EXPLANATION OF BOOK PROGRAMS

• Most programs are in Stata version 8.0, executed on a MSWindows PC with Stata 8.2.
  Stata 7 will usually be okay.  Exceptions where Stata 8 is needed include:
  (1) Estimates command (for tabulating regression results) is not available in version 7.
  Comment out occurrences of "estimates store ..."  and "estimates table ...."
  (2) Graphics commands (used to obtain the figures in  the book) changed substantially from 7 to 8.
  This only effects generating figures. If graphs are important, it is best to upgrade to Stata 8 as so much better.
  (3) In some places free Stata add-ons have been included. These are noted in programs.
  To download these programs e.g. knnreg in Stata give command "search knnreg" and follow directions.
  The Stata programs vary from very problem-specific code to code that potentially can be adapted to one's own needs.
• Some programs use Limdep version 7.0 and Nlogit 2.0, executed on an MSWindows PC.
• Some programs use SAS / IML. SAS version 8.0 used on a Unix machine.
  

• For Stata: as an example for chapter 4.5.3 we provide:
  - mma04p1wls.do      Stata program
  - mma04p1wls.txt      Output from this program
  - mma04p1wls.asc     The generated data as fixed width ascii data set
                                    [permits analysis with programs other than Stata]
• For Limdep: as an example for chapter 14.5.3 we provide:
  - mma15p3mnl.lim      Limdep program
  - mma15p3mnl.out      Output from this program
• For SAS: as an example for chapter 13.6 we provide:
  - mma15p2bayesgibbs.sas    SAS program
  - mma13p2bayesgibbs.lst     SAS output
  - mma13p2bayesgibbs.log    SAS logfile
  
• For data sets the extensions are: 
  - .dta    for Stata data set
  - .asc    for ascii (text) data set that is usually both space delimited and fixed width
• For descriptions of the data sets see the relevant program that uses the data set, and the associated output.
  

• Programs generally take little time to run.
• Exception is programs that entail simulation, including bootstrapping.
• Programs can be speeded up by reducing the number of simulations / replications, though final analysis should use many simulations / replications.
  

Back to Main Page