Tag Archives: longitudinal study
Just a quick note: some of the data which wanted and could not get the ICPSR, such as the Educational Longitudinal Study of 2002, is available at http://nces.ed.gov/edat/ — I will continue to look for data sources other than the better known but clearly … Continue reading
Using a text editor it was easy to convert the SAS catalog files which come with the WLS to simpler formats to be read by simple programs.
Creating a bit array with the Python Bitarray package will produce a one dimensional array of bits, any number of which can be put in a list to produce a two dimensional thing. I need them transposed. You can transpose an ordinary two dimensional list of lists in one line of code using map and zip. Continue reading
Coordinates for individual persons can be constructed by finding extreme points and using the similarity to these extremes as coordinates.
Given these coordinates, the same number of coordinates for the microvariable representing one value of a survey variable can be creating by taking the mean of the rows which have that bit set. Continue reading
I am still looking for better sources of the information most important to me, but I have had some luck with the Wisconsin Longitudinal Study. Just a bit, but some. On question, repeated on different instruments asks how close the respondent … Continue reading
Treating coded variables as probability columns will form a very large matrix, which can be processed to correct the data. This may be done interatively, because of limits to processing power and memory. Continue reading
It seems that data from the Wisconsin Longitudinal Study can be used without easily machine readable codebooks by using the catalog files from the SAS distribution with the CSV distribution. The fact that some raw data is hidden in constructed variables in unfortunate, but we can probably work with what we have, at least for prototyping. “Just the facts, ma’am, just the facts”, please, questions exactly as asked, answers exactly as received would be the most useful. Continue reading
To say that WLS codebooks are not digitized was wrong, as they are digital in the sense of being available as PDF and HTML files. They help me little, being not much use except to print out. Digital codebook data is also in files for a big statistical package like Stata, but these are of no use to me since they would require the purchase of a big expensive package for the sole purpose translating the codebooks into a simple CSV format. Though I remain frustrated, I was wrong. Continue reading