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

RE: st: cross-model testing on panel data

From   Maarten buis <>
Subject   RE: st: cross-model testing on panel data
Date   Wed, 7 Nov 2007 09:28:37 +0000 (GMT)

--- Elissa Braunstein wrote:
> I am using US state-level time series data on factors determining  
> male versus female employment

Are you estimating a model explaining some state level aggregate of
male and female employment like the proportion of women working? If so,
you should be very careful about the ecological fallacy. 

Consider the following example by Robinson (1950) he compares litteracy
status and immigrant status. Using individual data he finds that
immigrants are more often illiterate than non-imigrants (not
surprising). However if he looks at proportion of immigrants in a state
versus proportion illiterate he finds that states with a higher
proportion immigrants have lower proportion illiterates. Why?
Immigrants go to wealthy states and wealthy states have less

Gelman et al. (2007) have a more recent example looking at income and
voting republican: Rich people vote republican, but rich states (East
and West coast) vote Democrat.

Bottom line: what is happening at the state level (or any other
aggregate level) is a very poor indication of what is happening at the
individual level. The effects can easily be complete opposites. 

-- Maarten

Andrew Gelman, Boris Shor, Joseph Bafumi, David Park (2007) ``Rich
state, poor state, red state, blue state: What’s the matter with

Robinson, W.S. 1950. ``Ecological Correlations and the Behavior of
Individuals.'' Amercian Sociological Review 15(3):351–357.

Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

Want ideas for reducing your carbon footprint? Visit Yahoo! For Good
*   For searches and help try:

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