[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Pulling out observations based on a condition

From   Dmitriy Krichevskiy <>
Subject   Re: st: Pulling out observations based on a condition
Date   Wed, 29 Jul 2009 12:24:41 -0400

Sorry I should have said polychotomous instead of dichotomous.
I have a very large panel dataset 14000 people over 7 years quarterly
(there are about 180 variables). I wanted to examine the income
population dynamics. Thus I created  quintile variables for income in
each representative quarter. I then reshaped to long, got rid of
generated missing values, and created a transition probability
matrices via -xttrans- . But it appears there is something wrong with
the results even though tests say that the transition processes are
stationary. There is too much mobility and I don't know why.
I wanted to "pull out" (i.e keep or tag) people who transition from
lowest to highest quintiles (since this is really drastic) and take a
detailed look at them to see why it is happening. But I have no idea
how to achieve this.

2009/7/29, Howard Lempel <>:
> Dimitriy,
> You'll get a better answer for this question if you describe your data a bit
> (maybe by listing a representative portion of it) and are a little more
> specific about exactly what you want your output to look like.
> You say you have a "dichotomous" variable, which I think is what confused
> Jeph.  It seems that your "status" variable actually has 5 values.
> A couple questions that might show the type of detail that would be helpful:
> Does the transition have to take place in ONE quarter?  If someone starts in
> state 1 and increases by 1 state in each quarter until they get to state 5,
> should they be kept?
> If someone starts in state 1, transitions to state 5, but then goes back to
> state 1, do you want this person to be kept?
> Howie
> -----Original Message-----
> From:
> [] On Behalf Of Dmitriy
> Krichevskiy
> Sent: Wednesday, July 29, 2009 11:56 AM
> To:
> Subject: Re: st: Pulling out observations based on a condition
> Thanks a lot Jeph,
> First, to clarify - I would like to keep those that do transition. But
> I want to handpick those who transition from state 1 to state 5 - not
> just those that leave state 1 to go anywhere. Thanks again, Dmitriy
> *
> *   For searches and help try:
> *
> *
> *

Dmitriy Krichevskiy a. b. c. d. e. f. g. h. i. j. k. l. m. n. o. p. q.
r. s. t. u. v. w. x. y. z.
*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index