PNLMS PUDF README
This documentation is a guide to the installation and initial use of the public use data file (PUDF)
resulting from the 2013 Public Needs for Library and Museum Services survey (PNLMS)
conducted by the Institute of Museum and Library Services (IMLS).
The most important document for understanding the PUDF is the Data File Documentation. Critical
information necessary for interpreting the data are contained therein.
PUDF Data File Formats
The PNLMS PUDF is available in three formats, which allow the data to be read by common
statistical software packages. A user should download and employ the format that best
matches his or her interests, needs and the statistical software package to be employed.
The layout of all files is identical. Each is a flat file containing one row per household (3738) and
one column per variable (237). Variables are in the same order as they are listed in the
Data File Documentation. There are no fixed width fields, so there are no explicit position or length
characteristics associated with any of the files.
CSV Format
These are recommended for users employing Microsoft Excel and the R statistical system. Users
should be aware that if either version of the CSV file is used, for many analyses it will be
necessary to ignore values that are one of the following: LOGICAL_SKIP, DONT KNOW, REFUSED, and
MISSING. There are two CSV versions:
* PNLMS_PUDF_Label.csv contains data in the form of the value labels (character strings)
listed in the codebook.
* PNLMS_PUDF_Value.csv contains the same data, but in the form of the numerical values
listed in the codebook. For instance, “YES” becomes 1 and “NO” becomes 0. Responses
that are not present are coded as follows: -666 for LOGICAL_SKIP, -777 for DONT KNOW,
-888 for REFUSED, and -999 for MISSING. Users are cautioned that, without suitable
care, improper analyses (such as calculating the mean of a numerical variable) can be
performed on this file.
Both CSV files contain variable names in the first row, and should be read by software in a way
that recognizes this. Users should also be aware that some entries in the label version contain
embedded commas, and that re-saving it without recognizing these entries may be
problematic.
SPSS Format
The file PNLMS_PUDF.sav contains the PUDF in SPSS format. It was created directly from SPSS, and
includes variable names, variable labels, values, value labels and missing value codes. The latter are
assigned to -999 (MISSING), -888 (REFUSED), -777 (DONT KNOW) and -666 (LOGICAL_SKIP). For a few
variables, other values have been designated as missing. The SPSS file is self-contained, and needs no
additional files to be used in SPSS for Windows.
Weights and Variance Estimation
The PUDF contains two variables, PNLMS_NRFU_WEIGHT with respondent weights and
STRATUM with sampling strata that can be used to construct national estimates and calculate
associated standard errors. The method to be used is Taylor series linearization, which is
implemented in SPSS and R (in the “Survey” package). Users are cautioned that:
(1) the weights apply only to variables for which the unit of analysis is the adult respondent, and
not to variables pertaining to either households or children; and
(2) SPSS procedures do not appropriately compute standard errors when weighting is used, therefore,
users should adjust analysis plans accordingly.