Markup Notes
Notes on the mark-up process for
Eurobarometer 10, 1978 (ICPSR #7728):
The data in this study are available from ICPSR in two formats: a
card image file, containing three decks per respondent, and an Osiris
dataset constructed with a single logical record per respondent. The
mark-up describes the two formats as two separate files, the file
description section being repeated for this purpose. Then, in the
variables description, the location element is repeated, thus
accounting for the variables' location within each file, with an IDREF
referring back to the file's description.
For those variables where frequencies are given in a table format, we
have used the CALS tables model to mark up the category statistics.
In variables V44 to V48 (Q147 to Q149D) where political parties are
grouped by country, the use of category groups was precluded by the
fact that the same category value had been assigned to parties with
similar political orientations from each of the nine EC
countries. Instead, we used the country attribute at the category
level to designate country-specific category labels and statistics.
We used the "forward progression" and "backflow" elements to link
items based on the same pre-question text (see, for instance, V52 to
V63 for questions 154A to 165A).
Notes on the mark-up process for Current Population Survey,
1999: Annual Demographic File (ICPSR 2825):
In the mark-up for the March 1999 CPS we repeated the study
description section in order to differentiate between the versions of
the study distributed by ICPSR and the Bureau of the Census,
respectively. A separate study description for the ICPSR-distributed
version allowed us to incorporate in the marked-up document the
metadata prepared by ICPSR (abstract, universe and sampling
information, etc.) as well as the Consortium's data use statement.
The design of the CPS invited extensive use of the variable groups and
category groups nesting feature of the DTD; see the section describing
variable groups and, for nested category groups, variables HG-ST60
(1960 Census state code), HHDFMX (household and family status), POCCU2
(occupation of longest job by detailed groups), and WEIND (industry of
longest job by detailed groups).
The flags used in the CPS codebook to identify the various elements of
a variable description (name, location and width, universe, values
range, category labels, and values) allowed us to use macros for the
mark-up of this section, which saved us considerable time and effort.
Filler variables were omitted in the variable description
section. Their inclusion would entail renaming them, and we felt this
would have been an unnecessary step for our purposes.
