confidentiality

[Brian MacWhinney]

Dear Info-CHILDES,

  Thanks so far to Julie Masterson, Susan Ervin-Tripp, Amy Sheldon, Eve
Clark, Marten Erikkson, Lynne Hewitt, and Kathy Hirsh-Pasek for
comments on confidentiality.  I think that everyone agrees that the
researcher cannot "grant" permission unless permission was given by the
people being observed.  Conservatism is also a good idea, but even the
most restrictive concept of data usage envisions some form of
laboratory use of the data. To further thinking about this, we can
divide data into three types.
   The first type is new data from normally-developing populations.
Here, the availability of a graded set of confidentiality levels may
help a researcher communicate with subjects regarding the use of the
data.  Susan Ervin-Tripp says that Berkeley uses a graded set much like
this as a model for their current IRB review.  However, Marten Eriksson
and Amy Sheldon feel that eight categories are too many for parents.
This is true.  The list is there so that the experimenter can select
out of these eight levels the one or perhaps two levels that are
appropriate for the particular study.  The crucial point is that, being
able to refer permissions for a particular study back to this general
typology will allow us to be clearer in the future regarding the level
of permission being granted.
   A second type is new data from special populations.  Because there
is social stigma attached to being "different", it appears that these
subjects and their parents do not want to have data used outside of the
laboratory under any conditions.  However, it seems to me that levels
6, 7, and 8 maintain this notion of "only used in the laboratory" for
these populations.
   The third type is older data.  Here, the issue of subject permission
was often inexplicit.  The internet did not exist and so this method of
viewing data could not have been envisioned.  I roughly agree with you
that my Level 5 would be appropriate for older data of this type.
   Let my point out another issue that was mentioned today in
discussions with Catherine Snow, Lauren Resnick and others regarding
videos for teacher training regarding early native language oral
proficiency.  In such cases, it is important to have releases from each
parent of each child being filmed.  If a particular parent does not
give a release, that child has to be "kept off screen".  So there are
further complications here even for new data.
   Finally, there are the technical issues.  The blurring of voices
and faces is technically possible.  These are the audio and video
equivalents of the use of pseudonyms for transcripts.  If this is done,
how does our understanding of the confidentiality issues change.  I
would think that it would then change radically.  Am I wrong?
  It is easy to block network access to people who do not have
passwords etc.  However, as Amy Sheldon noted, it may not be so easy to
block people from making copies of video.  One possibility would be to
add a banner to videos indicating that the data are for research
purposes only and are not to be viewed outside of research laboratories
and (in some cases) not copied.  Can these banners be done in a way
that presents them being "erased"?  I'm not sure.

--Brian MacWhinney