Exciting News
Today we have released a test version of the rewritten analysis, which for the first time approaches functionality resembling the final first version to be released later.
You may download the test version from http://www.epidata.dk/testing.php
Please notice that it is still premature to use the new analysis for production every day use, but we urge users to challenge the new analysis. You could for instance replicate data-management procedures you do with ordinary Classic Analysis v2.2. rel 3 (Build 187)
The intention is to release the new analysis for production when the following has been developed: A: all functionality for data management (new projects, new variables, labels and editing of these). B: append and merge. C: basic reporting - e.g. for logfile overview
Already now you may test many of these aspects. After download and unzipping (files are provided in zipped archives) you will see the commands and syntax documentation.
We attempt to maintain the same syntax that you know, but to allow for the modernization some changes are needed: e.g. read and write epx, csv and dta files ; work on all operating systems (Linux, Mac, Windows) ; accept UTF-8; extend with special functions; sort data; manage encrypted data. You can read rec files, but not write them. Use epx for saving data.
We also aim for standardization across commands. E.g. number of decimals in output is always !Dx, where x is the number of decimals 0,1,2 etc.
For data analysis these commands are implemented: count , means, freq,
Some changes: use := instead of = . Put texts in " " . End all command lines with ;. Selection works on a single block. Options are mentioned with ! instead of /.
Some enhancements: A copy of all executed commands is saved (commandlog.pgm). Recent pgm in editor filelist. Multiple pgm sorting ascending and descending and more. See datasets (F2) in a project. Open encrypted files.
A frequency table of v17:
freq v17 !vl !cum ; // would show labels and values plus cumulative percentage.
To do the same table among those younger than 40 with no decimals.
select (age < 40) do freq v17 !vl !d0;
To remove all observations with age above 68. Use the command drop:
select (age> 68) do drop data;
A new construct is a loop, e.g. :
new global integer i ; for i:=1 to 5 do ? "This is a number counting: " + i ;
Which can be used to show means or frequencies for subgroups of another variable:
new global integer i ; for i:= 1 to 5 do select (area = i) do means age !by:=sex !t;
Please challenge function within WHAT is already implemented. Also notice that output format is still rather crude. Rest assured more will come during autumn. But do remember that you should NOT use production data - work on copies of real data.
Please also discuss function and problems on this list.
Best wishes for fruitful testing
Torsten Christiansen Jens Lauritsen EpiData Association, Denmark