next up previous
Next: 2000 CODEBOOK INFORMATION Up: No Title Previous: NOTES ON CONFIDENTIAL VARIABLES

2000 FILE STRUCTURE AND NOTE ON ``DATASET NUMBER'' AND ``VERSION NUMBER''


The data file for the AMERICAN NATIONAL ELECTION STUDY, 2000: PRE- AND POST-
ELECTION STUDY is constructed with a single logical record for each
respondent.  There are 1881 variables for 1807 respondents.


NES "Dataset number"
-------------------

In early 1999, each unique dataset in the NES archive was assigned a 
"Dataset number".  Dataset numbers for datasets from all archived 
NES studies are included in the NES "VERSION TABLE" described below.

"Versions" of NES datasets
--------------------------

The term "dataset" used by NES refers to the following associated
components:
     1-  ASCII data file (.dat file)
     2-  SAS and SPSS data definition files (.sas, .sps files) 
     3-  Codebook files (.cbk file(s)) ^^

Components of the initial release of a dataset will be identified as version
01.  According to this system, a corrected component of a specific dataset
is called a new "VERSION" of that component and is assigned a new "Version
Number." 

Because the initial release of a dataset is sometimes followed by corrections
to one or more components, a labeling method has been implemented to identify
the release version of the datset component(s). In practice, the version
labeling will allow the analyst to easily verify if he or she has the most 
up to date component(s) for that dataset.

The version number of a particular component file is written as the first
information in the machine-readable component file:

     1) In the ASCII data file (.dat file), the version number of 
        that data file is written in each record in columns 1-2.

     2) In the SAS and SPSS data definition files, the version number 
        of the file** is written in the very first line as a comment 
        similar to the following:
        * Version 01 SAS DATA DEFINITION FILE ;
                 or:
        * Version 01 SPSS DATA DEFINITION FILE

     3) In the codebook file**, the version number is written as the 
        first line similar to the following:
        VERSION 01 CODEBOOK


NES Dataset "Version Table"
--------------------------

The NES Web site (www.umich.edu/~nes) includes an NES Dataset "Version
Table" which can be used to identify the latest version of component files for
released NES datasets.

_______________

^^NOTE:  A codebook usually comprises 3 files, an 'intro' file, variable file,
and appendix file
**NOTE:  Since SAS and SPSS data definition files (.sas and .sps files)
are identified together as a single component, a new "version" of either
signifies a new "version" of both, even if only one data definition file
required correction. The "Note" field in the NES VERSION TABLE will indicate
if only one file has actually been corrected.
Similarly, since most codebooks are released as 3 files, a correction to any
one of the codebook files results in a new "version" of all 3 codebook files
at once. Again, the "Note" field in the NES VERSION TABLE will indicate if
only one codebook file has actually been corrected.  (All 3 codebook files
will include the version number in the first line of the machine-readable
file, as indicated above.)


Walter Mebane
Mon Nov 19 01:34:04 EST 2001