Dear statalist,
This is a question regarding the appropriateness of the techniques I have chosen for my phd. Unfortunately, I do not have a statistician at my uni who can provide me with an expert opinion on this- hence why I am posting this message on stata list. I am sorry for the list of questions that follow, but I would be truly grateful if any one can offer me advise.
I have panel data containing suicide rates for 74 countries over a long time series. I wish to test the relationship that these have with social variables within the country (ie. Unemployment rate etc.,).
Suicide rates are transformed using the square root transformation and social data is transformed using the ln transformation.
Currently I have been using fixed/random effects to estimate the relationship between these variables. I have also made a lagged model of these relationships using xtabond2.
However, I realise that researchers often use the poisson distribution in conducting analysis with suicide data. Considering this, should I stick to techniques that use this distribution (xtgee is one I know… I am not sure about any others that may also use this)? or can I still use the fixed/random effect models? Also, if I do need to use the poisson distribution, how can I conduct my dynamic model? (I do realise that Martin Weiss made some suggestions to me for this- thank you)
Thanks for your advice in advance.
*
* 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/
