[DDI-users] Nesstar Webview/Server and limits on # of variables in output datasets ?

[Bob McConnaughey]

Out of curiosity -  when i attempted to export all the questionnaire/data
files that I've put into a Nesstar database from the Webview program (8
files w/ 1 rec/subject in each; 1 file with 1 record for each day of subject
participation (lab data from urine samples) there was no "obvious" complaint
- however none of the Stata7/8 files or the SPSS sav files were readable;
the SPSS transport format (*.por) files were fine, albeit w/ truncated value
labels).  On the other hand when i created  a linked subsets from multiple
QX files (all cases) the new file was fine in all formats.  And when I
exported, say 3 of the datasets @ a go, the exported files were all
readable.

And exporting all the files in one go from within Nesstar Publisher worked
flawlessly (though the old 1980 SIC/SOC* codings-labels were not applied,
although they showed up w/out issue in the tables/frequency charts that
Publisher so nicely displays). Is that simply a problem of there being too
many possible variable values ( as it doesn't seem to be a label length
problem).  I deleted the formats/value labels that were not actually used -
so there are something less the 99 possible values which seems to cause
problems w/ the Stata codebook procedure).  I'm happy to include a proc
format statement for the SIC/SOC codes which a user could easily convert
into a Stata or SPSS statement - or the Stata/SPSS statements themselves but
I just wondered if this is a known issue or something that I'm mucking up on
my own.

I thought the Server/Webviewer combo might be getting upset by having one
dataset (the lab data) w/ multiple records (anywhere from 30 to 200) per ID
and the other 8 with just 1 record/ID but that didn't seem to be the source
of the unsuccessful data export from the Nesstar Webviewer 4.x (and ought to
have been a simple one to many merge ..if that's what i'd wanted, anyways)
Writing out the files one at a time worked fine with all the formats I
tested  - the missing SIC/SOC codes excepted.

An additional general question in re file structure.  Our data was
originally collected in a variety of data layouts.
1. QX data - 1 per person/per questionnaire. (221 records for each)
2. Prospective reproductive diaries/samples which were collected weekly,
using a "week" calendar. So anywhere from 8 to 30 weeks/cards of data, 7
days on each record, depending on how long the participant was enrolled
(Collection stopped either after 6 months or 4-8 weeks after a pregnancy was
clinically observed)
3. Lab data which was organized by id/day-date.

However much of the actual analysis was done using each menstrual cycle as a
discrete unit of analysis, so there are some analysis files that have 740
records w/ the various daily reproductive variables and  assays handled as
"horizontal" SAS arrays -days1-day90 (there were a few very long cycles,
unfortunately, even though most were ~ 31 days) as well as the various
demographic/behavioral/exposure variables.  Other analyses were done the
data set up with 27000+ records - with each one being assigned to the
ID/CYCLE N/Day w/in the cycle.  How far down the analytic road do users
expect to be led?  On the one hand we could provide the data and a file with
menses start/stop dates for each participant. Or we could provide a few of
the synthetic files both for the data and as examples. (And with the
"synthetic" files we could include many of the summary variables (ie rate of
hCG rise over the first 5 days of a conception cycle; or let users handle
the data as they will w/out much in the way of feeding them 25 yrs of
analysis -> 60+ papers.)

as always,
thanks

Bob McConnaughey

"Well, I too would be capable of killing for a book."
"I wouldn't recommend it.  That's how it starts. Murder doesn't seem like a
big deal, but then you end up lying, voting in elections, things like that"
 Perez-Reverte - *The Club Dumas*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.icpsr.umich.edu/pipermail/ddi-users/attachments/20110630/7add8720/attachment-0001.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: publisher labels.jpg
Type: image/jpeg
Size: 26543 bytes
Desc: not available
Url : http://www.icpsr.umich.edu/pipermail/ddi-users/attachments/20110630/7add8720/attachment-0002.jpg 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nesstar webview sans SICSOC labels.jpg
Type: image/jpeg
Size: 54696 bytes
Desc: not available
Url : http://www.icpsr.umich.edu/pipermail/ddi-users/attachments/20110630/7add8720/attachment-0003.jpg