John Magill wrote:
> I am trying to figure out if Panel Data Analsysis is the most appropriate procedure to use to analyze some data
> We are conducting an impact assessment of a development project is South Asia. There is a control group and a test group. We have administered a questionnaire at the beginning of the study, and are now going back in to administer a follow-up survey two years later. There are only two points in time -- that of the original survey and that of the followup survey. We want to know if the changes over time are significantly different for the test and control groups.
> Some of the variables are continuous varibles (income in time 1 versus income time 2), where we want to know which of the two groups (control or test) has experienced the greastest increase in income. Others are various forms of scales -- such as "How would you rate your income?": 1=Very good, 2=Good,3=Neither Good nor Bad, 4=Bad, 5 = Very Bad.-- where we want to see if perceptions have changed over time and which group has changed the most between time 1 and time 2.
> Is Panel Data (as in the "Longitudinal/Panel Data" manual) the appropriate approach to take in analyzing thse two types of questions with control and test groups, or is there a more appropriate procedure for doing this?
IMHO, you're better off just using OLS including a dummy variable for
the surveys. If you find autos, heteros or both at post-estimation,
switch to -prais-, -reg, cluster()- or (from SSC) -ivreg2-
respectively and re-run the models.
--
Clive Nicholas
[Please DO NOT mail me personally here, but at
<[email protected]>. Thanks!]
"Courage is going from failure to failure without losing enthusiasm."
-- Winston Churchill
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
