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

st: RE:Methodological issue: combining datasets from different ...

From   "Georgiana Bostean" <>
Subject   st: RE:Methodological issue: combining datasets from different ...
Date   Mon, 6 Jul 2009 13:04:56 -0700

Hello again and I am sorry to bombard your e-mail boxes!

Thank you, Edwin and Martin for your responses, and particularly Martin
for your suggestion-- I will take a look at that command. I wanted to
clarify and contextualize my question, as it appears that it was
interpreted as me asking the list to do my work for me. That is not my
intention at all.

I am working on my dissertation, and have consulted textbooks, a
statistician, and social scientists who are well-versed in methods about
my issue and have not yet found an answer. I could well just do a logistic
regression and not think twice about it, but as I said, my initial
reaction is that I somehow need to account for the fact that the surveys
are of different populations (though there are, for example, immigrants
from the origin population in one dataset and non-migrants in the other

So, I thought that someone on Statalist could point me to any literature
or provide comments on whether I should be doing some analysis other than
a logistic regression in this case. My understanding of Stata and of
statistics is not as sophisticated as I am sure some of yours are. But I
certainly am not asking that someone "do my homework."

I did not go into too much detail about the issue in my previous e-mail
because I didn't want to get into all of the specifics of my dissertation.

However, to briefly elaborate, my "outcomes" are categorical, dichotomous
and ordinal. The theoretical question is whether the migrant population at
destination differs from the non-migrant population at origin in several
specific outcomes, both ordinal and dichotomous. I want to compare the
"likelihood" (used loosely) of Y occurring by X (migration status), which
is why initially a logistic regression seems suitable. I plan to control
for a variety of factors, including both continuous (e.g., age) and
discrete (e.g., marital status) measures. And, ultimately, I would like it
to be generalizable to the populations in question (the question of
sampling design).

So, again, thank you Martin for your guidance. And if anyone has
suggestions about textbooks that may address such an issue, please do
share (I have been searching Hosmer-Lemeshow's Logistic book and there is
nothing that addresses this).

Thank you.

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index