[EpiData-list] New data format for testing and comments - tests needed

epidata-list at lists.umanitoba.ca epidata-list at lists.umanitoba.ca
Fri Oct 2 13:05:09 CDT 2009

An exiting phase of development is now ready for general comments and
testing. In june/july of this year the first cross platform experiments
were available published on the testing page

Crossplatform function must solve aspects such as:
- many different local character sets
- different standards for physical file formats
- the user should not do anything special when sending or receiving
files created on another operating system

On top of this another issue is coming up soon - the development of a
dedicated data entry module also allowing for GCP (Good Clinical
Practice) rules of transaction logging.

Therefore we have now experimented with development and specification of
a file format which contains as well the data as the metadata needed for
understanding the data. By doing so we are moving towards a new structure:
a. Datafile containing datadefinition and data.
    (parts of current rec and chk file). Suggested extension: recxml
b. Dataentry definition file containing control during entry
    (e.g. relate, range, mustenter etc). Suggested extension: chkxml
     - part b not developed yet

The proposed new data format is described here:

Just now users are encouraged to test the following:
1. Can the test application read your files ?
2. Can the test application write your files ?
3. In particular test - when possible - whether the data files can be
read on another operating system. E.g. saved by a Linux or Mac and then
read on a windows pc.

Get the test application from: http://www.epidata.dk/testing.php
And report to the epidata-list of failures and success.

Data formats supported are: REC, RECXML, ODF, DBF, XLS
A table of what can be imported and exported is shown here:

Operating systems supported are: Mac (OS X), Linux (32+64bit), Windows.
Mac PowerPC expected soon.

After a few weeks of comments on the format a new test application will
be created. Which will collect some time estimates of reading and
writing data with subsequent reporting to a central database. This
method should allow for future user tests of core modules with
documentation available on the central website.

Regards Jens Lauritsen
EpiData Association

More information about the EpiData-list mailing list