Previous: Format:

Up: PRELIM AUTOMATE

Next: Example:

Purpose:

This saves a step in writing and using JudgeIt command files. Rather than requiring you to run the program twice, once to estimate LAMBDA and SIGMA and a second time after you've changed the JudgeIt command file to include those values, this command will allow the program to be run only once. JudgeIt will go through the cycle once, as set up, to estimate these values, which will be averaged (by the PRELIM WRITE; command) for each value of num that you indicated in the PRELIM Pnum; commands. Then the file will be run again to do the analysis procedures with PRELIM set to these new estimates.

Note that you should still make sure that the PRELIM values estimated by the program make sense, and that the pooling of SIGMA and LAMBDA is appropriate theoretically. For example, you might have specified pooling to occur for only two types of elections, but by looking at the estimated values of SIGMA and LAMBDA, you might notice a trend that could be taken into account when pooling. Thus this automation procedure in no way lessens the impact of substantive decision-making.

When the values of SIGMA and LAMBDA are averaged during the PRELIM WRITE command as used in the automation procedure, odd estimates are truncated to values within acceptable intervals before pooling. In particular, SIGMA values below 0.02 or above 0.20 will be set to the appropriate endpoint. Values of LAMBDA outside 0.02 and 0.95 will be set to these endpoints as well.

If you wish to change some variables or analysis commands and rerun a command file with PRELIM AUTOMATE, you can speed this next run considerably. In the first output file, PRELIM AUTOMATE will print some ALIAS commands to define LAMBDA and SIGMA that you can copy to your input file, replacing the PRELIM AUTOMATE command. This will avoid the first pass through the command file by not requiring you to reestimate SIGMA and LAMBDA.