Re: st: Problems with FE and Hausman test

[Christopher Baum <kit.baum@bc.edu>]

<>
On Feb 19, 2014, at 2:33 AM, John wrote:

> My situation is as follows:
> Dependent Variable: % students graduating high school within 4 years
> Independent Varibles (8): Measures of administrative authority held by state over education matters (vs. local vs. shared). Each measure ranges from 1-30.
> Unit of Anaylsis: States (23)
> Time: 3 years
> Observations: 69
> 
> I've been trying to determine if I should account for fixed effects or random or pooled OLS but cannot get a Hausman score.

If these eight measures are time-invariant, they are state-level effects, which explains why they would all be dropped in a fixed effects model. Your model seems to ignore the macro time variation, and should certainly include two time effects. 

You could fit a random effects model, but that would suggest that average state-level performance is a random draw from a distribution of quality-of-educatiion factors. That ignores, for instance, that average teachers' salaries in Mississippi might be much lower than those in Massachusetts. We know that paying more doesn't necessarily translate into higher test scores, but I have a hard time believing that Mississippi and Massachusetts students' relative performance on standardized tests is due to random draws, and not something that can be explained by including a number of other relevant, measurable factors from the education literature. How the state oversees local education (and, importantly, funds it) is no doubt important, but it's not the only thing that's important.

Kit

Kit Baum
Professor of Economics and Social Work, Boston College, Chestnut Hill MA, USA
DIW Research Professor, Department of Macroeconomics, DIW Berlin, Berlin, Germany
baum@bc.edu  |  http://ideas.repec.org/e/pba1.html












*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/