a score attribute in the Value elements to be recoded to the score values, and the value labels will refer to those values. Multiple response variables are not recoded. The scores must be consistent with the variable type: a numeric variable cannot have string score values.
This command assumes that the xml file conforms to the Triple-S 2.0 specification. If it does not, the command may fail rudely. Information about the standard can be found at http://http://www.triple-s.org/.
NOTE: If the Triple-S website is down or the computer does not have internet access, and your sss file
includes the DTD reference, this command will fail. However,
removing the DTD reference will let it work. The DTD reference
looks like this.
<!DOCTYPE sss PUBLIC "-//triple-s//DTD Survey Interchange v1.1//EN" "http://www.triple-s.org/dtd/sss_v11.dtd">
If multiple response sets are present (multiple dichotomy or multiple category) the component variable names are generated from the corresponding input of type "multiple" by appending an underscore and numerical suffix. In the unlikely case where this would cause a name collision, paste and edit the syntax as needed.
For single variables and multiple category sets, if the values element
includes any scores, a custom attribute array named score is created.
Each array element has the form value|score, and if value includes
the vertical bar, it is escaped as \|.
Variables of type Date or Time are read as strings. You can use the Date/Time Wizard to convert these to SPSS dates if needed.
No attempt is made to shorten names or labels that exceed the maximum lengths supported in Statistics.
Triple-S names are case sensitive while Statistics names are not, so it is possible to have a name collision.
The Triple-S standard requires the comma as the field separator in csv format. If you are reading a csv file and are in a locale where comma is the decimal character, you may have to change your Statistics locale in order to read the file.
Language tagged strings are not currently supported.
- Apache 2.0
- JKP, IBM SPSS