OK...I'll take the bait. Marco mentioned that only those clearly interested
in sharing data were responding to the most recent controversial topic, and
he wondered if there were another side to the discussion.
BACKGROUND
While I am quick to acknowledge the value in combining resources where
possible, I am slow to embrace the idea of grand repositories of data. This
is easily traced back to my graduate training where a process was described
for conducting scientific studies. In essence, the research question leads
to a testable hypothesis, that defines the data that need to be collected,
that require planning for statistical analyses "a priori." In the best
situation, conscientious investigators are intimately involved with all
aspects of the study, and truly understand the links between hypothesis,
data collection, and statistical analysis. In my experience, this is a
challenge (though quite possible) to achieve in a single site study. It
becomes more of a challenge in multi-center studies, but can be achieved
through effective communication during project planning, and vigilant
monitoring during data collection.
THE PROBLEM
My concern for grand repositories of data is that this intimate appreciation
for the nuances in data collection will not be adequately achieved. This
has negative implications for the integrity of data going into the
repository, and for data being retrieved from it. If human nature were to
tempt some investigators to combine data that SEEM to be collected under
SIMILAR conditions, I would be concerned that increased variability could
obscure true differences that would have been detected in a more rigorously
conducted study. Most importantly, "disinformation" would take time to
correct in the literature.
A POSSIBLE SOLUTION
Like others have stated, I also believe that methodology sections need to be
detailed enough to allow for replication of studies, and therefore
validation of results by others. If this is done, colleagues will have a
good appreciation for those conducting similar work. Rather than creating
repositories of the data themselves, I would suggest a directory of
investigators willing to share data. In this situation, one investigator
will need to contact another, providing opportunities to confirm the
appropriateness of combining data. Science still benefits from larger data
sets, but to me, the integrity of the data would seem to be better
safe-guarded.
Frank L Buczek Jr, PhD
Director, Motion Analysis Laboratory
Shriners Hospitals for Children
1645 West 8th Street, Erie PA, 16415, USA
(814) 875-8805 voice, (814) 875-8756 facsimile
fbuczek@shrinenet.org
---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------
in sharing data were responding to the most recent controversial topic, and
he wondered if there were another side to the discussion.
BACKGROUND
While I am quick to acknowledge the value in combining resources where
possible, I am slow to embrace the idea of grand repositories of data. This
is easily traced back to my graduate training where a process was described
for conducting scientific studies. In essence, the research question leads
to a testable hypothesis, that defines the data that need to be collected,
that require planning for statistical analyses "a priori." In the best
situation, conscientious investigators are intimately involved with all
aspects of the study, and truly understand the links between hypothesis,
data collection, and statistical analysis. In my experience, this is a
challenge (though quite possible) to achieve in a single site study. It
becomes more of a challenge in multi-center studies, but can be achieved
through effective communication during project planning, and vigilant
monitoring during data collection.
THE PROBLEM
My concern for grand repositories of data is that this intimate appreciation
for the nuances in data collection will not be adequately achieved. This
has negative implications for the integrity of data going into the
repository, and for data being retrieved from it. If human nature were to
tempt some investigators to combine data that SEEM to be collected under
SIMILAR conditions, I would be concerned that increased variability could
obscure true differences that would have been detected in a more rigorously
conducted study. Most importantly, "disinformation" would take time to
correct in the literature.
A POSSIBLE SOLUTION
Like others have stated, I also believe that methodology sections need to be
detailed enough to allow for replication of studies, and therefore
validation of results by others. If this is done, colleagues will have a
good appreciation for those conducting similar work. Rather than creating
repositories of the data themselves, I would suggest a directory of
investigators willing to share data. In this situation, one investigator
will need to contact another, providing opportunities to confirm the
appropriateness of combining data. Science still benefits from larger data
sets, but to me, the integrity of the data would seem to be better
safe-guarded.
Frank L Buczek Jr, PhD
Director, Motion Analysis Laboratory
Shriners Hospitals for Children
1645 West 8th Street, Erie PA, 16415, USA
(814) 875-8805 voice, (814) 875-8756 facsimile
fbuczek@shrinenet.org
---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------