Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: 9999, matching data, fixed effect probit etc...

From   "Le Wang" <>
Subject   Re: st: 9999, matching data, fixed effect probit etc...
Date   Sat, 25 Nov 2006 01:26:35 -0600

My thoughts on the questions.

(1a) IPUMS-USA creates some routines that construct the family
interrelationship variables. You can check it out.

(1b) I think your expectation that husband and wife will give the same
answer is wrong. They are not necessarily the first partner for each

(3) I guess that you have three variables: var2, sex, year of birth.
If so, I think you should calculate mean of each category by female
and year of birth, which will be more meaningful.

(4) Follow Nick's suggestion.

Hope it helps.

On 11/24/06, Nick Cox <> wrote:
Nirina F. had problems sending this to the list.
I edited slightly using the -cmdname- convention.


-----Original Message-----

I have a few questions:

1- How do I create husband and wife variable in my data where
I have data on marital status, line number of respondent in
the household, the sex, etc..

Then  when I have identified the husband and wife, how do I test
whether they gave the same answer on the question of number of
kids they have. This is basically to test whether there was mistake in the data.

2- var2 is categorical having 4 answers,sex, year of birth.
Plot var2 on y axis, and year of birth on x axis but only for
female. I tried -tw histogram- but it doesn't give the graph
I would like to get.

3- I have the following:
year of birth: 4 digits
year of current marriage with 4 digit years but when respondent
don't know or for other reasons, they can respond to 9999
 so now, if I would like to calculate age at marriage, I get
for example 8069 (9999-1930)  for people having responded 9999
for the question year of current marriage.

Could you suggest me please how I can transform this into
meaningful values or anything else?

4- does anyone know how to do -probit- with fixed effect? it seems
that the estimates will be inconsistent so has anyone found a
way of getting away from the incidental parameter problem?

Thank you very much in advance.

*   For searches and help try:

Le Wang, Ph.D.
Minnesota Population Center
University of Minnesota
(o) 612-624-5818
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index