I am just beginning a project in which I will be using student-level data
from the National Assessment of Educational Progress (NAEP). I am not very
familiar with the data yet, but I understand that it not only uses a
complex survey design (with stratification and clustering), but also
something that is referred to as "matrix sampling" whereby no one student
is given the entire exam. For this reason, the National Center for
Education Statistics (NCES), which puts out the NAEP, provides 5 different
"plausible values" which represent different imputed scores for the
individual (that is, from what I understand, a guess as to what the student
"would" have gotten if he/she had taken the entire exam - based on how the
student did on the questions answered as well as the students observable
characteristics).
Apparently, there are some statistical packages, such as HLM, that have
components that will easily and conveniently utilize these "plausible
values" (multiple imputations) in analysis to not only obtain various
sample statistics (means, reg coeff, etc.), but also standard errors.
If anyone is familiar with a way to do this type of analysis in Stata, or
has worked with student-level NAEP data in Stata in other ways, I would be
very grateful for your thoughts/advice/suggestions.
Thank you.
-------
Brian A. Jacob
Assistant Professor of Public Policy
John F. Kennedy School of Government
Harvard University
79 JFK Street
Cambridge, MA 02318
ph - (617) 384-7968, fax - (617) 496-5747
web - www.ksg.harvard.edu/faculty/brian_jacob
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
