Thanks Torsten,
I have tried:
simple file: 1 numeric field, 4 records, one record has missing value complex file: 23 fields, 700+ records, lots of missing in numeric, text and dates
Stata export/import of complex file - dates and missing values are correct Excel export of complex file - everything looks OK Open Office export - simple file OK; complex file is not readable Open Office import - simple file reads OK; complex file (exported from Excel to .ods) reads missing numeric as zero, everything else looks OK CSV import of complex file - two adjacent missing fields are interpreted as a single missing field, so that fields do not line up with .qes and import fails - first record is "A123"<tab>1123<tab><tab>"F" for .qes with ____ #### ## <A>- "F" is read as third field, which is supposed to be numeric, instead of missing third field and "F" in fourth field.
Import should allow errors, up to some limit set by the user. In some cases, this is correct behaviour because the source data has errors that Epidata will identify, but still want most data to go in correctly.
Jamie
Torsten wrote:
I have posted a new version of the Core Module for testing on the EpiData website: http://epidata.dk/testing.php
With kind regards, Torsten Bonde Christiansen. Software Developer, Epidata.