Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "DeMichele, Matthew" <mdemichele@rti.org> |
To | <statalist@hsphsun2.harvard.edu> |
Subject | st: smoothing question |
Date | Mon, 24 Feb 2014 18:18:40 -0500 |
This may be outside the scope of the listserve, but I'm going to throw it out just in case someone has ideas. My situation is more methodological than software driven. I'm tasked with adjusting time-series data from 1980-2010 of about 480 respondents reporting administrative data for two time points in a single year. For instance, a respondent is asked in January of 2010 to report December 31, 2009 and January 1, 2009 populations. This is repeated annually. What we've noticed is that December 31, 2009 populations sometimes have large differences with January 1, 2010 (one day). It could be that an administrative or legal change took place, or even a recording error. My question is if anyone has advice for whether this seems to be best addressed with an ARIMA or exponential smoothing model. Or, if the Census X-12-ARIMA model/software if a better approach to address extreme values/trading-day adjustments. My apologies in advance if this is not something that fits the listserve. Matthew * * 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/