12th June 2009 at 5:02 pm #5915Paul J GrahamMember
I’m looking for resources (books, articles, websites, etc.) that might have something to say about stats for policy analysis. If you are in a Policy Analysis role, what are the key stats you are looking for in any particular social problem scenario? I can find lots of books on policy but I’m looking for something specific for applying stats in policy scenarios. Any assistance is greatly appreciated.15th June 2009 at 4:13 pm #5919Berkay OzcanParticipant
Can you clarify a bit more what do you mean by “social problem scenario” and “policy scenarios”? As far as I understood from your question, you are NOT interested in Policy Evaluation methods ( i.e. whether a specific intervention has an impact on social processes or not), am I right? What kind of policies do you have in your mind?
Berkay15th June 2009 at 5:43 pm #5918Paul J GrahamMember
Thanks for replying. Yes, my question is quite generally broad. To give you an example, in discussing methods issues with local health policy resarchers, they said that policy people are more likely to gravitate to simple cross tabs and assessments of strength of relationships rather than, say, sophisticated statistical measures or even methods such as Factor Analysis or Regression. Of course, this is dependent on the scenario. I have “health” specifically in mind as the policy area…but essentially I just wanted to throw out the question for feedback. Just what do health policy analysts rely on most for stats analysis? Are there specific methods that health policy analysts have when addressing a new health problem? Is there any standard or is it just too context dependent for any kind of reply to the query?
In my case, I’m trying to account for these ideas in relation to health knowledge transfer. Know the methods for better KT.18th June 2009 at 1:08 pm #5917Robert MorrisMember
You might take a browse through some policy specific journals (e.g., Criminology & Public Policy). As far as specific stats, the analytical procedure will depend on the style of data you are using. Is your data longitudinal? Is the unit of analysis at the individual or macro level? Is the data nested? Give me some more information and I can probably point you in the right direction.
If your data are longitudinal and you’re asking whether there is an effect… Most will be based on a regression model with the policy as an independent variable…
– Random/Fixed regression or GEE (i.e., panel regression)
If longitudinal and nested-
– multilevel model for change (growth modeling)
If cross sectional-
– a range of many GLM regression models (e.g., OLS, Logistic Reg., Neg. Binomial, etc.) depending on the scale of the outcome measure
If you’re interested in exploring the time to an event, a variety of survival analysis methods could be used (alone or in tandem with the above).
In all, it depends on your specific question.
hope this helps.
RM18th June 2009 at 1:09 pm #5916
- The forum ‘Default Forum’ is closed to new topics and replies.