Correct for Dependent Outcomes. Try to pick match fields with independent comparison outcomes (agreements and disagreements for one match field are not correlated with agreements and disagreements for a second match field on either matched pairs or unmatched pairs). Sometimes this is not practical because you would lose too much information, in which case you must set Agree X and Disagree X to correct for dependency. Often, two fields derived from the same reported field will have dependent disagreements – that is, all derived fields will tend to disagree together on matched pairs. For example, when non-crash E codes are recorded accidently for crash patients then all derived fields like StdCollide, StdVehicle, and StdPosition will disagree together on true Crash Hospital matched pairs for those patients – double or triple counting disagree weights. We can only correct a pair of fields with dependent outcomes so you could drop StdVehicle as a match field because it does not contribute much. Match specs like these examples should correct for StdCollide and StdPosition dependent outcomes:

 

Field A	Field B	Method	Tolerance	Pr Different	Agree X	Disagree X
StdCollide	StdCollide	Exact	0,0	0.0	0.99	0.70
StdPosition	StdPosition	Exact	0,0	0.0	0.99	0.70

Agree X = 0.99 means that you don’t expect strongly correlated agreements for the two fields. Set Agree Xs to the same values less than 1.0 for a dependent pair of fields. Also set Disagree Xs to the same values less than 1.0. Don’t use the same Agree X for more than one dependent pair. In this example, no other pair can have 0.99 as Agree X.