I will be out of the office starting 11/07/2008 and will not return until
11/10/2008.
If you require immediate assistance, please contact Celina Sousa at
extension 5300.
Thanks and have a nice weekend,
Alyshia
Hello,
Three suggestions on last version of Epidata Analysis 2.0 (very good on the
whole):
- a regression line in scatter graph is imperative like EPI 6 and all the
statistic softwares ++
- a few "set command" are desirable; for example :
* "set statistics = on" instead to write /t for all the commands
* "set table percentage = row" instead to write /r for all the "tables"
commands
- "set freq percentage= column" instead to write /c for all "freq" commands
A question about "lifetable" command (very useful and pratical) :
- In understand that the method is "actuarial" ou "lifetable"
with a range of time for the calculation of probability (1 by default).
The result is the same than SPSS with the command "lifetable" and time
range = 1. It's not Kaplan-Meier method because the probability is not
calculated at each event, (or calculation of remaining number at each
censored data).
Thank you your observation.
B. BRANGER.
PS : I modify this week the help on my site www.reseau-naissance.fr
<http://www.reseau-naissance.fr/>
I am more than just in favor of options, I believe that they are the
best possible approach, it makes everything so straight forward and
simple. Sometimes we want something to test, then we do it with the
option, else we take it out, and it is all in the same command line,
rather than a "set xx=on" then the command, then "set xx=off" again.
Options are very intuitive.
Hans
There will always be differences as to how users wish to run different
commands. Some time ago a discussion was on the list regarding the list.
The decision was to move as much as possible to options, since
otherwise the users have to issue different commands for changes.
As was suggested by B.Branger why not "get back" the "set statistics",
"set table percentage" etc.
The problem of this is the implementation in the menu system (dialogs),
which becomes a bit messy. The suggestion for users is rather to use the
following technique:
Add to your epidatastat.ini:
define tableoption __________ global
tableoption = " /t /R /C" // or whatever you prefer
or shorter:
define t __________ global
t = " /t /R /C" // or whatever you prefer
Now in your actual pgm file or at the command line, just issue:
tab v1 v2 @t
You might say "a bit annoying" to add @t, but still less than writing
the options.
Another aspect of the issue is that the set definitions mentioned are
part of the Epi Info v6 specification, but many users who did not do
statistical analysis with Epi6 are not aware of this by now.
regards
Jens Lauritsen
Epidata Association
B.Branger questioned whether the lifetable is "actuarial" or "Kaplan-Meier" principle.
The principle is to calculate changes in survival probabilities at each date,
where a change happens - either censoring or case status. In actuarial calculations there would be an
aggregation to say half year, month or similar. Therefore if you DO NOT use the /i for "Interval" the calculation
will be "Kaplan-Meier" technique. If you found an example where this is not the case let us know.
regards
Jens Lauritsen
EpiData Association
Hello,
Three suggestions on last version of Epidata Analysis 2.0 (very good on the
whole):
- a regression line in scatter graph is imperative like EPI 6 and all the
statistic softwares ++
- a few "set command" are desirable; for example :
* "set statistics = on" instead to write /t for all the commands
* "set table percentage = row" instead to write /r for all the "tables"
commands
- "set freq percentage= column" instead to write /c for all "freq" commands
A question about "lifetable" command (very useful and pratical) :
- In understand that the method is "actuarial" ou "lifetable"
with a range of time for the calculation of probability (1 by default).
The result is the same than SPSS with the command "lifetable" and time
range = 1. It's not Kaplan-Meier method because the probability is not
calculated at each event, (or calculation of remaining number at each
censored data).
Thank you your observation.
B. BRANGER.
PS : I modify this week the help on my site www.reseau-naissance.fr
<http://www.reseau-naissance.fr/>
A new build for EpiData Analysis has been placed for general download.
From v2.1.0.154 it is possible to give dates as a parameter to x-axis
limits in EpiCurve and Kaplan Meier Survival plots.
e.g. Epicurve outcome dayonset /xmin="01/10/1986" /xmax="25/10/1986"
Notice the DMY format.
A few other bugs have been solved. If you are not using Epicurve there
is no reason for updating.
A recent seminar in Canada forms part of the intensified development
ongoing this year. This collaboration is with the Assocation of Public
Health Epidemiologists in Ontario, Canada, see:
http://www.apheo.ca/index.php?pid=47
On the testing page (http://www.epidata.dk/testing.php) v2.1.1.154
includes two new commands released for tests and comments:
xbar /Range, xbar /Sigma , and ichart /MR - these will
perform double charts used in Statistical Process Control analysis and
visualisation. An extended user manual is in development and expected
ready for general use in about one month.
G-charts are expected within 1-2 weeks for testing.
Regards
Jens Lauritsen
EpiData Association