Caution re: merging in NELS:88 (through 1994) release

"For your information, I am sending this email documenting a problem I experienced with the NELS:88 data using the CD-Rom for the 1988 - 1994 waves. I do not know if this problem exists in the recently released CD-Rom for 2000 data.

Using the extraction software provided with the CD-Rom for the data, I checked off variables I wanted. Most of the data come from a file "F:\nels94\stmeg3.pub" but some of the data I was interested in (data about post-secondary experiences) come from a file
"F:\nels94\pse1994.dat." The data in these two files is structured differently. In the first file, the "stmeg3.pub," there's only one record per person. In the second file, the "pse1994.dat," there can be multiple records per person, depending on the person's post-secondary attendance history. In the second file, unique records are identified by the following variable combination: STU_ID + INCODE + INSTNUM.

It is not clear in the extract process (in making the .tag file) which variables come from which file, nor is it clear that the "pse1994.dat" file is constructed differently than the commonly used file, "stmeg3.pub." Given that the student identification numbers (STU_ID) can repeat in the "pse1994.dat" data, a user should NOT try to merge these data sets by simply using the STU_ID number.

Because I was unaware of this difference, I merged these two files and arrived at multiple records for people with post-secondary attendance. This led to errors in my analysis.

Final note, there is another file on the CD-Rom, "F:\nels94\inst1994.pub." I am not sure how this file is organized."

As reported by Molly Martin, June 6, 2003.