Helping users troubleshoot data problems is an inevitable part of our job. These situations call for close cooperation between the user and librarian. Keep in mind that its often true that the user may be far more familiar with the data and documentation that you are, and that this is not necessarily a bad thing. Frequently the problems are due to the user misunderstanding the variable coding or some other aspect of the documentation. It is often helpful to sit down with the documentation and have the user go over in detail what is happening. If the solution is not an obvious one like "oh! you've got the wrong columns!", then a detailed description of the errors is useful. The following is taken from a recent and harrowing example:
Dear Laura,
In my work with the ECA data I have come across the following problems, which I hope you will be able to help me with:
1) The frequencies listed in the codebook are incorrect.
According to page 10 of the codebook (by page 10 I refer to the number in the lower right hand corner of the pages), the total sample size for Wave 1 is 20,861. In the frequencies listed in the codebook (pages 1000 and on), however, the frequencies for each variable adds up to 18,572.
When I run frequencies on the data file, I find 20,861 cases. Would it be possible for me to obtain a printout of the frequencies for the full file from ICPSR?
2) The data contains codes that are not listed in the codebook. For example, the variable ECAAREA contains 55 cases who are coded as '29,' a value that does not appear in the codebook on page 163.
Would it be possible for me to obtain a codebook that lists all the values contained in the data?
3) Some listings in the codebook seem to be wrong. For example, the following entry appears on page 163:
WAVE Distinguishes different Waves of ECA data
For Wave 1 data: Wave = 1
The frequency of this variable is as follows:
The SAS System 1
14:05 Tuesday, April 15, 1997
Cumulative Cumulative
WAVE Frequency Percent Frequency Percent
--------------------------------------------------
1 20254 97.9 20254 97.9
2 430 2.1 20684 100.0
Frequency Missing = 177
The number 20,254 listed here does not match the frequency
listed on page
10 of the codebook, nor the frequencies listed on page 1000 and
on.
So, my final question: would it be possible for me to obtain a
codebook
that correctly describes the values of the variables?
Thanks for all your help! Please let me know if anything I
wrote above is
unclear.