Dear Statalist,
I have spent many hours trying to work this out, but all the
information I could find was on how to evaluate survey data where a
variable containing weights already exists. Everyone I know who works
with Stata is also only familiar with data analysis once the dataset
is correctly adjusted, or using experimental data. So even if the
answer is simple, I would be very grateful for your help!
I have a dataset from a survey that over-sampled highly educated
people. I'd like to weight it so that the data is closer to the values
in the true population, which I have from census data. The
distribution of levels of education in my sample looks like this:
highest |
level of |
education | Freq. Percent Cum.
------------+-----------------------------------
1 | 35 6.67 6.67
2 | 92 17.52 24.19
3 | 230 43.81 68.00
4 | 61 11.62 79.62
5 | 107 20.38 100.00
------------+-----------------------------------
Total | 525 100.00
Level 5 (tertiary education) is only 9% in the population, according
to census data, but 20% in my sample. Level 1 is 24% in the
population, but 6.67% in my sample. Etc.
Should I just create a new variable along the lines of
gen weight=0
recode weight (0=3.6) if educ==1
etc? Would this give me the weight variable that I'm looking for -
that I could then use in a svy: regression?
Thank you so much for your input.
Fran Brittan
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/