Q: Why is Epi2002 breaking down on freq * with a file containing boolean fields.

One question or aspect could be that field types <Y> in access (and thereby epi2002) are treated as having two values, either Y or N.

I have very little experience in using Boolean field types, but have had problems when exporting one file for a colleague from access due to the fact that all entered as Y were exported as Y and all others were exported as N regardless of data were entered or not.

My suggestion is therefore: a. make a small file in epi6 with the same three fields b. enten in epi6 data - say 5 records c. try the same freq * in epi2002 with that file

d. try with the "epidata" file in epi2002 to recode missing value to "N"

tell us if what works and what does not. If it works send me the new file and the one breaking down epi2002 and we will compare. If it does brake down my idea of not accepting a "missing" in a Boolean field is propably right and you could then in

recoding could also be done in epidata: recodeblock if m1ham = . then m1ham = "N" endif end

and run the recode in tools

regards Jens Lauritsen info@epidata.dk